Time: Mar 2021~ Jun 2022
Role: Developing algorithms for camera and intergrate devices under Robotic Operating System (ROS)
Skills Gained: Camera Calibration, Transfrom Matrix & Frame coordinatation, ROS, Motion planning
Developed algorithm for the husky car reaching desingated place without hitting the obstacles using lidar. And also allowed the robotic arm to detect and grab the tennis autonomously by using camera. After the tennis is grabbed, the husky car will go to arrive
By calibrating the camera using chessboard as well as the distance between the camera and robotic arm, we can then define a rotation matrix that helps us transform object's coordinatation into robotic arm's coordinatation. Since the camera only provid 2d coordination, we get the depth through the size of tennis we detect. The smaller we got, the larger distance from the camera. Afterwards, we applied inverse kinematic into our algorithm for arm so that it can reached to the correct coordination. The error is within 3cm so that it can grab it successfully.
For the Husky Car, we used a lidar as our sensor to detect the surroundings, and we applied Kalman-filer to balance between our model and the sensor data. We used ROS to integrate and communicate with others