Tentacle Robotic Gripper

The automation of sorting processes within the agricultural industry remains limited by the challenge of manipulating fragile and non-uniform produce without damage, as conventional rigid solutions are often unsuitable or prohibitively expensive. This project investigates the application of soft robotics to address this gap by developing a pneumatic tentacle gripper capable of grasping objects with variable geometries. The primary objective was to design a functional and cost-effective prototype capable of handling a payload of 200g, the average mass of a small fruit, while ensuring intrinsic safety during interaction.

The prototype methodology followed a rigorous analytical approach, commencing with a functional analysis and state-of-the-art review to validate the solution's originality. The gripper design relied on a Finite Element Analysis (FEA) to define the geometry of the internal chambers within the Dragonskin A-20 silicone tentacles, ensuring optimal curvature at a service pressure of 0.5 bar and minimizing physical iterations. Concurrently, the sizing of the 2-DOF manipulator arm was dictated by a torque requirement analysis: calculations revealed a demand of 2.40 Nm at the lower joint, exceeding the direct-drive capacity of the selected MG996R servomotors. Consequently, two pulley-belt transmission systems with reduction ratios of 4 and 1.5 were integrated into the laser-cut MDF structure.

Results demonstrate the validity of the proof of concept, with the prototype successfully grasping and lifting target loads through the controlled deformation of four unidirectional tentacles. However, a post-operational critical analysis highlights significant technical limitations, specifically insufficient structural rigidity at the shoulder resulting in parasitic movements, and the suboptimal reliability of the custom-fabricated timing belts. While the pneumatic system and electronic control via Arduino and MOSFET are functional, the absence of active pump regulation represents a major area for improvement regarding the energy efficiency of a future industrial iteration.

Project by Ayoub Hamam, Julian Meskens, Matteo Paolini, Ariadne Samaey & Henri Wauthier

Under supervision by Prof. Bram Vanderborght, Milan Amighi, Parham Mohamadi, Aleksander Burkiewicz & Dylan Sisavath.

Lasercutter; Prusa Mini 3D printer; Basic tools

BOM.xlsx

1. Introduction
2. Table of contents
3. Project motivation
4. Project working modes
5. Functional Analysis
6. Requirement list
7. Eco-Design analysis
8. Conclusion
9. State of the art and patent analysis
10. Festo Tentacle Gripper
11. Harvard Multi-Tentacle Gripper
12. Soft Robotics Gripper
13. SpiRobs Logarithmic Spiral Shaped Robot
14. Comparison criteria
15. Comparison Table
16. Patent Analysis
17. Conclusion
18. Conceptual design
19. Function identification
20. Morphological chart
21. Analysis of solutions
22. Concept generation: Arm
23. Concept generation: Tentacles
24. Concept generation: Gripper structure
25. Embodiment design
26. High-Level diagram
27. CAD-design
28. Material selection
29. Manufacturing process
30. Assembly
31. Design of sub-systems
32. Mechanical systems
33. Circuitry & sensors
34. Software and control
35. Integration guide
36. Demo
37. Critical review
38. Sustainability
39. Bill of materials
40. The team
41. Project Repo

One of the biggest issues in the agricultural industry, is the amount of intensive manual labour that is still required. From the picking of fruit and harvesting of vegetables, to the sorting and packing of the produce. This manual labour is needed because the fruits and vegetables are often too delicate and non-uniform for heavy machinery. Companies are working towards more automated processes, and the autonomous fruit sorting systems market is a profitable and growing market.[1] This automation is done with the intent to cut down on costs and labour and to drive up the percentage of produce that can be sold. Damaged products cannot be sold to consumers, and thus most automated processes cause quite a big waste of produce. To combat this waste and to accommodate a more efficient process, we decided to look towards the soft robotics field. So far, soft robotics have been mainly focussed on research, not real-world applications. We want to make a robotic gripper that can safely and softly grasp fruits and vegetables without causing damage.

Most automated sorting apparatus that exists on the market are big and bulky machines that sort very specific types of objects. So, if a farmer has multiple types of fruits and vegetables that need to be sorted, several machines are needed. To cut down on costs, a singular multifunctional robot will be developed that can be used to sort a multitude of objects.

So, the idea for this project is to create a soft robot that can grasp different fruits and vegetables.

The need for these machines can be seen in the market share of harvesting robots globally, which was set at 2.24 billion USD in 2024, and is expected to grow at a CAGR of 21.9% in the next 5 years.[2] Most of these robots are autonomous tractors, but there is also a definite need for the harvesting and sorting of fruit.

The target persona for this robot are thus agricultural companies that are looking to automate their fruit sorting process. These are orchards, or other fruit companies. The robot will be used by employees of the company, but the amount of needed personnel would be reduced.

[1] Reports, G. M. (n.d.). Global Market Research Reports and Consulting Company. Growth Market Reports. https://growthmarketreports.com/

[2] Grand View Research. (n.d.) Harvesting Robots Marker (2025-2030) https://www.grandviewresearch.com/industry-analysis/harvesting-robots-market-report

Functional Analysis

Arm-movement

A system that can move the robot-arm with a good level of accuracy. Ensuring that the robot is well placed to pick-up or drop-off objects. This all needs to happen at a reasonable speed and in a safe manner. The arm should be able to move in a wide range, to maximize the working area of the robot.

Tentacle actuation

A system that can actuate the soft gripper. The necessary forces to pick up different objects should be generated, and one should be able to alter the amount of force, based on the object that is picked up. The system should also be easily deconstructed, in case of a failure in the system.

Soft gripping

The gripper that is used needs to pick up a multitude of different fruits. To accommodate this, and to prevent the bruising of the fruit, a soft material should be used. This soft material needs to stretch and elastically deform when actuated but return to original form when actuation ceases.

Requirement list

Table 1: List of requirements with level of importance, with 1 the most important and 3 the least.

The requirement list seen in Table 1 is compiled to list out the most important criteria for the robot. The ‘importance’ column has a number from 1 to 3 where 1 is the most important and 3 is the least. This ranking is made to identify the focus-fields in the design of the robot. The ‘level’ column has target values that the robot should be designed to meet. This list is a checklist to identify if the robot meets the functional criteria. To rigorously ensure compliance with these criteria, the requirements are segmented into categories, each demanding a specific Verification and Validation methodology adapted to its physical nature.

For the geometry, kinematics, and materials categories, compliance will be validated through finite element method (FEM) (section ‘FE-analysis’) simulations and real-life testing. The forces and maximum torque on the motors used in the arm, as well as the pulleys needed to facilitate the movement of the arm was manually calculated and tested to prevent overexertion of the motors.

Operation, safety, and maintenance will shift from simulation to empirical validation. Critical metrics such as response time, acoustic noise levels, and pick-and-place reliability cannot be purely calculated; they must be verified through physical benchmarking.

Eco-Design analysis

To find the ecological impact of the robot, an analysis is done. This means that the ecological impact of this robot is looked at. There are certain parts where the robot gets a bad impact.

