Cooperative and Autonomous Mapping for Heterogeneous NAVs

Nano unmanned aerial vehicles (NAVs) have been increasingly used for mapping due to their advantages in weight and size, such as search, rescue and reconnaissance in confined areas. However, because the resources carried by NAVs are extremely limited, it is a challenge to achieve efficient autonomous mapping in 3D space. To address this challenge, this paper proposes a new autonomous mapping model and system architecture with heterogeneous NAVs. Firstly, this paper introduces an enhanced OctoMap-based map model designed for resource-constrained NAVs. In comparison to the conventional OctoMap map model, the proposed model exhibits a significant reduction of 85.94% in memory usage while maintaining the same number of voxel blocks. In addition, We propose an active exploration algorithm based on Next Best View (NBV) theory, which can achieve agile and efficient autonomous mapping. Furthermore, a merging rays strategy and a dynamic adjustment of Edge-NAV positions strategy are proposed to improve the communication quality between heterogeneous NAVs. The map model and exploration algorithms are deployed on real-world NAVs, and the correctness and effectiveness of the proposed algorithm are validated through extensive simulations and real-world experiments, accomplishing more than the exploration completion rate of 80% with less than memory pool usage of 60%.