Learning robust perceptive locomotion for quadrupedal robots in the wild

Abstract

Legged robots that can operate autonomously in remote and hazardous environments will greatly increase opportunities for exploration into under-explored areas. Exteroceptive perception is crucial for fast and energy-efficient locomotion: perceiving the terrain before making contact with it enables planning and adaptation of the gait ahead of time to maintain speed and stability. However, utilizing exteroceptive perception robustly for locomotion has remained a grand challenge in robotics. Snow, vegetation, and water visually appear as obstacles on which the robot cannot step~-- or are missing altogether due to high reflectance. Additionally, depth perception can degrade due to difficult lighting, dust, fog, reflective or transparent surfaces, sensor occlusion, and more. For this reason, the most robust and general solutions to legged locomotion to date rely solely on proprioception. This severely limits locomotion speed, because the robot has to physically feel out the terrain before adapting its gait accordingly. Here we present a robust and general solution to integrating exteroceptive and proprioceptive perception for legged locomotion. We leverage an attention-based recurrent encoder that integrates proprioceptive and exteroceptive input. The encoder is trained end-to-end and learns to seamlessly combine the different perception modalities without resorting to heuristics. The result is a legged locomotion controller with high robustness and speed. The controller was tested in a variety of challenging natural and urban environments over multiple seasons and completed an hour-long hike in the Alps in the time recommended for human hikers.

References in corpus (2)

• Learning agile and dynamic motor skills for legged robots
• Multi-expert learning of adaptive legged locomotion

Cited by in corpus (13)

• Reaching the Limit in Autonomous Racing: Optimal Control versus Reinforcement Learning
• Learning Robust Autonomous Navigation and Locomotion for Wheeled-Legged Robots
• Scientific Exploration of Challenging Planetary Analog Environments with a Team of Legged Robots
• Versatile Multi-Contact Planning and Control for Legged Loco-Manipulation
• Controlling the Solo12 Quadruped Robot with Deep Reinforcement Learning
• Learning-based Design and Control for Quadrupedal Robots with Parallel-Elastic Actuators
• Meta Reinforcement Learning for Optimal Design of Legged Robots
• Benchmarking Potential Based Rewards for Learning Humanoid Locomotion
• DeepCPG Policies for Robot Locomotion
• Hierarchical Quality-Diversity for Online Damage Recovery
• Informed Guided Rapidly-Exploring Random Trees*-Connect for Path Planning of Walking Robots
• Continuous Jumping of a Parallel Wire-Driven Monopedal Robot RAMIEL Using Reinforcement Learning
• Torque-Minimizing Control Allocation for Overactuated Quadrupedal Locomotion