The silicone of the tentacles is not recyclable, if they rupture and are not able to be repaired, they need to be thrown away. In case of very small holes, a repair is still possible most of the time.

The robot uses a modular connection system for the gripper, and most parts of the arm are equally replaceable.

The pneumatic system uses push-fittings, this means that in case of a blockage or tear in one of the tubes they can easily be disconnected. A malfunction in the valve or any other component can be solved by disconnecting the tubes and swapping out the affected components. For more information on the possible improvements, see the section ‘Sustainability’ in this paper.

Conclusion

The project is to build a manoeuvrable robotic arm that uses tentacles to securely grip fruits such as apples, bananas and oranges, pick them up, and move them to a different point in a predefined working area. The robot will be manually controlled and can move with two degrees of freedom. This, without dropping or permanently deforming the fruit. This entire process should respect the listed requirements in Table 1 as best as possible. In this project, a prototype is designed and build to serve as a proof of concept.

When designing a product for the market, doing research first is of great importance. This is because a product needs to be profitable for a company to not go under. Thus, an analysis is carried out where market competitors and potential users are identified. If a product is well designed but there are no buyers, it is a big waste of money. Reviewing competitors allows for an overview of functionalities and features that a product would need to be competitive. Based on the strong and weak points of these products, the places of innovation can be decided. Most of the tentacle robots that currently exist are used for research or demonstrations and thus have no practical application. To compare some features found in the state of the art, a focus is put on several soft gripper robots with differing actuating methods and uses. A summary is given in Table 2.

Festo Tentacle Gripper

Based on the limbs of an octopus, this gripper uses a singular flexible arm that is pneumatically actuated. The gripping force of the arm is enhanced by suction cups. The movement of the arm is controlled by a terminal that regulates the actuation. This means that the gripper is quite successful at picking up differently shaped objects and is thus incredibly multifunctional. The main downside of this product lies in its complicated actuating methods and price. Because it uses a multitude of sensors and vacuum technology, the price point for this robot is quite high, too high to justify in the fruit sorting industry. It is also not easily accessible to buy.[1]

Harvard Multi-Tentacle Gripper

This is another robot inspired by nature, more specifically by the tentacles that jellyfish use to collect their stunned prey. The mechanism consists of a multitude of small and fragile tentacles that curl around the target object when pneumatically inflated. When actuated, each small tentacle curls in many different directions, which makes them grip to objects next to them. These tentacles are quite weak on their own but when together, they can wrap around larger objects and pick them up. This robot was developed at the university of Harvard, mainly for research and is thus not available for purchasing. Furthermore, the production of the tentacles is a process that needs to be carried out meticulously, to avoid tears and other problems in the thin tentacles.[2]

Soft Robotics Gripper

Soft robotics gripper is a collection of many differently shaped grippers that all share some main criteria. Since they aren’t considered as tentacle grippers but still share the same idea since they are soft and pneumatic, these grippers are indirect competitors. The gripper is formed of a couple rubber parts that are pneumatically actuated. This means that they inflate, which causes them to deform. The deformation then brings them closer to the target object, curving around it. The pressure of the gripper around the object then allows the object to be picked up. Most of the time these grippers also have a mechanical component where the soft parts are closed around the target object and then inflated to pick it up. This mechanism makes them a more affordable gripper, but as a major drawback it will have more difficulty with some more irregular shapes. This gripper does not have a big deformation and thus cannot really be described as a tentacle gripper, but its function is similar.[3]

SpiRobs Logarithmic Spiral Shaped Robot

The SpiRobs robot is another university designed robot, mainly made for research. This means that it is not an available product that can be bought by companies. The robot uses a long spiral shaped arm, that is made of a soft, deformable plastic parts. When actuated, the tendons in the corners of the parts are pulled, moving the arm. When the arm touches the target object, the further tightening of the tendons causes the plastic to deform. This deformation translates to a strong grip on the object.[4]

Comparison criteria

The comparison between the different existing grippers will be done based on six criteria: the type, the actuation, the flexibility of the gripper, the complexity of the control mechanism, the application, and of course the price.

The type is of importance, since a different type of gripper may be more useful for a certain object. A large amount of thin and long tentacles might be great for grasping cylindrical objects, but quite incompetent at grasping a spherical object, as the individual tentacles would slip off due to lack of friction. Choosing which type of gripper our own robot will be, will choose which other robots we’re competing with.

The actuation is closely correlated with the type and influences the level of complexity of the robot greatly. It is a trade-off between consistency and complexity.

The flexibility of the gripper is a parameter that makes a trade-off between versatility and reliability. A flexible gripper might be able to grasp more types of objects but will work less consistently.

The complexity of the control is correlated with the repairability, user friendliness and cost. A simpler system might work less efficiently but might be easier to implement.

The application decides whether our own design will serve as an academic example or have industrial purposes. As it influences the profit model of the product immensely, it should be included in the comparison.

Since most of these robots can’t be found on the market and have no public information on their manufacturing process, it makes it impossible to know their price. However, a lower cost of production is always a positive, and known references helps giving a vision of the total cost to aim for.

Comparison Table

Table 2: Comparison table of the state of the art. The highlighted cells show the best aspects of their designs, based on the goals of this project.

Patent Analysis

The main area of innovation for this robot is the gripper. The robot arm can be considered as a standard robot arm. This patent analysis is only considered for the gripper.

Because this robot is made by students at the University of Brussels, only patents that are active in Europe are of importance.

US9464642B2[5]

‘Soft robotic actuators’

This patent pertains to the shape of inner chambers in a pneumatically actuated tentacle (Figure 1). It is for one-sided pneumatic channels and their shape and curvature. The tentacle has a wall that is thicker that the rest, to give it a preferential expansion direction. The pleated inner chambers to allow for ease of movement is also patented.

This patent is only active in the USA and does not limit the design.

EP3529011B1[6]

‘Soft actuators’

The focus of this patent lies in the connections between inflatable chambers and a rotating core (Figure 2). This core has 4 degrees of freedom and moves together with the actuated tentacles.

The patent does not apply to the gripper or its shape, only the connection and is not applicable for the design.

GB202004276D0[7]

‘A learnt approach for the design of magnetically actuated shape forming soft tentacle robots’

The robot in this patent uses a magnetic elastomer. This means that the tentacles are actuated using an electric field (Figure 3). The use of these and their shapes is covered. Since magnetic elastomers are expensive and difficult to use, they will not be used in this project. This patent does not limit the design.

EP3245028B1[8]

‘Soft robotic device with fiber Bragg grating sensor’

The focus of this patent lies in the integration of a fiber Bragg grating sensor into its core (Figure 4). This sensor is an optical one that is part of the sensors needed for autonomous control. Since there is no plan to use such a sensor, it does not impact the design

US11565406B2[9]

‘Multi-tentacular soft robotic grippers’

This patent covers the gripping procedure for autonomous grippers (Figure 5). This procedure is used to identify the object and then correctly sort them. The patent is only active in the USA, and its procedure is not needed for the robot, it does not limit the design.

