Modelling of multi-robot system for search and rescue

• Main Author: Poy, Yi Ler
• Format: Final Year Project / Dissertation / Thesis
• Published: 2023
• Subjects: TJ Mechanical engineering and machinery
• Online Access: http://eprints.utar.edu.my/5818/1/MH_1904296_Final_POY_YI_LER.pdf; http://eprints.utar.edu.my/5818/

The field of robotics has seen an increased interest in multi-robot systems, which bring a new set of challenges to the table. One of the key aspects in multi-robot systems is the path planning problem, which involves finding collision-free paths for each robot to reach their respective destinations while optimizing various performance metrics. This report focusses on developing a novel multi-robot path planning algorithm based on the Modified Particles Swarm Optimization (MPSO) algorithm for dynamic environments. The MPSO algorithm introduces a new path planning scheme for determining robot’s waypoints. Unlike the normal PSO algorithm which initializes the particle swarm at the robot’s starting position and iteratively determining each waypoint until a completed path is generated, MPSO algorithm initializes the particle swarm within a predefined search space and searches for the global best position within it to determine a specific robot waypoint through iteration updates. Moreover, to cope with dynamic environments, a combination of global and local path planning methods is introduced. The PSO algorithm functions as a global path planner, determining the complete path for each robot, whereas a sensor-based obstacle avoidance algorithm serves as a local planner to avoid collision with dynamic obstacles during navigation. In this project, this sensor-based algorithm is known as the Obstacle Avoidance Algorithm. The simulations conducted using MATLAB demonstrate the superiority of the MPSO algorithm over the PSO algorithm in terms of average path length and execution time of all robots in all three proposed scenarios: 16 meter shorter and 7.1 seconds faster in the first scenario, 17.89 meters shorter and 6.14 seconds faster in scenario 2, and 6.18 meters shorter and 8.47 seconds faster in scenario 3. The impact of the MPSO parameters on the simulation results is also studied to determine the best PSO parameters that achieve the best performance. It was found that the number of populations set to 75 and dynamically adjusts the value of inertial weight, the cognitive and social parameter provides the best performance in terms of shortest path length and execution time. In conclusion, this project shows that the MPSO algorithm is capable of generating a better path compared to the normal PSO algorithm in terms of average path length and execution time, making it a promising algorithm for multi-robot path planning in dynamic environments.