Towards Autonomy and Mobility for a Tethered Robot Exploring Extremely Steep Terrain

摘要

Mobile robots are well suited to explore environments considered too costly, time consuming, and hazardous for human inspection. However, a recent push to explore increasingly extreme environments has required the development of robust mobile platforms that can navigate steep, cluttered terrain, operate for extended periods, and relay information to a remote operator. Applications of these systems include terrestrial and planetary geologic survey and infrastructure inspection, where remote observation is not a viable option. This thesis chronicles the design, development, and testing of the physical platform and autonomy functions for the Tethered Robotic eXplorer (TReX), a novel mapping robot capable of navigating near-vertical terrain while supported by an attached electromechanical tether; the tether provides continuous power and communication to and from the robot, but also constrains motion due to its finite length and susceptibility to entanglement in cluttered environments. A tethered robot can avoid entanglement by (i) mapping intermediate anchors (locations of obstacle-to-tether contacts), and (ii) autonomously retracing its outgoing path to sequentially unwrap the tether from obstacles. We approach (i) by formulating incremental and batch solutions to a new TSLAM problem using tether measurements to aid odometry and map anchors, and handle (ii) using visual route following in conjunction with autonomous tether control to manage tension and assist the robot to repeat previously driven paths on steep terrain. This work concludes with a geologic surveying mission, where TReX is teleoperated to explore and map a steep, tree-covered rock outcrop in an outdoor mine.