Conclusion

From the state of the art, it can be concluded that the proposed design is original enough to be an innovation and will have no conflicting competition on the market. It will expand on the growing market for soft grippers but not directly copy any product. The innovation that our design is introducing is a specialty in moving fruits with care and being a commercial product rather than an academic tool.

From all these patents it can also be concluded that there is freedom to operate. There are no direct conflicts with the technology or design aspects used in the patents. There are also no pending patents on the methods used. It is safe to start the development of the robot.

[1] Festo. (2025). Tentacle Gripper. https://www.festo.com/us/en/e/about-festo/research-and-development/bionic-learning-network/bionic-grippers-and-soft-robots/tentaclegripper-id_33321/

[2] Harvard. (2022). Tentacle robot can gently grasp fragile objects. https://seas.harvard.edu/news/2022/10/tentacle-robot-can-gently-grasp-fragile-objects

[3] Schmalz. (n.d.). Finger Grippers mGrip. https://www.schmalz.com/en/products/vacuum-technology-for-automation-301607/vacuum-components-301608/area-gripping-systems-and-end-effectors-306363/finger-grippers-312388/finger-grippers-mgrip-405170

[4] Wang Z., Freris N. and Wei X. (n.d.). SpiRobs: Logarithmic spiral-shaped robots for versatile grasping across scales. https://www.sciencedirect.com/science/article/pii/S2666998624006033

[5] Ilievski, F., Chen, X., Mazzeo, A. D., Whitesides, G. M., Shepherd, R. F., Martinez, R. V., Choi, W. J., Kwok, S. W., Morin, S., Stokes, A., & Nie, Z. (2010). US9464642B2 - Soft Robotic Actuators - Google Patents. https://patents.google.com/patent/US9464642B2/en

[6] Manfredi, L. (2016, October 24). EP3529011B1 - Soft actuators - Google Patents. https://patents.google.com/patent/EP3529011B1/en

[7] University of Leeds & University of Leeds Innovations Ltd. (2020, March 24). GB202004276D0 - A learnt approach for the design of magnetically actuated shape forming soft tentacle robots - Google Patents. https://patents.google.com/patent/GB202004276D0/en

[8] Lessing, J. A., e.a. (2015, January 12). EP3245028B1 - Soft robotic device with fiber bragg grating sensor - Google Patents. https://patents.google.com/patent/EP3245028B1/en

[9] Yerazunis, W., Solomon, E., & Inc, M. E. R. L. (2020, November 23). US11565406B2 - Multi-tentacular soft robotic grippers - Google Patents. https://patents.google.com/patent/US11565406B2/en

Function identification

The main goal of the robot is to pick up differently sized objects. This means that the pickup process of the robot needs to be variable, to allow multifunctionality. The prototype robot is designed with a controller that is handled by a human operator. This controller is used to move the arm and actuate the tentacles. The main functionalities of the robot, based on the requirement list in Table 1, consist of:

1. The robot arm should be reliably controllable in space.
2. The gripping force should be sufficient to lift fruit.
3. Good controllability, not too complex and no danger for operator.
4. Reliable enough to grasp fruit of varying sizes and softness without causing damage.
5. Good repairability: easy to troubleshoot and repair.
6. Low cost of materials
7. Easy fabrication, to keep the manufacturing costs low.

These features represent the seven main criteria on which the morphological chart, given in Table 3, is based on. In this chart the seven main features are listed together with the different ways of achieving them. These features are chosen to give the gripper and arm all the required functionalities listed above. The highlighted options in the table are the chosen means of achieving this feature for the product. From the morphological chart the functions are found for which solutions need to be found.

Morphological chart

Table 3: Morphological chart comparing the different design choices to achieve certain functions. The highlighted cells are the options that are chosen for the prototype.

Analysis of solutions

The main area of innovation for this project lies in the design of the tentacles. Their shape and actuation are very defining for the type of robot that is built in this project. In the morphological chart in Table 3, the choice was made to use for silicone pneumatic unidirectional tentacles. This choice was made after analysing different types of tentacles and actuations. The other design choices, made for the arm and the control system have also been well thought over. This is why an overview of the pros and cons of each design choice is given in this section, which will explain why some choices were made. There are also some extra choices that needed to be made that followed from the choices made in the morphological chart.

Tentacle actuation and material

The actuation of soft grippers can be divided into three main groups.

One consists of mechanically actuated tentacles, using tendons that pull on the tentacle to deform them. In this case the tentacles have a structure that consists of interconnected triangles and is made of a material that can deform easily like a soft and thin plastic. The tendons are pulled and the material deforms. The gripper itself is then soft in the sense that the material can deform because it needs to, to be actuated. It is still less soft than ideal for this project, and the control over these mechanical tentacles is more complicated than for pneumatic actuation.

The second category is made up of fluid-controlled tentacles. In this case a tentacle is made up of an elastomer with one or multiple interconnected chambers. These inner chambers are filled with a fluid in case of actuation. The shape of the tentacle and inner chambers are designed in such a way that it lends to the bending of the tentacle in a certain direction. The fluid used can vary, but the most logical one is air, as it does not cause problematic leaks or other issues. Additionally, the pressure in the tentacles is not high enough to justify using a hydraulic system, when a pneumatic one suffices. Since the tentacles can be made with casting, the fabrication is simple enough for the project. Furthermore, it has room for a lot of design choices, especially in the design of the internal chambers.

The third option is the use of electroactive polymers. This means that the tentacle is made of a polymer that, in case of actuation with an electric pulse, moves the tentacle by deforming. This material is quite expensive, highly experimental, and complex to control, thus not a viable option for a student project.

It is also possible to combine some of these actuation types in something called a hybrid type of actuation. But this way of working is very complicated, in both control and fabrication, and would both cost and weight more. This is why the choice was made to use a simple pneumatically actuated system.

Furthermore, since the goal is to keep the tentacles as simple as possible, the pneumatic network of chambers will be unidirectional, which means that the tentacles can only move in one direction. This is mostly because multi-directional tentacles are not needed for the type of gripper made in this project. For every extra direction that a tentacle to moves, an extra set of internal chambers needs to be added that can be actuated separately. This would increase both the cost, the diameter of the tentacle (and thus weight), and increase the complexity of control and manufacturing.

Size and number of tentacles

The amount and size of the tentacles are closely linked together. This is because tiny tentacles are quite weak and thus need a higher number to pick up an object. Larger tentacles weigh a lot more and thus should not be too numerous. Medium sized tentacles seem like the best option, because they are easier to manufacture than the small tentacles, where air bubbles in the material or small weaknesses have a large impact. Furthermore, they use less material than the large tentacles. Because the motors should not be overexerted, the load at the arm should be minimized.

