Blog

Plan of Action :

1. Collect materials and build frame, including funnel
2. Purchase and assemble throwing mechanism
   • Design control and compression of spring piston
   • Find mapping between spring and distance thrown
   • Control with software to angle and shoot
3. Integrate computer vision
   • Use Kinect to gauge depth to targets and human shapes

Weekly Updates :

Week 1 : We formed our groups and started coming up with designs for a robot that can play catch.

Week 2 : We came up with the main design that we will be working off of for the robot. We presented the idea for the robot, named FunnelBot, to the class. A generalized 3D design for FunnelBot was produced in SolidWorks by Nick Bira, shown below:

Week 3 : We started testing out different components of the design. Nick Bira laser cut a cardboard version of the funnel using a repeating, octagonal CAD file he designed. This design includes holes for future testing of potential air vortex functionality using computer case fans. In the lab, we taped together the pieces and tested out if the funnel design would work as expected. We also debated if we were going to make the funnel more circular or leave it as it is. The shape is adequate for now, and we intend to test various ways of accomplishing the vortex. Alex and Courtney worked on the design for the aiming portion of the firing mechanism. Utilizing a BeagleBone Black and a Seeed motor bridge cape from a previous project along with Nick's laser cut gears, we were able to test out our gear design for the horizontal and vertical aiming mechanism. For the testing gears we attached the cardboard gear to timing belt pulley which then attaches to the Shinano SST42D1020 stepper motor. After wiring up the motor to the motor bridge cape we built a small Python program to control the motors to move the gears into roughly the desired direction. Since this is a test of concept with cardboard gears we are allowing for a larger margin of error. Below is an short video of the gear system prototype:

Week 4 : This week ended up being the week of prototyping the shooting mechanism, purchasing correct parts, and getting the vision of the FunnelBot working. This week was a challenge to get all of the parts of the projects we planned on getting this week done. Nick Bira was gone at the Robosoft conference in Italy until the May 1st, so he was busy for most of the week and was unable to physically work on the project. Alex and Courtney worked on the project through the flu, so most of their work this week's work was designing and programming. Alex worked on getting the vision for the FunnelBot working utilizing a GoPro, two wireless video transceivers, two LiPo batteries, and a TFT display, as can be seen below. He spent most of his time savaging parts from previous projects and making connectors to connect everything together. Courtney worked on developing the shooting mechanism. The current design utilizes a gear system to compress the spring using the shaft of the piston, that is geared to match the compression gear. The compression gear that is being used does not have teeth around the whole gear. This is done so that the spring will be fully compressed once the gear reaches the last available tooth. The part after it is smooth which allows the piston to be fired by the spring. The parts for the design are in the process of being ordered and shipped to begin the assembly of the shooting mechanism.

We expect to get in the parts for the final moving aiming assembly by the end of this week, along with the parts for the firing mechanism early next week. From this we will begin piecing the parts together and assembling a frame to accommodate all of the parts.

Week 5: Testing all remaining concepts, and bringing the parts together for the midterm demo

This week we assembled several important systems. First, the rotational mounting system for our piston arm was assembled from the Vex robotics turntables we previously purchased and just recently received. The servos that connect to the turntables lacked shafts that could reach further into the turntable, so while we completed the two-axis build with laser cut wood, we are waiting on these shafts to have full control. We also programmed the servos in preparation. Continuing with the funnel, we re-cut the pieces from heavy-duty cardboard, building on the design and adding reinforcing circumferential components. We stuck with cardboard after initial tests with wood, deciding that manufacturing time, weight, structural vulnerabilities, and ease of construction were all too great with wood to justify using it. Cardboard, when reinforced properly, should stand up to the stresses of our application.

We gathered together the vision system from the previous week in preparation for our demo, and went about working on the piston firing mechanism. After some research and discussion, we settled on a gear and linear piston system, in which the piston and spring is compressed by turning a motor-powered gear. After reaching full compression, the gear is missing teeth on one half, which allows the piston to then rapidly return to its state, delivering the impulse to shoot the ball. We are currently experimenting with components from a Nerf gun, and will be creating an enclosure and method to mount it onto our robot base soon.

Our individual contributions are:

• Nick:
  • More CAD of the turntable attachment bases, allowing us to connect them to each other and create the pivot system for the piston
  • More CAD to reinforce the previous funnel design
  • More laser cutting of heavy duty cardboard to allow for a more rigid and durable funnel design, made with cardboard
  • A Challenge: Initial construction of a wooden funnel was attempted, before settling on cardboard after reevaluating wood properties and their failure for desired robot properties.
• Alex:
  • More code to run the servos that will connect to the turntables once the longer shafts arrive
  • Vision system testing
  • Assembled the turntable from laser cut wooden parts
  • Challenge: Shafts for servos were too short, had to order more
• Courtney:
  • Wiring together multiple PC fans for use in the vortex formation
  • Testing a PC power supply for running these fans simultaneously
  • Website maintenance
  • Challenge: Finding components at Home Depo, and multiple runs to find correct screw size (M4 instead of M5)

Goals for next week:

• Integrate servos into turntable assembly once we receive the longer shafts
• Add more fans to the funnel and test vortex formation
• Create enclosure for piston assembly and test throwing distance abilities

Week 6: Work towards a functional piston assembly, purchasing remaining components, transferring controls system to RC transmitter.

