Our PhD. student Miloš Prágr presented his work on "Online Incremental Learning of the Terrain Traversal Cost in Autonomous Exploration" (co-authored by our other students Petr Čížek and Jan Bayer, all supervised by Jan Faigl) at the prestigious Robotics: Science and Systems (RSS) 2019 conference in Freiburg, Germany. The presented work is a contribution to our development of a self-improving robotic system that can operate in an unknown environment and exploit its experience for improving its performance, the research that is conducted in the Computational Robotics Lab.
In this paper, we address motion efficiency in autonomous robot exploration with multi-legged walking robots that can traverse rough terrains at the cost of lower efficiency and greater body vibration. We propose a robotic system for online and incremental learning of the terrain traversal cost that is immediately utilized to reason about next navigational goals in building spatial model of the robot surrounding. The traversal cost experienced by the robot is characterized by incrementally constructed Gaussian Processes using Bayesian Committee Machine. During the exploration, the robot builds the spatial terrain model, marks untraversable areas, and leverages the Gaussian Process predictive variance to decide whether to improve the spatial model or decrease the uncertainty of the terrain traversal cost. The feasibility of the proposed approach has been experimentally verified in a fully autonomous deployment with a hexapod walking robot.
The conference Robotics: Science and Systems has a long history of bringing together researchers in all areas of robotics from around the world for an engaging and focused week. The program includes invited talks as well as oral and poster presentations of accepted papers. The three-day main session is preceded by two days of workshops and tutorials. This year it took place in Freiburg, Germany. More info can be found at the conference website.