Learning Agile Locomotion via Adversarial Training

08/03/2020
by   Yujin Tang, et al.
5

Developing controllers for agile locomotion is a long-standing challenge for legged robots. Reinforcement learning (RL) and Evolution Strategy (ES) hold the promise of automating the design process of such controllers. However, dedicated and careful human effort is required to design training environments to promote agility. In this paper, we present a multi-agent learning system, in which a quadruped robot (protagonist) learns to chase another robot (adversary) while the latter learns to escape. We find that this adversarial training process not only encourages agile behaviors but also effectively alleviates the laborious environment design effort. In contrast to prior works that used only one adversary, we find that training an ensemble of adversaries, each of which specializes in a different escaping strategy, is essential for the protagonist to master agility. Through extensive experiments, we show that the locomotion controller learned with adversarial training significantly outperforms carefully designed baselines.

READ FULL TEXT
research
11/11/2020

Learning Agile Locomotion Skills with a Mentor

Developing agile behaviors for legged robots remains a challenging probl...
research
04/02/2020

Learning Agile Robotic Locomotion Skills by Imitating Animals

Reproducing the diverse and agile locomotion skills of animals has been ...
research
09/21/2022

Recent Approaches for Perceptive Legged Locomotion

As both legged robots and embedded compute have become more capable, res...
research
10/05/2016

Supervision via Competition: Robot Adversaries for Learning Tasks

There has been a recent paradigm shift in robotics to data-driven learni...
research
07/02/2022

Learning fast and agile quadrupedal locomotion over complex terrain

In this paper, we propose a robust controller that achieves natural and ...
research
03/08/2023

Breaking Symmetries Leads to Diverse Quadrupedal Gaits

Symmetry manifests itself in legged locomotion in a variety of ways. No ...
research
11/11/2020

Zero-Shot Terrain Generalization for Visual Locomotion Policies

Legged robots have unparalleled mobility on unstructured terrains. Howev...

Please sign up or login with your details

Forgot password? Click here to reset