Dynamic Pickup and Delivery Task Allocation in Multi-Agent Robotic Systems: Vacancy Chain Approaches Evaluation
Shahd E. Askar, Sameh Eid, Yasser I. Elshaaer · 2024
Multi-Agent Robotic Systems (MARS) face the challenge of efficiently dynamic allocating pickup and delivery tasks. Although these tasks are complex and interconnected, many existing research efforts prioritize improving task allocation efficiency while neglecting the critical aspects of overall completion time and load balance in dynamic task allocation scenarios. This gap often results in suboptimal performance, with increased operational delays and inefficiencies. To address this problem, this paper proposes a modified vacancy chain approach with two different strategies: greedy and probability based. These strategies are designed to dynamically optimize task allocation, reducing completion time and balancing the workload among robots. The methods are validated through extensive simulations in the Robot Operating System (ROS) and Gazebo environments. Results demonstrate that while the greedy strategy is highly effective in speed within static settings, the probability-based strategy offers better adaptability and workload distribution in dynamic conditions. These findings could contribute to enhancing the efficiency and responsiveness of multi-robot systems in real-world applications.