A total of four tentacles is used to maximize grip and account for the possibility that one of the tentacled might miss the object, seeing as it has a predetermined deformation. So, in case of the misalignment of the gripper, the object might only be in contact with 3 of the tentacles. This will still allow the object to be picked up.

Tentacle air-chamber

Since pneumatical actuation was chosen (so the tentacles move with air), some extra design choices must be made about the chambers and tube-network. The inner chamber or the air channel inside of the tentacle decides how the tentacle will curve when air is pumped into it.

To start with, the directionality of the channels should be considered. Directionality refers to the directions in which the air can move in the channel. A bidirectional channel can have air moving in and out of the same channel. For unidirectional channels, two are needed: one to let the air in and one to let it out. This directionality is needed to control the air in a tentacle. Bidirectionality in the chamber does need an external system, like a solenoid, to control whether the air should move in or out of the tentacle. Bidirectional channels are chosen for this project, as they are easier to produce and introduce less points of failure.

The further shape of the channels will be explored through finite elements analysis to find the optimal shape of them without having to produce tons of sub-optimal tentacles. The main design will be done by making an inner core that has certain sections that are extruded. This creates parts of the tentacle that have more or less material and thus can either inflate more easily or resist moving.

Gripper structure and tentacle fixation

For the structure of the gripper, the choice falls between making it one solid piece or making it modular. Modularity would make the fabrication slightly more complicated, since more parts need to be made and the assembly of those parts need to function correctly. The advantage of modularity would be the repairability, since if a part breaks, it can be replaced individually.

Another functionality of the gripper structure is how it fixes the tentacles in place. Should they be loose or fixed? Making them loose would create problems with the gripping, as the location of the tentacles would change when grasping an object. This is why the tentacles should be fixed in place, so that the gripping force when actuated is more reliable.

Arm design and motor choice

Since the goal of the prototype is to lift an object from the ground, only two degrees of freedom are required. In a final design, extra degrees of freedom could be added to the arm in case it requires the ability to sort or move the objects to another place. This is why the prototype arm is built in two parts, each actuated using a motor.

The prototype is kept small to avoid high costs in material and actuators. The choice of using hobby servo motors mainly lies in its low cost and low complexity to control. Since it has an internal encoder, a UWB signal can be sent that tells it to rotate to a certain angle. The total rotation range of the servo is limited to 120°, but since the arms of a robot arm don’t need the ability to do even a full rotation, this won’t be a problem. This choice does mean that a direct drive wouldn’t be possible for this project, as more torque is needed than the available servos are able to deliver. This way using stepper and DC-motors can be avoided, which is beneficial as both are heavier and more complex to control. If these other motors were used encoders would need to be added. Since the choice goes to simplicity over high power or continuous rotation, servo motors are the best choice. The use of servo motors in robotic applications is also very common.

There is no need for the accuracy level of a stepper motor, since the control of the robot is manual, and the user will visually decide where to move the arm. The movement of a stepper is also less smooth. Furthermore, stepper motors have the big drawback that they continuously need power to apply a static torque. This means that, when turning off the power, the arm would not be able to support its own weight and collapse on the ground. This would mean that a brake or locking mechanism would have to be designed and added to the arm.

DC motors can deliver a lot of torque when combined with a gearbox and are most suitable for continuous rotation. They are not often used for tasks with low speeds and holding positions. The control also needs an encoder and controller, which in combination with the gearbox, makes this option larger, heavier and more complex than a hobby servo. Even though a DC motor often costs less than a servo motor, the extra cost of the other needed components makes the servo a cheaper option.

Linear actuators are also a possible option but are too expensive and way stronger than needed for this project. This is why they won’t be considered as a viable option.

Torque transmission

Since the available servo motors cannot deliver sufficient torque to actuate the arm in direct drive, a transmission is needed. The designing process of making gears and attaching them to the arm can get really complicated and can cause for a lot of problems. This is why a pulley system was chosen, which is relatively easy to design, print and implement in the arm.

Control design

For the controls, it was chosen that the prototype operates on manual control. This is both cheap and easier to implement than any automatic system and sufficient for a prototype that serves as a proof of concept, which is to grasp and lift an object. Furthermore, since the control of the robot is simple, the provided Arduino Uno was used. There was no need for a stronger microcontroller.

Concept generation: Arm

Since we’re not reinventing the arm in any original or innovative way, no concept generation is done for the arm. The arm will be kept simple, so that it can do what it was made for, which is to demonstrate the lifting aspect for the proof of concept.

Concept generation: Tentacles

From this analysis several possible concepts can be developed. These will then be compared in Table 4 to find the ideal solution for the robot. The main concepts concern the shape of the tentacle and its air channel. These concepts can be compared with each other and subjected to an FE analysis. This FE analysis is done to find the pressure needed to actuate the tentacles and to monitor the curvature of the tentacles.

Concept 1: Simple internal air-channel

Figure 6: Concept 1; Sketch of simple air channel with thicker wall on one side.

This first concept for the realization of the air channel, in Figure 6, uses a single chamber with a constant cross-section. It sits just off-centre to achieve a preferential direction of curving when inflated, due to difference in wall thickness. This design is the simplest and the easiest to manufacture, but the curvature of the tentacle is not optimized for grasping, as the whole tentacle will deform equally.

Concept 2: External channel with chambers

Figure 7: Concept 2; Sketch of the external air channel design.

The second option, in Figure 7, is a tentacle that has an external air channel. This would mean that the tube that supplies air to the tentacle is connected to internal chambers individually. This would allow more control and good reliability, but it would be very impractical to have a multitude of connections that need to be made airtight. If any air leaked out from the connection tube-chamber, the tentacle would not function properly. It would, however, allow a modular design. Any problems with one chamber can be fixed by changing the pipes so that chamber is no longer used. This would mean that the gripper can still be used without all its chambers functional. This design does use more material as a tube and connectors run along the length of the tentacle, and care should be taken to prevent them from limiting the deformation of the tentacle.

Concept 3 & 4: Internal air channel with chambers

These next two concepts are closely related. The first, in Figure 8, is a simple tentacle with constant outer diameter, the second, in Figure 9, has a shape caused by a constant wall thickness around the air channel. What makes them similar is the shape of the air channel. It is a channel that has several interconnected air chambers, with a bigger cross section, along its length. These chambers allow the curvature of the tentacle to be chosen in the design, so a preferential direction and curvature is reached. The exact shape of these chambers was decided using finite elements modelling, as seen in the section: ‘FE-analysis’.

Figure 8: Concept 3; Sketch of internal air channel with multiple interconnected chambers (constant outer diameter).

Concept 3 has the advantage that it’s easier to produce. A simple 3D-printed cast can be used. It is the most practical design proposed, as seen in Table 4, and eventually the one chosen for the prototype.

Figure 9: Concept 4; Sketch of internal air channel with multiple interconnected chambers (constant wall thickness).

