We built an autonomous packing robot for quick commerce using the open-source LeRobot SO-101 arm (Won 1st Place!)
retained text
My team and I recently tackled the problem of high labor costs, security and slow picking times in the quick commerce industry. We ended up winning 1st place at a recent hackathon with our proof-of-concept, We built an automated dark store where robots handle the order packing autonomously. For the hardware prototype, we utilized the LeRobot SO-101 (SO-ARM101) 3D-printed robotic arm. Because the LeRobot ecosystem is open-source and backed by Hugging Face, it was the perfect platform to rapidly prototype the teleoperation, data collection, and imitation learning workflows needed to get the arm picking and placing reliably within the time constraints of a hackathon. We put together a 1-minute pitch video demonstrating the setup in action. Watch the demo here: [https://x.com/DsouzaJovian/status/2110028875459932364?s=20](https://x.com/DsouzaJovian/status/2110028875459932364?s=20) I would love to get feedback from this community on the kinematics and our approach. I am also happy to answer any questions about our experience training the SO-101, managing the vision datasets, or the general hardware configuration we used to pull this off!
document fetched 2026-10-14 00:38 · canonical https://reddit.com/r/robotics/comments/1x533ol/we_built_an_autonomous_packing_robot_for_quick