Intelligent trajectory planning using grey wolf and jellyfish algorithms for multi-robot-multi-tasking system

The increasing demand for automation in sectors such as advanced manufacturing, warehousing, vertical farming, and autonomous inspection has propelled the development of multi-robot systems capable of performing a broad range of tasks with constrained resources and under time-sensitive conditions. However, traditional trajectory planning approaches, including graph-based methods and sampling-based algorithms, face significant scalability and optimisation challenges when confronted with large-scale, heterogeneous, and collaborative multi-robot scenarios. This research addresses these challenges by introducing novel intelligent trajectory planning algorithms, namely, the Elitist Preservation Differential Evolution Grey Wolf Optimiser (EPDE-GWO) and the Tent Chaos Lévy Flight Dynamic Weight Jellyfish Search (TLDW-JS), which aim to enhance global search capacities, convergence speed, and stability in complex multi-robot-multi-tasking (MRMT) and multi-robot- collaborative-multi-tasking (MRCMT) environments. In the first part of this research, a mathematical model was formulated to determine the essential constraints and optimisation objectives for multi-robot trajectory planning. The EPDE-GWO algorithm was then devised to solve the trajectory planning problem for multi-robot-multi-tasking scenarios, in which each task is completed by a single robot. By integrating Differential Evolution strategies and Elitist Preservation into the baseline Grey Wolf Optimiser, the EPDE-GWO demonstrated superior performance compared to standard approaches such as the Genetic Algorithm (GA), Simulated Annealing (SA), Particle Swarm Optimization (PSO), and Grey Wolf Optimizer (GWO). Simulation results showed that the EPDE-GWO achieved the shortest average path length (815.1 metres), reducing the average path length by approximately 24.6%. Additional improvements were observed in convergence speed, solution stability, consistency, and scalability. Building upon these advancements, the second part of the research focused on multi- robot-collaborative-multi-tasking scenarios, in which multiple robots with complementary functions jointly execute complex tasks under route synchronisation requirements. To address the heightened complexity of these applications, the TLDW- JS algorithm was proposed, incorporating Tent Chaos for enhanced initialisation, Lévy Flight for global exploration, and a Dynamic Weight adjustment scheme to balance exploration and exploitation. Comparative simulations demonstrated that, compared to the baseline algorithms GA, Dung Beetle Optimization (DBO), PSO, and Jellyfish Search (JS), the TLDW-JS achieved the shortest average path length (707.2 metres), reducing the average path length by approximately 34.3%. It also exhibited faster convergence speed, greater stability, and better consistency, effectively overcoming the JS algorithm's dependency on high-quality initial populations and its limitations in avoiding premature convergence. Overall, this research provides evidence that making carefully tailored improvements to bio-inspired algorithms can significantly bolster the efficiency, reliability, and adaptability of multi-robot trajectory planning. The findings underscore the practical feasibility of using the EPDE-GWO and TLDW-JS in real-world robotic applications, laying a robust foundation for future research on advanced cooperative strategies, heterogeneous robot coordination, and optimisation frameworks that would further enhance multi-robot system performance.

Text FK 2024 77 - Full Text.pdf Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (6MB) Text FK 2024 77.pdf Restricted to Repository staff only Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (7MB)

Actions (login required)

View Item