P18227: Soft Robot 2.0
/public/

Subsystem Build & Test

Table of Contents

Team Vision for Subsystem Level Build & Test Phase

During this phase, the ComQuaT team planned to demonstrate functionality and share test results for each subsystem, including

We achieved these objectives except

Opto-isolator Circuit Test

Test was performed by connecting a switch between the valve and the power supply on the isolator circuit. After setup, the MOSFET was found to be unnecessary. The valve and switch were connected to the output side of the optoisolator. The switch was then toggled several times in rapid succession. This produced voltage spikes over 0.8V and up to 3V which is high enough to damage the input of the Arduino pins. The following figure shows these voltage spikes measured on the oscilloscope:
Voltage Spike Measurement

Voltage Spike Measurement

After consulting with Dr. Indovina in the electrical engineering department, a flyback diode was added across the valve. This reduced the voltage spikes on the input to 0.2V maximum. This can be seen in the following oscilloscope measurement
Reduced Spike Measurement

Reduced Spike Measurement

Power supply Variation Test

This test was performed in order to ensure that fluctuations on the power supply would not cause resets on the Arduino. A 25V power supply was connected to the 5V step down regulator and 12V step down regulator. The 12V step down regulator was used to power the RC motor driver. The oscilloscope was then attached to the output of the 5V regulator. Large voltage swings of 2V in amplitude were measured when the motors were started. This would cause resets on the Arduino if connected to the 5V line.
Reduced Spike Measurement

Reduced Spike Measurement

To fix this issue, the Arduino will be connected to a separate power bank to ensure no resets while operating.

Arduino Valve Switching Test

The Arduino was connected to a prototype of the isolator circuit board. The output of the isolator circuit was then connected to 5 of the valves. The Arduino code was run to trigger the valves for 0.5 seconds every 1.5 seconds. They were programmed to have a phase difference of 0.25 seconds. It was expected that the valves could be heard triggering in this pattern. Initially, only one of the valves was heard firing. This was fixed by adding a BJT to the input of the isolator circuit and reducing the input resistance. The final circuit configuration was then given by:
Revised Valve Control Circuit

Revised Valve Control Circuit

After implementing these circuit changes, the Arduino code was run. When the Arduino code was run, the valves could be heard triggering in this pattern. The following video shows the Arduino powering the circuit and the clicking of the valves can be heard: Arduino Valve Switching Video

Expected Result: Hearing the valves clicking in the pattern
Observed Results: Valves were heard clicking in the expected pattern

RC Control Test

The code for the RC was written on the Arduino. The USB host shield and motor driver shield were connected to the Arduino. The XBOX 360 controller dongle was connected to the USB host shield and connected to the controller. The code was structured to measure the displacement of the left and right controller sticks and scale them between 0 and 255.
RC Forward Control

RC Forward Control

RC Reverse Control

RC Reverse Control

RC Clockwise Rotate

RC Clockwise Rotate

RC Counter-Clockwise Rotate

RC Counter-Clockwise Rotate

RC Forward Movement Veer Left

RC Forward Movement Veer Left

RC Forward Movement Veer Right

RC Forward Movement Veer Right

RC Reverse Movement Veer Left

RC Reverse Movement Veer Left

RC Reverse Movement Veer Right

RC Reverse Movement Veer Right

Motor Driver Shield for Arduino

Motor Driver Shield for Arduino

The motors were connected to the motor driver with an additional voltage source for the motors. The XBOX controller was then used to input the speed controls. When the controls joy sticks on the the XBOX controller were moved, the motors varied in speed and direction as expected. This confirms the functionality of the RC system. These motors will be mounted on the frame, ideally with 2 additional motors in parallel for differential drive.
Expected Result: Motors would move in the directions specified in the control diagrams with the specified speed
Observed Results: The motors moved in the proper directions and with reasonable angular speeds for each shaft

Pneumatic Supply System Test

Three tests were performed on the 3-way valve supply system to estimate fill rates and ensure head losses through valves wouldn't be overwhelmingly high. Test 1 connected both compressors in parallel directly into a balloon to maximize flow rate and minimize line losses. Test 2 added the 3-way valve and manifold system between the compressors and balloon. Test 3 re-configured the compressors in a series configuration to maximize delivered pressure.

Parallel Compressors Directly Into Balloon

Parallel Compressors Directly Into Balloon

Parallel Compressors Into 3-Way Valve

Parallel Compressors Into 3-Way Valve

Series Compressors Into 3-Way Valve

Series Compressors Into 3-Way Valve

When adding the 3-way valve/manifold assembly a drop in flow rate is observed. We suspect the 3-way valves have relatively high head losses due to their narrow internal channels. It appears from the tests that the series configuration inflates slightly faster than the parallel, which was not expected at first. The high pressure configuration reduces each individual pump's load and allows more air to be pushed through the 3-way valve. In a full configuration with 12 lines splitting the flow we expect line losses through individual valves to be significantly less than during these tests, and a parallel configuration may work better. Fortunately the nature of flexible tubing and barbed fittings allows for easy, repeatable re-configurations to best suit our needs.

