For the week of September 27th – October 3rd, I focused on helping finalize the Iteration 1 system design and narrowing down the CV approach for our 6-DOF robot arm. In addition, I helped finalize the master parts list, including the required components, quantities, and pricing. With our overall Iteration 1 design done, we are now in a better position to move forward with more implementation-related work.
One of my main areas of research was determining the best approach for AprilTag detection. After looking over documentation and comparing OpenCV-based AprilTag detection with using a YOLO26 model, I determined that OpenCV is the better choice for our application. Since OpenCV already has a library dedicated to detecting AprilTags, a YOLOv26 model trained on these tags would be less appropriate. OpenCV should also provide a simpler and less computation-heavy on the NVIDIA Jetson Orin Nano while integrating well with the Intel RealSense D455. This approach will allow us to focus the vision system on reliable AprilTag detection and tracking without adding unnecessary model-training time.
I also continued working with the team on the overall Iteration 1 design and component selection. With the parts list now largely finalized, we have a clearer picture of the electrical and mechanical components required for the system. This will help us move toward integrating the camera, computing hardware, servo motors, and other components into the final design.
For next week, I aim to:
- Prototype OpenCV AprilTag detection and tracking on the NVIDIA Jetson and Intel RealSense D455.
- Start configuring motors and power supply once they come in.
- Continue refining the Iteration 1 design and preparing components for system integration.