NeBula: Quest for Robotic Autonomy in Challenging Environments; TEAM CoSTAR at the DARPA Subterranean Challenge

This paper presents and discusses algorithms, hardware, and software developed by the TEAM CoSTAR (Collaborative SubTerranean Robots), competing in the DARPA Subterranean Challenge., it presents the techniques utilized within the Tunnel (2019) and (2020) competitions, where CoSTAR achieved 2nd and 1st place,. We also discuss CoSTAR's demonstrations in Martian-analog surface subsurface (lava tubes) exploration. The paper introduces our autonomy, referred to as NeBula (Networked Belief-aware Perceptual Autonomy). is an uncertainty-aware framework that aims at enabling resilient and autonomy solutions by performing reasoning and decision making in the space (space of probability distributions over the robot and world). We discuss various components of the NeBula framework, including: (i) and semantic environment mapping; (ii) a multi-modal positioning; (iii) traversability analysis and local planning; (iv) global motion and exploration behavior; (i) risk-aware mission planning; (vi) and decentralized reasoning; and (vii) learning-enabled adaptation. discuss the performance of NeBula on several robot types (e.g. wheeled,, flying), in various environments. We discuss the specific results and learned from fielding this solution in the challenging courses of the Subterranean Challenge competition.