Tunnel Vision: High-Speed Mapping for Enclosed Terrain

Autonomous systems can augment human operators in exploring indoor environments. In some cases, such as search and rescue scenarios, it is essential to create a map as quickly as possible, and some loss of fidelity may be acceptable if the mapping can be expedited. This work proposes using a high-speed aerial vehicle with low fidelity scanning to map long indoor corridors, such as train or highway tunnels at a rate of ~ 50 m/s with a tunnel of ~1000+ m length. The system operates on lightweight hardware with a novel mapping algorithm. The algorithm utilizes data collected with commercial off the shelf Flash LiDAR and mmWave Radar sensors to develop a two-dimensional map of a corridor for human operators. The system operates at a rate much faster than existing autonomous systems. A prototype of the proposed system was built and demonstrated to potential operators inside a 70 m tunnel. It was able to map the tunnel at a speed of approximately 5 m/s and a spatial resolution of 10 cm. The demonstration provides a proof-of-concept for high-speed mapping and exploration of indoor environments.