Scalable character animation based on reinforcement learning

It's natural to see virtual characters and moves reasonably in terms of the laws of physics, that is, human-like, which has been a subject of long-standing research in the field of gaming as well as computer graphics.

In the project "A Scalable Approach to Control Diverse Behaviors for Physically Simulated Characters", a project in which Facebook Jungdam Won participated as the first author, divide various motion sets into clusters, learn “experts” optimized for each cluster, and then It deals with learning how to combine to create a single “general controller” that can handle various actions.

Basically, it is an approach based on reinforcement learning, but I think it is highly useful in that it can be easily extended by adding any action. Share the project page link and the github link.

A Scalable Approach to Control Diverse Behaviors for Physically Simulated Characters – Facebook Research

Human characters with a broad range of natural looking and physically realistic behaviors will enable the construction of compelling interactive experiences. In this paper, we develop a technique for learning controllers for a large set of heterogeneous behaviors.

facebookresearch/ScaDiver

Project for the paper “A Scalable Approach to Control Diverse Behaviors for Physically Simulated Characters” – facebookresearch/ScaDiver