Hybrid Hierarchical Swarm Robots for Planetary Exploration

Autonomous multi-robot systems offer potential for planetary exploration by distributing tasks and reducing reliance on a singular platform. This work presents an implementation of a hybrid hierarchical swarm architecture including a central "queen'' robot with extended sensing and computational capability and multiple "worker'' robots with limited onboard resources. The queen robot performs global planning and assigns navigation tasks based on a high-level map representation, while worker robots execute local path following with minimal but independent sensing capabilities. A feasibility demonstration in the SpaceROS lunar simulation environment shows that task assignment and path execution can occur with reduced communication requirements, as worker robots continue to operate locally during communication interruptions. Experimental results demonstrate that worker rovers relying solely on wheel odometry exhibit waypoint tracking errors of 3-22 m depending on path geometry, while workers with access to global-frame pose information (simulating ideal queen-provided corrections) achieve consistent sub-meter accuracy (mean error 0.23-0.26 m). We achieved at least 85% reduction in navigation error against 3 different scenarios in multiple runs. These results describe the potential and outline the limitations of a hybrid hierarchical model for future planetary surface missions.