We only had a 10-week quarter and limited real-robot time since we shared one robot among 7 teams. If we had more time, we'd love to implement the following:
Test with more objects in the YOLO dataset, and test with more facial recognition subjects
SAM3 implementation for better detection
Optimize robust navigation (mapping + obstacle avoidance) for real-life
Troubleshoot the eGPU so we can test our GraspNet code in real life