Introduction to Robotics — University of Pennsylvania
Cube Stacking Competition
Python
ROS
Inverse Kinematics
Computer Vision
Motion Planning
A competitive robotics project from the Introduction to Robotics course: autonomously detect, grasp,
and stack cubes in the correct order using a robotic manipulator arm, all under real-time constraints.
The full pipeline integrates three subsystems:
- Perception: a computer vision module detects cube positions and colors from an RGB-D camera feed, estimating 3D poses for the planner.
- Motion Planning: a path planner generates collision-free joint trajectories to approach each cube, using forward/inverse kinematics for precise end-effector positioning.
- Control: a ROS-based controller executes the planned trajectories, handling grasping and release actions via a gripper interface.
Our team's implementation was robust enough to complete the stacking task efficiently under competition
time constraints, handling variations in cube placement.