Rough Terrain Navigation for Legged Robots using Reachability Planning and Template Learning

Abstract

Navigation planning for legged robots has distinct challenges compared to wheeled and tracked systems due to the ability to lift legs off the ground and step over obstacles. While most navigation planners assume a fixed traversability value for a single terrain patch, we overcome this limitation by proposing a reachability-based navigation planner for legged robots. We approximate the robot morphology by a set of reachability and body volumes, assuming that the reachability volumes need to always be in contact with the environment, while the body should be contact-free. We train a convolutional neural network to predict foothold scores which are used to restrict geometries which are considered suitable to step on. Using this representation, we propose a navigation planner based on probabilistic roadmaps. Through validation of only low-cost graph edges during graph expansion and an adaptive sampling scheme based on roadmap node density, we achieve real-time performance with fast update rates even in cluttered and narrow environments. We thoroughly validate the proposed navigation planner in simulation and demonstrate its performance in real-world experiments on the quadruped ANYmal.

Elevation Mapping for Locomotion and Navigation using GPU

“…An important benefit of performing elevation mapping on GPU is that data is already loaded into memory and we can therefore perform efficient terrain analysis using neural networks without any processing overhead caused by data transfer between CPU and GPU. Specifically, we deploy a simple Convolutional Neural Network (CNN) model trained to output traversability values for robot navigation [45]. It is implemented in PyTorch [36] and is light enough to run at full map update rate.…”

“…All authors are with Robotic Systems Lab, ETH Zurich the sum of geometric features [45]. Besides mobile robot navigation, legged robots also benefit from perceiving their surroundings to control their locomotion.…”

“…We introduced a raycasting method for visibility clean up and an exclusion area to remove virtual artifacts when the robot gets close to them. Furthermore, we integrate a learningbased traversability filter [45], which gives a traversability value based on the local geometry and enhanced the elevation map through improvements such as upper bound calculation or overlap clearance for navigation tasks. For locomotion, we integrated various smoothing filters used in [23], [22] and plane segmentation used in [16].…”

“…We demonstrate our framework through extensive experiments. Our implementation supported legged locomotion and navigation research and formed the basis for perceiving surrounding terrains used by learning-based controllers [30], [38], [29], model-based controllers [23], [22], [16], or others such as navigation [45], [47] or learning occlusion filling [40]. In addition, the proposed elevation mapping solution was successfully used for underground exploration missions during Defense Advanced Research Projects Agency (DARPA) Subterranean Challenge [9] where team CERBERUS [8], [42] deployed our mapping on four quadrupedal robots, ANYmal [21].…”

Coupling Vision and Proprioception for Navigation of Legged Robots

“…[12] describe a coupled navigation and locomotion framework which reasons about foot placement by estimating foothold costs from an elevation map. Foothold scores can be estimated heuristically [14,18,32,40,52,80] or learnt [34,44,50,51,79]. Other methods forgo explicit foothold optimization and learn whether a given section of terrain is traversible [11,24,86].…”