Multi-expert learning of adaptive legged locomotion

Abstract

Achieving versatile robot locomotion requires motor skills which can adapt to previously unseen situations. We propose a Multi-Expert Learning Architecture (MELA) that learns to generate adaptive skills from a group of representative expert skills. During training, MELA is first initialised by a distinct set of pre-trained experts, each in a separate deep neural network (DNN). Then by learning the combination of these DNNs using a Gating Neural Network (GNN), MELA can acquire more specialised experts and transitional skills across various locomotion modes. During runtime, MELA constantly blends multiple DNNs and dynamically synthesises a new DNN to produce adaptive behaviours in response to changing situations. This approach leverages the advantages of trained expert skills and the fast online synthesis of adaptive policies to generate responsive motor skills during the changing tasks. Using a unified MELA framework, we demonstrated successful multi-skill locomotion on a real quadruped robot that performed coherent trotting, steering, and fall recovery autonomously, and showed the merit of multi-expert learning generating behaviours which can adapt to unseen scenarios.

References in corpus (8)

• Learning agile and dynamic motor skills for legged robots
• Robots that can adapt like animals
• DeepMimic: Example-Guided Deep Reinforcement Learning of Physics-Based Character Skills
• Whole-Body Nonlinear Model Predictive Control Through Contacts for Quadrupeds
• Learning Locomotion Skills Using DeepRL: Does the Choice of Action Space Matter?
• Learning natural locomotion behaviors for humanoid robots using human knowledge
• Contact-Implicit Trajectory Optimization using an Analytically Solvable Contact Model for Locomotion on Variable Ground
• Learning Whole-body Motor Skills for Humanoids

Cited by in corpus (16)

• Learning robust perceptive locomotion for quadrupedal robots in the wild
• Concurrent Training of a Control Policy and a State Estimator for Dynamic and Robust Legged Locomotion
• Learning Free Gait Transition for Quadruped Robots via Phase-Guided Controller
• Lifelike Agility and Play in Quadrupedal Robots using Reinforcement Learning and Generative Pre-trained Models
• Adaptive 3D descattering with a dynamic synthesis network
• ViTAL: Vision-Based Terrain-Aware Locomotion for Legged Robots
• Learning Bipedal Walking for Humanoids with Current Feedback
• Imitating and Finetuning Model Predictive Control for Robust and Symmetric Quadrupedal Locomotion
• Terrain-perception-free Quadrupedal Spinning Locomotion on Versatile Terrains: Modeling, Analysis, and Experimental Validation
• Learning to Navigate in a VUCA Environment: Hierarchical Multi-expert Approach
• Nature's All-in-One: Multitasking Robots Inspired by Dung Beetles
• Efficient Learning of Control Policies for Robust Quadruped Bounding using Pretrained Neural Networks
• Real-time Digital Double Framework to Predict Collapsible Terrains for Legged Robots
• An Interpretable Neural Control Network with Adaptable Online Learning for Sample Efficient Robot Locomotion Learning
• Growable and Interpretable Neural Control with Online Continual Learning for Autonomous Lifelong Locomotion Learning Machines
• Agile perceptive multi-skill locomotion for quadrupedal robots in the wild