Concept 4 is more difficult to produce. It needs a specialised cast that would be difficult to print without internal supports. This means that a silicone outer mould would probably be needed to produce the tentacle. This would require us to print a 3D mock-up of the tentacle, encasing it in silicone and then using that mould to cast the final tentacle. In a final product this design would use the least amount of silicone, as more sophisticated casts would be available, but here the process of making one tentacle would take a lot of time and resources. This would lead to a lot of holdups and waste when prototyping.

Comparison of the concepts

Table 4: Concept generation of the internal air channel of the tentacle with a weighted scoring system, where the more points a concept has, the better it is.

Note that the values and weights in Table 4 pertain only to the prototype and its production. The final product has different areas of focus and production possibilities.

Synthesis of selected concept

Considering the ease of fabrication without making compromises in the curving of the tentacle, concept 3 is the preferred design. It is a simple tentacle with a fully enclosed connected chamber. It does not require a lot of silicone compared to the others, is controllable and reliable, and does not cost a lot to produce. It scores well on all aspects mentioned in the concept generation.

Concept generation: Gripper structure

The gripper holds the tentacles in place and connects them to the robotic arm. It’s important that the tentacles are fixed in a good position in order to apply enough pressure on the object to pick it up. In addition, it’s essential that the tentacles, when not actuated, are able to move around the object so that pressure can be applied from four sides. Three concepts were generated as improvements on each other.

Concept 1: Circular plate

Figure 10: First concept of a gripper structure.

A first design, seen in Figure 10, consisted of a simple circular plate, with a detachable stick that connects to the arm. In the plate there were holes preserved for separate connector bits, that on one end would attach with standard 6mm tubes, and on the other end could be inserted into the tentacle. This design was a quick draft and has several issues. First of all, the stick can’t handle a large amount of torsion and might break or experience a lot of fatigue as a result. The connector bits couldn’t hold the tentacles that well, as the inside pressure would just blow them off. This design is also relatively big and bulky.

Concept 2: Fixed holder

Figure 11: Second concept of a gripper structure.

The second design, seen in Figure 11, was in improvement on the first. It features a more secure connector for the tentacle, providing outside pressure onto the tentacle to prevent it from slipping off when pressure builds inside. The stick was also redesigned to be less flimsy and more symmetrical. While it takes longer to produce, this design was preferable to the first.

Concept 3: Adjustable holder

Figure 12: Final concept for the gripper structure.

To allow a positioning of the tentacles that can be changed depending on the way the consumer wants to use the gripper, a gripper is designed with an inner rail-like structure. This alteration to the design, seen in Figure 12, was made because the tentacles were found to be too close together to pick up bigger objects in prototype testing. By using a gripper where the tentacles can be moved, the gripper can be used for a greater number of different objects. The tentacles are attached to this gripper using a top and bottom part that secure the tentacle to the gripper. This piece has screws, which allows the tentacle to be unscrewed, slid along the inner rail, and then screwed securely again. If a consumer wants to pick up very big or small objects, they only need to move those tentacles.

Comparison of the concepts

Table 5: Concept generation of the gripper’s structure with a weighted scoring system, where the more points a concept has, the better it is.

Synthesis of selected gripper structure concept

As the third design features a better gripping force, more controllability, more reliability and easier repairability than the other designs, it’s a clear winner. It takes a bit longer to manufacture and takes some time to assemble, but performs way better afterwards, so it’s worth the time investment.

Figure 13: High-level block diagram of the robot’s system.

The ‘High level diagram’ in Figure 13 is a diagram that is used to show the interactions between the different subsystems of the gripper. It groups the components into subsystems and shows how those are interconnected. From this diagram it is easy to see that there are three main subsystems, the pneumatic subsystem, the mechanical one and the controller. The pneumatic subsystem refers to the entire system with a valve, a diode, a pressure regulator and the tentacle. The mechanical subsystem is the system that moves the arm, so the motors. The controller is the one that delivers the inputs to the different systems. With a button to open or close the valve on command. The potentiometers then can be used as dials, turning them turns the motors.

CAD-design

For the CAD-design, mostly Autodesk Inventor was used to draw parts and make assemblies. The gripper was made in Siemens NX.

Design of arm

The arm only had a few requirements: to be able to lift the required weight, to have a good range of motion, to be controllable precisely, to be able to move at a reasonable speed, to be safe for use, and the arm structure should make optimal use of the material. The full Inventor assembly of the arm can be seen in Figure 14 and the main part of the gripper in Figure 15. In comparison, a picture of the fully assembled arm in real-life can be seen in Figure 16. A bill of materials and complementary technical drawing are found in Table 6 and Figure 17 respectively.

The controls, movement speed and range of motion are discussed in the section ‘Design of sub-systems’. It’s important to understand that while “whole” plates are mostly used for the bigger pieces instead of cutting out material, this does not impact the amount of used MDF. At most, it would be possible to save a few grams here and there, but when a piece is cut from a plate using a laser cutter, it’s the outer frame of the design that decides how much MDF is used in the process. Additionally, for the humerus and ulna pieces of the arm, it was most important for them to have a good structural integrity, and weight was only a secondary concern. For the X-shaped pieces across the arm, it was more useful to cut out the unnecessary material, as this gives a person more access to the inside of the arm while assembling or performing quick repairs.

Figure 14: CAD model of the final assembly of the arm, made in Inventor.

Table 6: Bill of materials for the arm assembly.

Figure 15: Technical drawing of the assembly of the arm. Including numbering that correspond to the item numbers used in the bill of materials (Table 6).

Design of the gripper

Figure 16: CAD model of the main part of the gripper, made in Siemens NX.

Figure 17: Fully assembled tentacle robot, for comparison with the CAD designs in Figure 14 and Figure 15.

Material selection

Prototype

The materials used for the final product and the prototype are different. This is because the prototype is dependent on the materials and production tools. Because of this difference, this report focuses on the prototype materials. These materials need to be cheap, easily accessible, sufficiently strong and easy to manufacture parts out of.

Structural components are made from MDF. This is done because a laser-cutter is available at the FabLab. The thickness of the used MDF is 4mm, to give sufficient strength but keep the weight of the components as low as possible

The connection pieces are made of PLA, again because of availability. This is done to allow for the fast prototyping in case of design changes. The connections need to be exact for the parts of the pneumatic system, seeing as air leaking out will compromise the inflation of the tentacles.

The material of the tentacles needs to be soft and stretchy to deform under the air pressure and form around an object and at the same time it would have to be strong enough to hold up an object without tearing. It also needed to be easily used in manufacturing, and relatively cheap. Those requirements led to the selection of silicone, a soft and flexible rubber, more specifically a silicone with a shore of A-20. Because of availability and price Dragonskin A-20 was used. A shore of A-20 allows for the compromise between flexibility and strength that is needed for the tentacles in a flexible gripper.

For the casting mould (see section ‘Manufacturing process’), PLA is used. This is because it does not react with the silicone, and because it can be 3D-printed with a good precision.

