Muscle driven character animation controlled by reinforcement learning
RLMuscle: Muscle-Driven Character Animation via Reinforcement Learning
Graphics
Summary
Making animated characters look realistic is hard because muscles move and change shape in complex ways. This paper presents a way to automatically find how each muscle should activate so the character moves correctly, without artists doing it by hand. The method uses artificial intelligence to learn muscle controls and simulates how muscles contract in 3D for better realism. It also uses real muscle activity data to guide the process, making the movements more like those of humans.
What this means in practice
- •For game developers: Create more physically accurate and visually realistic animated characters by automatically generating muscle movements that match real motions.$Commercial implications: Enables games to feature lifelike characters with biomechanics-based muscle animations, providing improved immersion and visual fidelity.
- •For virtual reality content creators: Develop realistic human avatars with muscle-driven movements that respond naturally in immersive environments without manual tuning.
Authors
Chunlei Li, Siyuan Yu, Yang Gao, Shuai Li, Peng Yu, Aimin Hao, He Wang
Abstract
Body deformation caused by muscle contraction plays a vital role in the visual realism of character animation, yet in most 3D animation pipelines artists have to manually specify muscle activations, which is time-consuming and gives no guarantee of physical realism. To address this, we propose a new method in which a reinforcement learning (RL) controller discovers per-muscle activations that track reference motions without manual tuning. Existing RL muscle controllers stop at 1D muscles; we go one step further, keeping the 1D model inside the RL control loop as a fast surrogate, and using a 3D volumetric simulation to produce the final deformation. Because both stages share the very same constitutive law, the muscle deformation is produced by the same activations that generated the motion. We further add a measured electromyography (EMG) prior that steers the redundant activation solution toward real human muscle recruitment. To achieve stable, efficient, and realistic simulation, we propose a novel geometric multigrid for Extended Position-Based Dynamics (XPBD) that is built on a coarse tetrahedral cage with barycentric-coordinate prolongation and closes a V-cycle on tetrahedral meshes. Within this GPU-parallelized solver, we inject the same Hill-type force-length law as an energy-based fiber constraint that turns per-fiber activation into realistic muscle contraction and bulging. We demonstrate volumetric muscle simulation on a 194-muscle human body with a diverse set of motions (walk, kick, swim, overhead-squat, etc.), simulating the full-body model at 18 ms per frame on a single RTX 4090. Our work takes a step toward character animation that is not only visually convincing but also biomechanically grounded. The website with code and videos is at https://chunleili.github.io/project-page-rlmuscle/.