Optimised multi-robot path planning via smooth trajectory generation

• Main Author: Loke, Zhi Yu
• Format: Final Year Project / Dissertation / Thesis
• Published: 2024
• Subjects: QA75 Electronic computers. Computer science; T Technology (General)
• Online Access: http://eprints.utar.edu.my/6549/1/MH_1903026_Final_LOKE_ZHI_YU.pdf; http://eprints.utar.edu.my/6549/

The deployment of multi-robot system (MRS) in real-world applications like warehouses and manufacturing plants has increased the importance of path planning algorithms for MRS. Compared to a single robot, an MRS is more effective and robust in completing tasks, even when one robot breaks down. Particle swarm optimization (PSO) outperforms conventional methods like artificial potential fields (APF), the Dijkstra algorithm, and the A* algorithm in path planning for mobile robots. PSO focuses on finding the local and global best position of each particle through iterations, calculated based on a fitness function whereby the Euclidean distance between a particle's next waypoint and the target point is calculated. However, there is a need for optimizing smooth trajectory generation in multi-robot path planning. The application of parametric curves like the Bezier curve, Dubin's curve, and non�uniform rational B-spline (NURBS) curve is common for generating smooth trajectories. This project uses the Bezier curve equation for smooth trajectory generation as it is computationally inexpensive and easy to form desired curves. Smooth trajectories enable efficient traversal, shorter travel times, and energy conservation by limiting unnecessary movements and abrupt changes in direction. Collision avoidance is achievable through careful coordination of robot trajectories, preventing collisions and improving MRS safety. This project develops an enhanced PSO algorithm (EPSO) for smooth trajectory generation of MRS, aiming to reduce path length, execution time, and turn points, thereby increasing efficiency and conserving energy. A MPSO algorithm, without path smoothening, is used for comparison. EPSO parameters like swarm size, control points, inertia weight, and acceleration coefficients are tuned appropriately. Simulations for MPSO and EPSO are conducted five times for average results. In conclusion, EPSO outperforms MPSO in generating pathways with shorter path length, lower execution time, and fewer turn points, making it an effective solution for optimizing multi�robot path planning.