Final Design

Structural components are made of aluminium. This choice was made based on the requirements listed out in Table 1. Using Granta-Edupack (Figure 18) an elimination process was carried out to find the ideal material. The gripper needs to be strong enough to withstand the stresses and strains that come from the deformation of the tentacles and the pick-up process. It also must stay at a manageable weight so, so that any arm it is connected to will be able to move the gripper in space without the need of heavy-duty motors. Aluminium is strong, light and cheap to both buy and to process. Other light metals like titanium or magnesium are either too expensive or may cause dangerous reactions when ingested. As this gripper is intended to be used in the fruit-processing industry, it is important to use a material that is not dangerous when ingested. So, the chosen material is food grade aluminium.[1]

For the material of the tentacles of the final design, the same kind of silicone can be used as in the prototype.

Figure 18: Chart for material selection (Granta-Edupack).

The choice of silicone for tentacles is common in the state of the art, as well as aluminium for a robotic arm.

Manufacturing process

Prototype

The manufacturing used for the components differs between the prototype and the final product. For this report only the prototype is considered. The manufacturing processes are chosen partly on the availability of materials and machines in the FabLab.

Laser cutting was used for most of the arm-components. This was decided because laser cutting wood is both faster and a more sustainable process than printing plastic parts, while still allowing for a quick and precise translation of CAD-models. This is mainly used for parts that are flat and big, like the arm structure.

3D-printing was used for components that have more complicated shapes. This is additionally useful because 3D-printing allows for quick prototyping and testing of the parts. This method was used for the connection pieces and the tentacle mould. The mould was 3D-printed to allow precise design and to make demoulding easier, as silicone does not stick to PLA.

The tentacles were cast, using the 3D-printed mould, seen in Figure 19. Casting is the easiest way to make a silicone shape. The riskiest part is the demoulding of the core of the tentacle, the part that becomes the air chamber. Extra caution is used when pulling out the core. A pressurized air gun can be used to break the seal between the silicone and the PLA, to ease the core-removal. The cast is screwed together, to allow easy demoulding and reuse.

Figure 19: Technical drawing of the 3D-printed mould used for casting the silicone tentacles.

Final Design

Table 7: Selection process for the manufacturing of the gripper’s structure.

For the main gripper parts, since they are 3D solid shapes made of aluminium, milling is the preferred manufacturing process. The selection of this process was done by comparing a multitude of industry standard manufacturing processes (see Table 7). The designs for the gripper parts that need to be milled are made with reduction of waste in mind. Milling uses a solid block of material, where material is removed until the desired shape is reached. To minimize the ecological impact of this product, the parts of the gripper are made to be quite regularly shaped. The main gripper part detaches from the connection piece, so both can be produced separately and material is saved.

Table 8: Selection process for manufacturing of the silicone tentacles.

The tentacles are to be cast, using a mould. As seen in Table 8, casting is the easiest way to make a silicone shape. The moulds will be placed in a vacuum chamber to prevent the formation of air-bubbles in the silicone which could cause weak points. The specific type of polymer casting chosen for this project is mould casting, where a solid plastic core is inserted into the wet silicone to act as a placeholder for the air channel. When the silicone is cured, the core is removed and air can flow into the tentacle.

FE-analysis

To test the design of the inner chamber for the tentacle a finite elements analysis was done. This means a virtual prototyping approach was adopted. This strategy allowed different design solutions to be analysed and compared prior to physical prototyping, reducing the number of experimental iterations required.

The fabrication of the tentacle with the selected geometry (Concept 3 in section ‘Concept generation: Tentacles’) requires an external mould for silicone casting and an internal core to form the internal cavity of the tentacle. During the prototyping phase, the external geometry was kept identical for all tentacles, while only the internal core geometry was varied. This approach made it possible to investigate the influence of the internal cavity on the deformation behaviour of the tentacle, isolating the effect of the core geometry.

The development process followed an iterative ‘design and test’ approach. Each core was first designed in a CAD environment (Siemens NX) and subsequently analysed through finite element simulations in Ansys Workbench. The obtained results were compared with those related to different core geometries, enabling a comparative evaluation of the tentacle behaviour. Based on these comparisons, the geometries were progressively modified until a configuration providing an acceptable deformation behaviour with respect to the project objectives was identified.

A finite element (FE) analysis was performed to evaluate the tentacle behaviour under pneumatic actuation. The tentacle was modelled as a solid body made of silicone elastomer and analysed under large deformation conditions. The internal cavities were explicitly modelled and loaded by applying a uniform internal pressure. The base of the tentacle was fully constrained, while the remaining external surfaces were left free.

Special attention was given to the definition of the material model, shown in Figure 20. The Dragon Skin Shore 20 silicone was modelled in Ansys Workbench using a second-order Ogden hyperelastic model, suitable for describing the non-linear mechanical behaviour of the material. The density was set to 1070 kg/m³, and the material was assumed to be incompressible by appropriately defining the incompressibility parameters.[2]

Figure 20: Used material parameters in Siemens NX.

The definition of the pressure load represented one of the main challenges of the simulation. Due to the presence of large deformations and the use of a hyperelastic material model, the internal pressure was applied gradually through multiple load steps under quasi-static conditions (Figure 21). This strategy improved numerical stability and promoted solution convergence, avoiding issues related to abrupt changes in the deformation state.

Figure 21: Applied internal pressure over time (Siemens NX).

Simulation results show that the asymmetric inflation of the internal cavities induces a bending motion consistent with the intended working principle. The curvature increases with increasing internal pressure, while maximum deformation values are mainly concentrated near the cavity edges and in transition regions between different wall thicknesses, identifying the most critical structural areas.

Despite some limitations related to the number of nodes and convergence issues, the simulations proved sufficiently reliable to provide qualitative insights for comparing different core geometries. The adopted virtual prototyping approach therefore effectively guided the tentacle design process and supported the selection of a configuration compatible with the project objectives and prototyping constraints.

Some results for different core shapes are given in Figure 22, Figure 23, and Figure 24.

Figure 22: Simulation and results for a design with first one larger and then one smaller air chamber.

Figure 23: Simulation and results for a design with first one smaller and then one larger air chamber.

Figure 24: Simulation and results for a design with two equally sized air chambers.

Assembly

Arm structure

The design of the component intended for laser cutting is managed to facilitate assembly via screws or an interlocking joint, which ensures simple and efficient production. An overview of the number of used screws is given in Table 9.

Table 9: Overview of standard parts used for each part of the arm.

It should be assumed that for each screw and bar used, at least one washer and nut is needed to fasten it.

Assembly should be done from the shoulder up. First, the shoulder pieces are put in the slits in the baseplate, which should be bolted to a steady surface. If this is not possible, a counterweight should be placed on the shoulder pieces to prevent the arm from tipping over when lifting items. L-profiles are used to connect the shoulder pieces to the base blate. The servomotor should also be attached to the shoulder using screws. Bearings should be put in all the prevised spots, after which the assembly should look like Figure 25.

Figure 25: Assembling the shoulder.

