Collaborative Autonomous 3D Reconstruction for Heterogeneous Multiple UGVs in Complex Environments

Unmanned Ground Vehicles (UGVs), with the ad- vantages of endurance, payload capacity, and high autonomy, are widely utilized in search and rescue, surveillance, and exploration. The 3D reconstruction of unknown environments by multiple UGVs is a fundamental task in multi-robot systems. However, diverse terrains and complex surface structures of en- vironmental objects pose significant challenges to efficient multi- robot collaborative reconstruction. To address these challenges, it is essential to adopt a heterogeneous multi-robot system equipped with various sensor types and chassis configurations to traverse and cover the environment efficiently. This work focuses on map reconstruction and collaborative planning for heterogeneous multi-robot systems to enable real-time exploration of complex environments. We propose an incomplete surface element (ISE) extraction method based on terrain awareness and observation quality to support collaborative map fusion and view planning. A view generation technique employing minimal view set sampling is introduced to reduce the number of task views while adapting to the motion and sensory constraints of various robots. Addition- ally, a hierarchical view planning framework is designed with a clustering-based matching algorithm to achieve near-optimal task allocation, effectively coordinating the view tasks of robots with different capabilities. The feasibility of the proposed method is validated through a simulation scenario, demonstrating efficient environment exploration and high-completeness surface coverage.