ME405 Term Project
Cal Poly Mechatronics 2 Term Project: Design and create an automated nerf turret that uses an infrared camera to track and shoot a target (the target being a person). The created turrets will compete against one another in a 'deul', where the first turret to shoot their target wins. This project was completed as a group with Scott Brown and Trevor Foley during Senior year at Cal Poly.
Hardware Overview
Our design consists of a large, circular bsase that connects to a mating section through a custom-created ball-bearing (two 3d printed runners with a set of metal spheres between) and a pivot joint attached to the base that allows the nerf gun to pivot up and down. The base is driven by one of the provided Ametek/Pittman motors whose 3D-printed gear mates to the inner teeth of the base (already created; was originally the magazine for the nerf gun before being altered for our purposes) to cause rotation in the yaw direction. The pivot joint was mounted to the base using a set of screws and causes rotation of the gun in the pitch direction through the use of the other provided Ametek/Pittman motor and a mating gear (laser cut from acrylic) that is directly screwed into and attached to the nerf gun with the use of 2 screws. This was all controlled using an STM32L476RG Nucleo. For more info see the 'Hardware Info' section at the bottom.
Software Overview
The dueling robot uses a set of tasks (specified above) called in a main program to run and target an opponent succesfully. The robot uses MicroPython to control two motors and adjust the pitch and yaw axes in order to follow the target (found with the provided thermal camera). For more info see the 'Software Info' section at the bottom.
Results
After creating our mechanical design and troubleshooting our software and code, we were able to successfully track and hit our target multiple times during the dueling competition (once first, and once second). We did have two rounds where our robot missed our target, however the shots were only slightly off from the target (which leads us to believe that heat fluxuations in the room caused by a gathering of people near where the target chose to move to may have been the cause). During the final phase of testing, we found our robot to be very accurate - more so than it appeared in the competition - as it successfully tracked and shot the testing target numerous times without fail (this was with limited gatherings of external heat sources).
Lessons Learned/Recommendations
After completing this lab, we have learned how to create motor controller classes for both direct DC and servo motors and how to adapt this code to multiple other types, as well as how to integrate encoders into code and use them to track/measure position, create cooperative tasking classes, and how to utilize other components (such as an infrared camera) to create complex, moving systems. What seemed to work best for our device was the creation of as few external classes as possible (to avoid memory allocation issues with the Nucleo) and the creation of separate conditional code and Kp/motor constant values for the rotation of our device in different directions (in order to account for the backlash in the gearing system of the pitch axis). Our biggest recommendation for anyone attempting to build on our work would be to redraw the FSM's in order to fully understand the flow of our code, before attempting to improve on or utilize it, as this is a complicated program that uses multiple different classes cooperatively.
