ArmLab: A Project on Robotic Manipulation and Computer Vision
This project was part of ROB550: Robotic Systems Lab for my MS in Robotics at the University of Michigan. I was one of the members of a team of three, and the objective of the project was to implement algorithms in order to perform simple detection and manipulation tasks for a 5-DOF robotic arm.
Relevant Skills and Topics:
Robotic Manipulation
Python
Computer Vision
Motion Planning/State Machine design
Algorithms implemented:
Camera calibration methods: Estimate the camera extrinsic parameters in order to convert pixel coordinates to world coordinates
Forward/Inverse Kinematics: Determine the position of the gripper (end effector) given the joint angle read-outs from the arms joint motor encoders (forward kinematics), and vice versa (inverse kinematics)
Object Detection Algorithms: Detect blocks with controlled size and colors within the cameras field of view
Path Planning/Autonomy: Performs tasks such as sorting or stacking blocks based on their size and color.
Arm working in action to drop off a block onto a stack.