Dynamic Role Allocation Algorithms for Heterogeneous Robot Swarms in Disaster Recovery

Abstract

Disaster recovery activities demand quick adaptation and coordination due to the complex and unpredictable nature of the environment. Robotic swarm solutions that feature heterogeneous capabilities in terms of mobility, sensory ability, and payload, among others, are ideal for improving the performance and effectiveness of such an endeavor. However, it is imperative for such robotic swarms to have effective algorithms that help with dynamic role allocation in accordance with the capability of each agent and the prevailing circumstances. This paper provides a novel role allocation algorithm that includes considerations of heterogeneity, self-evolving agentic logic, and dynamic reallocation. The algorithm will be assessed using suitable simulation tests and performance measures. It will compute a suitability score for each assignment and use a utility function to allocate roles dynamically to ensure optimized task coverage, reduce redundancy, and minimize time wastage during the process. Simulations will incorporate disaster situations that require activities such as searching, cleaning, and victim assistance, among others. Performance measures TCR, SCE, ATAT, RU, and RR show that the new algorithm outperforms existing ones such as static allocation, centralized optimization, and reinforcement learning. The results show that the algorithm is able to achieve an average success of 92% in completing tasks, an efficiency of 88%, and a minimal task allocation time of 0.35 seconds. These results demonstrate the versatility, scalability, and computational efficiency of the algorithm. They confirm its capability of taking advantage of the diversity of the resources and ensuring reliable performance under different conditions, making it possible to be applied in practical scenarios of disaster recovery missions.