Implementation of collaborative/cooperative tasks using a dual arm setup in ROS/MoveIt with two OpenManipulator-X robotic arms and D435i RealSense cameras. Development in C++, Python, Linux and ROS/MoveIt for motion planning and perception.
This project was an assignment in one of my robotics modules.
I completed this project during my Internship at Micochip Technology.
This was my second year project, the task was to use a Raspberry Pi, a Neural Comute Stick 2, a webcam and computer vision to detect and recognise recyclable objects. While the project was successful at recognition, the project used YOLOv3 and COCO dataset which wasn't great with deformed objects. A model which was solely trained on trash and recyclable objects would have worked a lot better, but due to time (and knowledge) this was not possible at the time. However this project demonstrated the use of a hardware accelerator to boost the video capabilities of the Raspberry Pi to perform computer vision
This was my first year project and received an A. A working physical prototype was contructed to balanace a ball on a beam. This demostrated the control theory such as PID controllers in a real application. This was a "group" project, but I ended up doing mostly everything.
This is something I worked on over a summer. I applied my skills in embedded systems and electronics to create it.
I bought this small guy to learn more about humanoid/bipedal robots.
Using an AMG8833 sensor by panasonic, and a Raspberry Pi to make a thermal camera