DECENTRALIZED CONTROL APPROACH FOR MULTI-AGENT PATH FINDING IN OMNI WHEELED ROBOTS

• Main Author: Ghani, Abdul
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/79340
• Institution: Institut Teknologi Bandung

Multi-Agent Path Finding (MAPF) is a method of finding the best path without collision. In the final project, MAPF is implemented on an omni-wheeled Rover with the aim of moving laboratory equipment from one place to another. The application of MAPF is carried out with a decentralized control approach which means that each Rover performs path planning and collision avoidance to move laboratory equipment. This research begins with designing a path planning algorithm. In path planning, the D* Lite algorithm is used which can perform path planning well and can also re-plan the path quickly if there is a change in the terrain information received. In the next stage, communication is designed using WebSocket as a communication medium between rovers that provides task, position, and speed data for each rover. These data can be used as a reference for collision avoidance design. In collision avoidance, velocity obstacles are used to avoid collisions on dynamic obstacles that will be passed. The research ends with the implementation of the design made on the Rover. From the experiments conducted, Rover can do path planning well and in its implementation Rover can follow the path that has been designed using the D* Lite algorithm. In the velocity obstacle algorithm by testing on the available ground mat, Rover can solve 4 out of 5 classic MAPF problems namely swapping conflict, following conflict, vertex conflict, and cycle conflict.