Moisture Content % Robot

Concept

To create a system to automate the detection and recording of moisture content in wooden boards for self-shaping panels.

Relevance

While the design of self-shaping bilayer wood structures can be integrated into  computational workflows, there is still a high level of dependence on real-world material properties to predict outcome curvatures. These properties include layer thicknesses, wood species, and year-ring angle, all of which only need to be recorded once. Another important property is the moisture content, which changes all the time as it equalizes with the humidity of the air around it. In order to gain an understanding of the potential curvature space for a given set of stock, constant monitoring and recording is required. Manual recording for large sets of stock is prohibitively time consuming, while state of the art methods such as CT scanning are prohibitively extensive. For these reasons we set out to create a cheap automated system which can sense and record the moisture contents of wooden boards.

This robot was designed and fabricated by Alex Reiner, Edgar Schefer and Alina Turean as a part of the Computational Design and Digital Fabrication seminar in the ITECH master program.

Robot body

The body of our robot is composed of 3 main elements:

• A platform with integrated living hinge wheels
• A pencil holder
• Conductive prongs

All of the components were 3D printed.

Electronics

We used a mix of Arduino Starter Kit parts as well as additional components:

From Arduino Starter Kit

• 1x Arduino Uno R3
• 1x Small breadboard
• 3x 10MΩ resistor
• 2x sg90 Mini Servo Motors

Additional components

• 4x 28byj-48 Stepper Motors
• IR Sensor
• Wires (Female to Male, Male to Male). 

Prototyping tools

• Rubber Bands (for extra grip on 3d printed wheels)
• Super glue
• Craft knife
• Screw driver
• Screws
• Wood Board for moisture sensing
• Moisture Meter for calibrating the Arduino ohm meter system

Moisture Content Robot Functions:

The Moisture Content Robot has three main functions. The first is to measure the moisture content along the board - this data is then stored within an excel sheet which can later be implemented into a curvature prediction model. The second function is the physical material manipulation - where the robot maps the moisture content on the board using a high amplitude to represent a high moisture content and a low amplitude for low mc. The final function is the physical sorting of the boards using both the data bank and the drawn amplitude in communication with the computational model as guides for the placement.

Moisture Content Measuring

To gather moisture content measurements we created our own moisture meter by measuring the resistance within the wood. Since we know that current is equal to the volts divided by the resistance - the function for resistance can be easily extracted. We were then able to find many tables correlating the resistance and moisture content for different wood species, however for our amplitude we currently use the resistance value as it has an exponential rise in resistance the drier the wood is - thus giving us more variation in the amplitude.

Physical Material Manipulation

To create a better robot to user communication we translated the measured Resistances into the physical space through normalizing and multiplying by the higher and lower bound of the reachable pen holding servo motor movements, resulting in an amplitude informed by the wood moisture content.

The code is fairly straight forward:

IR Sensor

The robot first sets its default state and enters with an if statement triggered by the the IR sensor, as long as the robot does not sense an edge/the end of the wood board, the code continues to run. When the robot senses the edge - it stops.

Wood Moisture Sensor

The robot is equipped with two prongs installed on a servo. Every few seconds the code initiate the wood moisture sensing state, where it triggers the front servo to turn 90 degrees to place our moisture sensor(the prongs) on the wood board. It receives the resistance and normalizes the value to define the range of our sine wave, ending with the servo moving to its default state again.

Stepper Motors

The stepper motors control the main four wheels that move the robot along the board. The motors stop while the wood moisture sensor is activated. Once a value has been taken, the motors initiate movement by a set distance and enters a while loop until it reaches this settled value.

Servo Motors

The wood moisture sensor moves up and down using a servo motor. However, our pencil holder also rotates with a servo motor. The servo movement for the pencil holder is defined by 3 conditions, which are mainly distinguished by a direction value. If the direction is positive it moves to our upper bound of the amplitude by one step,  if it’s Negative to the lower bound and if it reaches one bound, it will just flip the direction. The last if statement, in regard of our IR sensor, prevents our robot from falling of the board. By just breaking out of the loop and initializing a reverse movement.

Physical sorting of the boards

The moisture information gathered could also be logged in a larger dataset, and used to better understand the self-shaping design space of a set of stock material. The curvature gradients achieved in self-shaped bilayer panels can be done through sorting the active layers by moisture content of the stock. The boards with a lower starting moisture contents have a smaller potential curvature range than those with higher starting moisture contents.

We were successful in creating an automated system for sensing and recording the moisture content of wood boards but many aspects leave room for improvement. The accuracy of our moisture sensor was limited by the large range of resistances in the wood between wet and dry states. The pattern encoded by the marker on the board was difficult to read and the relationship between amplitude, wood resistance and moisture-content required significant calibration.

The data recording aspect of our system would need to be re-thought in the future. A system for recording and keeping track of moisture data over time, directly informing our design tool, would be superior to one which encodes temporary data by making permanent marks on the wood.

Additionally, due to time spent on calibration, tuning and maintenance, this system would only become superior to manual data recording at scales much larger than a single board. This is why we began to envision a secondary, two-axis system, which could lift the mobile robot between boards on a rack. This would allow for the recording of moisture data for an entire set of stock, and give up-to-date information to our design tool for self-shaping curvature prediction.