Conclusion

Discussion

Ultimately, with Dish-n-Dash, we were able to create an end-to-end system that detects, grasps, and sorts objects as compostable or dishware. Beyond just at schools, this technology can be generalized to different regimes where we need to pick, place, and sort items based in relatively deterministic environments. Our sensing component (DETR + camera calibration + transforms), planning & motion components (IK) together create the robust system that is Dish n Dash!

Challenges

• Calibrating the robot's motion to ensure precise and reliable sorting was time-consuming.
• Classification errors occurred with items like bananas in unusual orientations.
• Hardware limitations, such as short cable lengths, led to occasional collisions between the camera and gripper.
• Handling mixed items in bowls required simplification of grasping strategies.

Future Revisions

Given more time and resources, we propose the following improvements to enhance Dish-n-Dash:

• Incorporating advanced grasp generation techniques, such as Grasp Anything, to better handle individual items in mixed bowls.
• Expanding the classification model to include more waste categories for better sorting accuracy.
• Implementing retry loops for error handling, ensuring that items are picked up and sorted correctly every time.
• Upgrading hardware components to reduce latency and improve reliability in robotic movements.

Final Thoughts

The Dish-n-Dash project demonstrated how robotics and computer vision can address real-world challenges in waste management. With further refinement, the system has the potential to make a significant impact in educational and public spaces, promoting sustainability and reducing waste contamination. We look forward to seeing how this concept evolves in the future!