This week our efforts were focused on completing the piston assembly, testing its efficacy with the beach ball, and transferring the controls from the phone application to a more reliable RC transmitter. Nick designed and 3D printed several components to integrate the NERF piston from last week into a piece that could be attached to the rotational assembly. One of the components is a fist attached to the end of the piston arm, to increase surface area/contact with the beach ball, as well as adding mass and general charisma to the assembly. Courtney and Alex decided to work on the controls system while waiting on turntable parts. The robot is now able to be controlled by a DX7 RC transmitter which gives the robot increased accuracy for the aiming controls. Once the parts come in, estimated for later this week or early next week, the controls and the turning system will be combined. The main issues that we have had this week were mainly due to waiting on shipping and having parts not arriving with all of the needed components.

Our individual contributions are:

• Nick:
  • CAD of the Piston Assembly, combining the NERF Piston with our spring and additional attachment points
  • 3D Printed components (mutiple revisions based on 3D Printer tolerances)
  • Assembled the parts in the picture shown below
• Alex:
  • Got interface between RC receiver and Beaglebone Black working in python
  • Set up demo between Beaglebone Black and servo (waiting for servo shafts).
• Courtney:
  • Found extra transceiver for the improved controls
  • Wired up components for the control system
  • Calibrated the DX7 RC transmitter

Goals for Next Week:

• Integrate servos into turntable assembly once we receive the longer shafts
• Add more fans to the funnel and test vortex formation
• Finish the piston assembly with the gearbox control system
• Complete frame and base

Week 7: Worked towards assembling all of the parts together on the frame, added fans to the funnel, finished piston and gear box assembly

This week we were finally able to get the aiming portion functioning again since the parts came in. We were also able to assemble a majority of the frame. The remaining elements of the frame are supports to make the finished robot more stable and for adding a "shell" to protect the internal components and make the robot more presentable. Since we have the base of the frame completed, we were able to attach the aiming and firing mechanisms to the robot. Alex was able to continue work on the code for both of these mechanisms. Courtney worked on completing the funnel catching component of the robot by adding the fans and installing a mesh across the fans to decrease the chance of the ball getting caught. One issue that we have run into is making the transceiver communicate with the BeagleBone, since the transceiver works 5 volts and the BeagleBone works at 1.9 volts. We were able to work around this in the earlier builds because we could work directly with the transceiver to test the motors, but now that we have to work with multiple motors, we have to go through the BeagleBone cape. For a short term fix we are going back to the phone application while the part for the transceiver is shipped.

• Nick:
  • Finished assembling the firing mechanism
  • CAD of the gear box system for the firing mechanism
  • Laser cut parts for the aiming turntable system
  • Assembled the central frame and attached the various independent systems to the frame
• Alex:
  • Assembled the frame and attached other systems to the frame
  • Worked on code for the aiming and firing systems
  • Programmed the fans for control
• Courtney:
  • Assembled the funnel and attached fans
  • Helped assemble the frame
  • Wired up the motors and some of the fans for testing.

Goals for Next Week:

• Make the frame more secure
• Add connecting parts to transport ball from funnel to launcher
• Finish the coding elements
• Fix some controls issues
• Test and refine the firing element

Week 8: Worked towards getting the controls mechanism working again, attaching the vision system, finishing the firing mechanism, connecting fan lights to controller.

This week we worked on finishing assembling all of the parts together and found ways to fix mechanical and software issues that came up last week. We worked on getting the RC controller to communicate with the BeagleBone using a Teensy 2.0. We tried to fix the firing issues by designing and 3D printing a restraint to go on top of the firing mechanism guide the piston and ensure it doesn't veer from it's firing path. The research involving color signaling also occurred, and will be included on this website once the data is processed and plotted. We finished attaching all of the fans and added a mesh so that the ball doesn't get attached to the fans.

• Nick:
  • Worked on HRI research for the robot
  • Fixed mechanical bugs of the firing mechanism
  • Attached intermediary system between catching system and firing system
• Alex:
  • Got RF controller connected to BeagleBone via Teensy to enable Remote Control of motors and firing system
• Courtney:
  • Finished attaching fans to funnel
  • Attached GoPro vision system

Goals for Next Week:

• Add visual elements, masking the internals
  • Spray paint and Wrapping
• Finish firing mechanism improvements
  • Explore Flywheel design in case of piston failure
• Achieve Successful game of catch!

Week 9:

• Finalized Piston assembly and tested it. Due to its less than stellar performance, we moved on to a flywheel design. See below:

After deciding on flywheels, we quickly designed and 3D printed a first attempt.

Once we took further measurements and carefully tested the motors, we reprinted and completed our final throwing design.

Testing of the motors and flywheels can be seen below:

Week 10: Competition and Final Adjustments

Overall What We Learned:

• BeagleBone does not have what is needed to do RC control, so use a Teensy 2.0
• Piston design, though original is not reliably accurate and is hard to have work consistantly
• Don't make changes to something and not fully test it before competition day
• Always have a backup plan for when parts don't work
• Just because you ordered two day shipping does not mean it will arrive on time, sometimes processing the shipment can take a week
• "Plug 'n Play" parts work great when you have the intermediary systems