This project demonstrates that multiple robots can collaborate to explore an unknown environment without GPS to generate a map with reasonable accuracy. This addresses exploration as a problem in the planning layer of the robotics domain. A frontier-based exploration methodology will be explained. The main approach uses Rapidly-expanding Random Trees for frontier detection, and map merging for information sharing. The developed algorithm also makes use of existing solutions for SLAM, and path planning. The algorithm runs in a network of robots, and a master machine on ROS. The project shows that using multiple robots can reduce exploration time compared to using a single robot.
Know basic robotics engineering, preferably robotics software and control.
Sean Chok
I am a recent mechanical engineering graduate from HKU, with a minor in computer science with a passion in cloud services and robotics. I am also a co-founder of Deploifai. We're building MLOps tools for ML developers so that they don't need to worry about cloud infrastructure management. Our platform provides ready-to-go GPU servers for model training and data science, and application containerization with model inference endpoints, and other ML-related cloud services.