Computer Integrated Manufacturing final project. My role was coding, wiring, and helping with CADing.
- Raspberry Pi 3B+ or higher (might have compatibility issues with the 5)
- Compatible 640x480px webcam (we did not use a picamera for this, and our camera was YUYV only).
- 4x mecanum drive wheels
- 4x motors (check limitations section)
- motor driver
- 2S LiPo battery (or 2 1S LiPos in series) for motors
- USB Power bank (5V 3A minimum) for pi
- 537T101 McMaster Carr extrude of length x amount of: 28.25in x 2, 16.25in x2, 13.25in x2, 14 x 4, 19.25 x2
- 2 1/8" thick planks of wood cut to these dimensions: 29.825in x 17.825in and 23.825in x 17.825in
- M5 screws
- 5537T007 McMaster Carr drop-in ball nuts
- 5537T935 McMaster Carr L-brackets
- Something to make an electronics casing and motor holder out of (3d printer, laser cutter, etc.)
- 4976N25 T-slot nuts
- Allen keys, drill, file, M5 tap-wrench
Install both the OpenCV and dt-apriltags libraries on the raspberry pi
pip install opencv-python
pip install dt-apriltags
- Only translational pose estimation was used meaning the robot cannot easily turn corners by analyzing the rotation of the paper.
- The project ended up being very rushed because we were given a month under the amount of time we would normally have to build
- The project was initially meant to be scaled up to ~double its current size and have enough structural integrity to support a human. It was also meant to use DW3000 modules and an ultrasonic sensor sonar to make it able to follow a person or make its way to them off a button-press. However, due to the time constraints mentioned above, we settled for april-tag following
- We had to restart creation from scratch many times due to design inputs and newfound material limitations in the classroom
- This was my first time using a Raspberry Pi and my first python script since sophmore year so it is probably not that well optimized.
- The camera should have been elevated higher so the tag could be followed while attatched on someone's back
- The motors we used came from a DIY car kit so we did not have any information about its torque specifications.
ignore my groupmate trying to be humble lol
The back-right wheel was loose in the second video causing the jitter we fixed it later but did not end up taking another video
IMG_2792.mov
IMG_2793.mov
I plan to remake this project on a smaller scale to go forth with our inital UWB tracking plan My autodesk inventor student license expired so I won't be able to convert the CAD files to STLs until I renew it in college.
(Part of the builds I moved from google docs to GitHub starting from 6/19/26)