Dynamic obstacle handling in multi-robot coverage

Tay, Wing Le (2024) Dynamic obstacle handling in multi-robot coverage. Final Year Project, UTAR.

Preview

PDF Download (4Mb) | Preview

Abstract

The field of robotics has witnessed a growing interest in multi-robot system for coverage task such as static coverage, which presenting a new set of challenges to table. One of the key aspects in multi-robot coverage is dynamic obstacle handling, which involves robots avoiding dynamic obstacle during coverage task while reaching their respective goal positions. This report focuses on developing a multi-robot coverage algorithm with dynamic obstacle handling abilities, known as the Modified Lloyd’s algorithm with Velocity Obstacle (VO). This modified algorithm introduces the Velocity Obstacle which calculates avoidance velocities for robots to avoid dynamic obstacle. Unlike the existing Lloyd’s algorithm which primarily focuses on achieving complete coverage, the Modified Lloyd’s algorithm with VO can determine the time to collision based on the relative distances and velocities of robots and dynamic obstacle. It then generates a new avoidance velocity for robots, enabling them to avoid dynamic obstacles while achieving complete coverage. The simulations were conducted by using MATLAB to demonstrate the superiority of Modified Lloyd’s algorithm with VO over existing Lloyd’s algorithm in terms of average of total number of collisions between robots and dynamic obstacle during coverage task. All five robots successfully avoided dynamic obstacle during coverage tasks, achieving complete coverage with zero collisions. The Modified Lloyd’s algorithm was also tested under two different scenarios to evaluate its functionality. In first scenario, as the aggressiveness of dynamic obstacle increased, the algorithm remained capable of handling them, although there were occasional collisions at very high aggression levels. In the second scenario, where the starting position of dynamic obstacle varied, the modified algorithm consistently handled dynamic obstacle and achieved zero collisions during coverage tasks. The impacts of the modified algorithm’s parameters on simulation results were also studied to determine the optimal parameters for achieving better performance. It was found that with number of 500 iterations and a safety margin of 5, the algorithm provided better performance in terms of shorter execution time and lower average number of collisions.

Actions (login required)