Next, the first pulley system should be assembled. M6 threaded bars are put through the bearings. The squared shape joint fits through the hole in the big pulley, which is necessary to transmit the torque to the humerus. At this point, the humerus should also be attached to the shoulder. The motor should be screwed to the small pulley, and the assembly should look like Figure 26. The M6 bar in blue should be fastened using two nuts and some washers, to prevent the bar from moving along its axial axis.

Figure 26: Assembling the humerus.

The ulna, second motor, second pulley system, and motor-to-pulley connection can all be mounted at this point, as seen in Figure 27. The motor-to-pulley connection should be glued to the pulley using superglue and screwed to the motor. A hexagonal joint is used for the power transmission to the ulna. Although putting both motors on the same size of the arm is not necessary, it is recommended for cable management.

Figure 27: Assembling the ulna.

The X-shaped structural pieces are to be screwed onto the humerus and the ulna in order to strengthen the arm against torsion. Two tensioner pieces need to be screwed onto the shoulders, an M6 threaded bar should be placed in the slits and fastened in place. It’s important that this bar pushes against the T5 belt, to maximize tension in it. The final assembly of the arm should look like Figure 14.

A complete list of used parts is given in Table 6. A complementary technical drawing can be seen in Figure 15.

Gripper

Connection pieces were modelled in Siemens NX to link the tentacles to the system.

This design (Figure 28) includes a connector (Figure 29, Figure 30, Figure 31) that inserts into the silicone and features an internally matched shape to prevent the tentacle from coming loose. The connector is also designed with an integral through-hole to facilitate air passage. This connector piece is the fastened to the tentacle using a zip-tie, to ensure airtightness.

The second modelled part is the structure designed to support all the tentacles. It consists of four inner half-hourglass shapes engineered to perfectly mate with the corresponding hourglass geometry of the tentacle connectors. The assembly is secured by screwing the remaining outer half-shells together, effectively closing the structure. The screws used here are M3x8 and they are screwed through the top piece of the 2-part connection tentacle connection piece. This top piece holds the tentacle securely in the inner rail-like structure of the gripper.

This design allows for easy interchangeability in the event of a tentacle failure, and it provides ample space for organized cable and tubing management.

The main gripper piece is designed to be as flat as possible to allow easy printing and further modularity. The way the gripper attaches to the robotic arm is by the upright piece on the gripper. It is screwed into the main piece with M3x16 screws. The upright piece has a hole at the top. When connecting the gripper to the arm, the bar at the end of the top arm part. This bar connection serves as a hinge that is not actuated, allowing the gripper to self-level and remain in a downwards position no matter the position of the arm.

Figure 28: Assembled view of the gripper.

Figure 29: Assembly for the gripper, from top to bottom: the upright connector, the tops of the tentacle attachments, the main gripper piece, the bottom tentacle attachment.

Figure 30: Technical drawing of the assembly of the gripper, including the four tentacles.

Figure 31: Technical drawing of the assembly of a connector piece and a tentacle

Pneumatic system

To allow for interchangeability, a push-fit system is used for the pneumatic system. The assembly is seen in Figure 32. This means that all tubes can be easily disconnected in case of issues. The air pump is plugged into an external outlet and turned on for the entire use of the robot. In a final product this would be replaced by a pump that is actuated by the microcontroller. The pump is then connected to the first tube by use of a screw on push fitting. This first tube leads to a pressure regulator, the ‘AFR2000’, that is chosen so the pressure in the system never exceeds the safe limit for the tentacles. From this regulator another push-fitting leads tube that is connected to a 3/2 normally closed solenoid valve, the ‘Heschen 3/2-NC-Solenoid’. On the other side of the solenoid is another tube that leads to a pressure regulator. From the regulator a tube is connected to push-fit split connection to split the singular tube into 4 connectors which lead to the gripper connection pieces. All the fittings used are standard parts and the tubes are 6mm plastic tubes. The only custom part in this assembly is the gripper connection piece, which is also part of the gripper assembly and thus listed in the assembly section of the gripper.

These components are then secured to the robot in the cable management system. The tubes are tied to the arm pieces using zip-ties and the valve and regulator are stored out of sight in the box beneath the robot. The regulator is accessible on the side of the box to read the pressure in the system and adjust the maximum pressure that is let through to the tentacles.

Figure 32: Assembly of the pneumatic subsystem.

[1] Form X., (z.d.). Dragon Skin.https://www.formx.nl/products/silicone1/dragon-skin-series/dragon-skin-20---1kg.php

[2] Alumeco. (n.d.). Food contact materials. https://www.alumeco.com/certificates-and-declarations/food-contact-materials/

Mechanical systems

Arm

While the mechanical arm isn’t as innovative as the gripper or the pneumatic system, it’s still an important aspect of the prototype. It is needed for the demonstration, and thus for the proof of concept. The main function of the arm would be to lift targeted objects off the ground as well as being able to support the weight of the gripper and tentacles. Additionally, it is used to showcase that the gripper is strong enough to grip on to an object under a multitude of angles. To do this, it was calculated that the motors should be able to practice 2,40 Nm and 0,79 Nm on the bottom and top joint respectively, assuming a healthy safety margin of 1,5. The chosen motors, MG996R servo’s, couldn’t deliver this torque through direct drive, so a pulley system was implemented with reduction ratios of 4 and 1,5 respectively. This ensured that the motors wouldn’t fail during lifting, if the picked-up load was much higher than predicted.

Because of logistic issues, the right T5 belts for the motors couldn’t be bought and delivered in time, so they were to be produced by the students. To create a loop, the exact number of teeth the belt should have, could be counted on a strip of T5 belt, then cut at that point. A small piece of belt could then be glued on the flat side of both ends, to connect the two together, creating a loop. For this, superglue was the optimal choice. An alternative method concerning drilling a hole through the two ends and connecting them with a screw and a nut was tested but deemed less ideal due to the risk of the screw getting caught up in one of the pulleys.

For the bigger pulley system, a crossbar with adjustable height was designed to function as a tension wheel. This was proven to be necessary in the testing phase to ensure the belt wouldn’t slip. For the smaller pulley system, a similar idea could work, but it was not implemented as testing proved it wasn’t necessary, as the belt had just been manufactured a bit better than the other one.

The final CAD of the arm is presented in Figure 14. Additionally, all CAD files can be found in the project repo.

Tentacle

The tentacles are the main innovation of the system. The tentacle must withstand the pressure from the air pump without popping or permanently deforming. The way they deform is dependent on the shape of the internal air chamber. Several iterations were made for the inner chamber of the tentacle to deliver the correct bending of the tentacle at the delivered pressure. The selection and design of these inner chambers was done mainly by FE analysis. This was needed because casting tentacles without any calculations would lead to waste of product and take a long time, as each printing and casting takes around 12 hours. To see the results of the FE analysis, refer back to section ‘Concept generation’.

