Autonomous Visionbased Landing and Terrain （自治Visionbased着陆和地形）.pdf

Autonomous Vision-based Landing and Terrain Mapping Using an MPC-controlled Unmanned Rotorcraft Todd Templeton, David Hyunchul Shim, Christopher Geyer, and S. Shankar Sastry∗ Abstract In this paper, we present a vision-based terrain mapping and analysis system, and a model predictive control (MPC)- based flight control system, for autonomous landing of a helicopter-based unmanned aerial vehicle (UAV) in un- known terrain. The vision system is centered around Geyer et al.’s Recursive Multi-Frame Planar Parallax algorithm [1], which accurately estimates 3D structure using geo- referenced images from a single camera, as well as a modu- lar and efficient mapping and terrain analysis module. The vision system determines the best trajectory to cover large Figure 1: An electrically-powered, MPC-controlled UAV areas of terrain or to perform closer inspection of potential testbed for vision-based landing and terrain mapping. landing sites, and the flight control system guides the vehi- cle through the requested flight pattern by tracking the refer- ence trajectory as computed by a real-time MPC-based op- To execute a safe landing, a UAV first needs to find a timization. This trajectory layer, which uses a constrained level space that is free of debris. It is usually impossible system model, provides an abstraction between the vision or impractical to rely on a preloaded digital elevation map system and the vehicle. Both vision and flight control re- (DEM) due to its limited resolution and possible inaccuracy sults are given from flight tests with an electric UAV. due to obstacle movement. UAVs are typically no larger than manned helicopters, an