Autonomous Spot: Long-Range Autonomous Exploration of Extreme Environments with Legged Locomotion

Abstract

This paper serves as one of the first efforts to enable large-scale and long-duration autonomy using the Boston Dynamics Spot robot. Motivated by exploring extreme environments, particularly those involved in the DARPA Subterranean Challenge, this paper pushes the boundaries of the state-ofpractice in enabling legged robotic systems to accomplish realworld complex missions in relevant scenarios. In particular, we discuss the behaviors and capabilities which emerge from the integration of the autonomy architecture NeBula (Networked Belief-aware Perceptual Autonomy) with next-generation mobility systems. We will discuss the hardware and software challenges, and solutions in mobility, perception, autonomy, and very briefly, wireless networking, as well as lessons learned and future directions. We demonstrate the performance of the proposed solutions on physical systems in real-world scenarios. 3 The proposed solution contributed to winning 1st-place in the 2020 DARPA Subterranean Challenge, Urban Circuit. 4

Risk-Sensitive Motion Planning using Entropic Value-at-Risk

“…Measurement noise and moving obstacles were considered in [15], wherein the authors devised an MPC-based scheme for path planning with CVaR safety Spot and Husky robots exploring a subterranean environment in Valentine Cave, Lava Beds National Monument, California. Obstacle avoidance in unstructured environments incurs higher mission risk due to lack of global positioning [9]. constraints when a reference trajectory is generated by RRT * [18], and extended to a Wasserstein distributionally robust formulation in [16].…”

“…1.Spot and Husky robots exploring a subterranean environment in Valentine Cave, Lava Beds National Monument, California. Obstacle avoidance in unstructured environments incurs higher mission risk due to lack of global positioning[9].…”

STEP: Stochastic Traversability Evaluation and Planning for Risk-Aware Off-road Navigation

“…We tested our risk-aware traversability software during our fully autonomous runs. The planner was able to navigate the robot safely to the every goal provided by the upper-layer coverage planner [8,23] despite the challenges posed by the environment. Figure 10 shows snapshots of elevation maps, CVaR risk maps, and planned paths.…”

LOCUS: A Multi-Sensor Lidar-Centric Solution for High-Precision Odometry and 3D Mapping in Real-Time

“…The first approach relies on prior knowledge of the environment that can be acquired by a human supervisor, for instance in single floor exploration of urban environments. For stair-climbing robots, the initiation of a stair mission can be used to deactivate FGA, and then reactivate it when the stair mission is complete through the input of a human operator (see [26] for an example). In the second approach, an IMU monitor can be used to detect periods when the robot has near-zero roll and pitch over a sufficient time period to activate FGA, and then deactivate it when this condition is no longer met.…”