AdaptNet: Policy Adaptation for Physics-Based Character Control

Abstract

Motivated by humans' ability to adapt skills in the learning of new ones, this paper presents AdaptNet, an approach for modifying the latent space of existing policies to allow new behaviors to be quickly learned from like tasks in comparison to learning from scratch. Building on top of a given reinforcement learning controller, AdaptNet uses a two-tier hierarchy that augments the original state embedding to support modest changes in a behavior and further modifies the policy network layers to make more substantive changes. The technique is shown to be effective for adapting existing physics-based controllers to a wide range of new styles for locomotion, new task targets, changes in character morphology and extensive changes in environment. Furthermore, it exhibits significant increase in learning efficiency, as indicated by greatly reduced training times when compared to training from scratch or using other approaches that modify existing policies. Code is available at https://motion-lab.github.io/AdaptNet.

SIGGRAPH Asia 2023. Video: https://youtu.be/WxmJSCNFb28. Website: https://motion-lab.github.io/AdaptNet, https://pei-xu.github.io/AdaptNet

References in corpus (6)

• Character Controllers Using Motion VAEs
• Robust Motion In-betweening
• ASE: Large-Scale Reusable Adversarial Skill Embeddings for Physically Simulated Characters
• Unpaired Motion Style Transfer from Video to Animation
• ControlVAE: Model-Based Learning of Generative Controllers for Physics-Based Characters
• Composite Motion Learning with Task Control