Development of a multi-agent task allocation and hot-zone detection system for mars exploration using data mining techniques

The unique challenges of autonomous exploration and sampling on the surface of Mars and other planetary bodies as part of a multi-agent system in uncertain and data-scarce conditions are non-trivial. The problem statement of this paper is a Mars exploration for multi-agent systems with hot zone detection, dynamic task allocation, and shared mapping capabilities. This paper investigates the use of cooperative multi-agent systems for Mars exploration with dynamic task allocation and hotspot location for maximum scientific return of Mars exploration missions. Using a data mining approach in pseudo-structures, the methodology applies clustering to locate regions of high-value samples, uses rules to formulate agent support and re-tasking, and data fusion to create a shared terrain knowledge system. The system’s performance was tested with a synthetic dataset with over 1000 simulated exploration events. Sample collection summary: total success reports – 483, total failure reports – 517, total samples collected – 1944; that is, the success rate is approximately 48.3%, while the failure rate is approximately 51.7% of a total of 1,000 reports. The experiments show the framework can improve the efficient collection of samples and demonstrate the promise of data-mining coordination methods for scalable autonomous planetary exploration. The multi-agent team worked in an efficient and coordinated manner, collaborating closely to identify the sample locations and collect them from the hot-zone area. As demonstrated in the results, the agents operated together seamlessly, ensuring accurate detection and successful retrieval of all required samples