For the actuation of the final inner chamber a pressure is needed of 0.5 bar. This means that a simple air pump can be used to actuate the tentacles. To allow the air to escape the tentacles at the correct time, valves are used. Because it is a single chamber tentacle, a 3/2 solenoid valve is sufficient. The solenoid allows air to flow towards the tentacles when it is powered and lets the air escape when it is not. This ensures that, in case of a malfunction, the air will be let out of the tentacle, so the tentacles are not actuated in case of failure. This also allows for easy implementation of control systems, that can cut operation in case of failure.

Gripper

The full design of the system contains the tentacle, pieces to connect the tentacle, the tube, the valve and the air pump.

The cad files and explanation of different parts of the gripper can be found in the ‘assembly’ section above.

For the size of the main gripper part, some testing was done. This was done because if the tentacles are too close together, they will clash and not pick up anything. This gave a distance of 7.5 cm between the tentacle and the centre of the gripper as the optimal distance for picking up small fruit.

The final assembly of the gripper is presented in Figure 28. Additionally, all CAD files can be found in the project repo.

Circuitry & sensors

This subsystem is composed of one valve, two motors, two potentiometers, one MOSFET, some resistors, a diode, a button and the Arduino Uno. The motors and the valve are powered independently with power supplies because of their needed voltage (respectively 12V and 6V), which can’t be supplied by the Arduino Uno.

Each potentiometer consists of a supply pin (VCC), a ground pin (GND), and an output pin (wiper). Both potentiometers are connected to the Arduino's 5V rail and ground. The output pins are wired to analogue inputs A6 and A7, allowing the system to read the variable voltage and map it over to a PWM duty cycle to send to the servo motors with pin D5 and D6. This duty cycle determines the motor angle.

The MOSFET acts as a switch, controlled by the Arduino, for the 12V power supply. The gate is connected to digital pin D7 through a 1kΩ resistor. When D7 is driven HIGH (5V) upon receiving a signal from the button connected to D2, the MOSFET becomes conductive. This allows current to flow from the 12V source through the valve to ground, thereby activating it and letting the air flow through.

The schematic of the implemented electronic system is seen in Figure 33. This circuit was first implemented and tested on a breadboard, and afterwards soldered into a PCB, achieving a cleaner, more compact look and with more reliable connections. Wires were cut to the minimal length and winded together if possible. The electronics were mounted onto a separate board together with the valve to minimize the mess of cables and tubes around the robot, as seen in Figure 34.

Figure 33: Electrical scheme of the used circuitry. Made with circuit-diagram.org.

Figure 34: Control/electronics board, including the solenoid valve on the left and the Arduino Uno on the right.

Software and control

The code of the robot allows us to control the motors and the valve through a remote to move the arm to a certain position and grasp the object precisely. For the tests beforehand, the computer was used directly as a remote.

Arduino code consists of three main parts:

Definition & Tools

Figure 35: Arduino code for importing libraries and defining variables.

In this opening section, seen in Figure 35, the code prepares all the necessary components for the robot.

1. Libraries: It imports Servo.h to control the motors and EEPROM.h to allow the robot to remember its state even after being turned off.
2. Pin assignment: It defines specific pins for two potentiometers, two servos, a MOSFET to control the valve, and a physical button.
3. Variables and objects: It creates the servo objects and establishes the variables that will store sensor data and the robot’s current status, such as whether the tentacles are active.

Initialization

Figure 36: Arduino setup() code.

This section, seen in Figure 36, runs once when the robot is powered on to ensure that everything starts in the correct state.

1. Hardware setup: It attaches the servo objects to their control pins and configures the button using an internal pull-up resistor to ensure stable readings.
2. Output states: It sets the initial power levels for the valve and the constant 5V pin.
3. Memory recovery: The robot reads its permanent memory from the EEPROM. If the robot was in an active state when it was last turned off, it automatically restores that state, allowing the robot to remember its previous position.

Main Loop

Figure 37: Arduino code of the main loop.

The loop section, seen in Figure 37, handles the continuous logic of the robot by repeatedly executing two primary functions.

1. Servo management: The “handleServos” function constantly reads the position of the potentiometers, maps those values to degrees, and moves the arms in real-time.
2. Button management: The “handlePhysicalButton” function acts as the interface logic. It checks if the button is pressed, toggles the valve state, updates the hardware, and saves the changes in the memory. It also includes a debounce delay and a safety loop to ensure one click only triggers one action.

The assembly of the different sub-systems can be found in the ‘Assembly’ section above. This section of the report focuses on the final connections between those subsystems that need to be made in order to have a fully functioning robot. The arm can be attached to the gripper by putting the bar at the top of the small arm through the hole at the top of the gripper connector piece. This connection can be seen as a loose hinge which allows for the gripper to freely hang down no matter the position of the arm, acting as a self-balancer.

The pneumatic system is then connected by plugging the tentacle-tubes into the tentacle attachments. The tubes are then secured to the arm and gripper structure with zip-ties to function as cable management. The air intake of the pressure regulator is the connected to the air supply. Power supply cables for the motors and for the valve are then connected, and the cables that go from the Arduino to the motors and valve are connected as well.

When all these connections are made and checked for security, the robot can be operated safely.

A montage of the prototype was made for the mechatronics course and can be seen on YouTube: https://youtu.be/5TBewZ9wyTY

While the prototype works, a lot of optimizations can be done. For the arm, both pulley systems should have a fastener to ensure tension is high in the belt. The shoulder tends to move a bit when picking up loads, which is counteracted by putting a weight on top of them, when it would be better to mount more L-profiles to it. A better T5 belt should’ve been bought, as the DIY T5 belt sparks little confidence in us, despite it working fine for now. Some more material could probably have been removed from the arm without impacting the structural properties too much, although this could be considered nitpicking.

A better pump could’ve been found or even made custom for this project. Sadly, the many different options tested out failed and the lab’s compressed air supply was used. The electronics could’ve been hidden more discreetly in the design of the arm.

The tentacles, while sufficient for a prototype, weren’t as durable as hoped for. This is due to non-optimal casting techniques, including the lack of a vacuum chamber to prevent bubbles. Also, because of practical reasons, the casts had to be moved around a lot while still in the mould, which couldn’t have been great for the consistency of the tentacle.

In a world where the effects of climate change are felt and seen, it is essential to consider the ecological impact of products. Therefore, it is essential to consider the environmental impact of automated machinery that is supposed to replace humans. If this robot would hit markets, one would want it to be as sustainable as possible. The used materials should be able to be repurposed or recycled.

Repairability was already a great consideration in the design of the robot, lending to a design with connectable tentacles and tubes. In case of a local failure of the system, the affected parts can be disconnected and replaced with minimal impact on the system.

In the prototype, the air pump is continuously turned on during robot operation. This is done because the commercial air pump cannot be actuated by the Arduino. When looking at energy saving methods, integrating the air pump into the circuit and turning it ‘On’/’Off’ at the exact times air is needed will limit the energy used by the system.

Bill of materials Excel: BOM.xlsx

Drive link: Project_Repo