Estimated Power Consumption

Chart of Estimated Power Consumed by our Equipment, individual and groups

Chart of Estimated Power Consumed by our Equipment, individual and groups

The following chart shows the power consumed by several pieces of our equipment. If the value is Italicized then the value was assumed based off of testing done to one of the items. We currently do not have the motor for the Stepper so the results for that are unknown. Ultimately, we are looking at 1.5-2.0 A for constant current drawn. This is because certain pieces of equipment will not run at the same time, or operate in a pattern. An example would be that we will not run the motors for the RC while performing the actions for the robot.

Leg Chamber Fabrication

Best Print Result to Date

Best Print Result to Date

3D Printing leg chambers that match the requirements set by the ComQuaT team has been a significant challenge. A prototyping SME from the Construct, Mike Buffalin, remarked that the leg subsystem of our design could be its own MSD project and would still be very challenging. A list of problems encountered so far includes...
Print Surface Shift - Cause Unknown

Print Surface Shift - Cause Unknown

A new geometry was created for the leg chambers in order to minimize issues with the overhang angles. Overhang angles of less than 30 degrees are very difficult to print and are inconsistent at best. The circular design of the original leg chambers means there are several areas in the design where overhang angles are less than 30 degrees. The hexagonal geometry allows the filament to form a solid base to build on and ensures that all overhang angles are greater than or equal to 30 degrees. A trade off of this design is the flat portions at the top of the print that are more difficult to completely cover than with a rounded top.
Hexagonal Model on Slicing Software Buildplate

Hexagonal Model on Slicing Software Buildplate

Original Model on Slicing Software Buildplate

Original Model on Slicing Software Buildplate

Another challenge of fabricating inflatable legs is interior support. In solid objects 3D printers will fill the interior of the part with support structures, this allows the walls to remain stable and supports the top surface. Any traditional infill will block the flow of air to the leg chambers, leaving only two options: No Infill or Removable Infill. Below is an image of a print-in-progress that shows the dissolvable PVA (Polyvinyl Alcohol) support structures on the interior of the model. These support structures are intended to maintain the quality of the print's interior as well as dissolve in water after the NinjaFlex has been set.

Hexagonal Geometry Model with Soluble Supports

Hexagonal Geometry Model with Soluble Supports

Critical Decision Making The legs of the ComQuaT robot system are integral to the success of the project. The decision to use 3D Printing as the preferred method of fabrication for our design relied heavily on the project from University of California - San Diego, and unfortunately the levels of details included in the publication of the UCSD project, as well as the time and resources available to our team, are not sufficient for us to produce leg chambers that are similar at this time. For that reason we have begun to leverage past and present projects' research into silicone casting inflatable muscles through Dr. Lamkin-Kennard.

Silicone Casting

As an alternative method of fabrication, we have decided to cast the legs out of a flexible silicone. The Dragonskin material from Smooth On can be adjusted to achieve the desired rigidity. Two methods of casting are being explored using different molds. On recommendation from the Robotic Otter team, we are casting leg chambers in multiple steps. Each half of the leg will be cast using a vacuum chamber to ensure we minimize air bubbles in the mixture. The two halves will then be joined, again under a vacuum. Alternatively we will attempt to cast the entire leg as one piece, utilizing dissolvable 3D printed mold features for removal.

Risk and Problem Tracking

Snapshot of our updated Problem Tracking Chart

Snapshot of our updated Problem Tracking Chart

The Risk Assessment Chart has not changed since our last review.

Snapshot of our updated Risk Assessment Chart

Snapshot of our updated Risk Assessment Chart

Draft Layout for ImagineRIT Poster

Proposed draft of project poster for ImagineRIT

Proposed draft of project poster for ImagineRIT

Plans for next phase

By our next demo in the last week of March, we plan to have achieved:

Individual Plans:

Role Individual Plan
Project Manager Conor McKaig
Lead Engineer Zach DiLego
Electrical Engineer Cameron Taylor
Software Engineer Sean Bayley
Hardware Engineer Zach Hayes
Purchasing & Materials Marie McCartan
Comm. & Customer Contact Jamie Mortensen

Home | Planning & Execution | Imagine RIT

Problem Definition | Systems Design | Preliminary Detailed Design | Detailed Design

Build & Test Prep | Subsystem Build & Test | Integrated System Build & Test | Customer Handoff & Final Project Documentation