Autonomous Pick-and-Place System
Built a robotic pick-and-place system using the Franka Emika Panda arm, capable of handling both static and moving blocks.
Highlights
- Object Detection:
- Used AprilTags + OpenCV to detect block poses in real time
- Dynamically adjusted poses to align coordinate frames and avoid grasping failures
- Control System:
- Closed-loop motion pipeline with:
- Pose correction
- Inverse kinematics (pseudoinverse method)
- Gripper feedback for grasp validation
- Closed-loop motion pipeline with:
- Static Blocks:
- Averaged multiple scans for noise rejection
- Used re-scans and ID-verification for robustness
- Achieved 100% grasp success rate in simulation
- Dynamic Blocks:
- Implemented pre-programmed sweeping trajectories for pickup on a rotating platform
- Achieved 71.4% grasp success rate in simulation with minimal computation
| 👉 GitHub Repository | 📄 Full Project Report (PDF) |