InstantMimic speeds up physics learning for realistic character control
InstantMimic: A High Performance System for Learning Physics-based Skills in Seconds
GraphicsRobotics
Summary
Making computer characters move realistically using physical rules takes a lot of computing time, slowing down how quickly these skills can be learned. The authors found that existing methods waste computing power, especially when using both CPUs and GPUs inefficiently. They created InstantMimic, a tool that runs the whole learning process entirely on the GPU, cutting training time to just a few seconds. This makes it easier and faster to teach characters complicated physical skills and allows for quick experimentation with settings.
What this means in practice
- •For game developers: Train physics-based character motions rapidly to create realistic animations in games within seconds instead of hours.
- •For robotics engineers: Enable fast learning and testing of control policies for robots that require complex physical interactions and agile movements.
Authors
Ikjun Choi, Geonho Leem, Jungdam Won
Abstract
Physics-based character control is a long-standing challenge in computer graphics and robotics, requiring policies that satisfy complex dynamics while producing realistic motion. Recent Deep RL approaches, particularly imitation learning methods such as DeepMimic, have had broad impact beyond animation, influencing robotics by enabling agile and expressive behaviors. While these approaches achieve impressive results, they remain computationally inefficient to train in practice. Despite GPU-accelerated simulation, we find that end-to-end pipelines often underutilize hardware due to overheads outside the physics solver, caused by fragmented GPU kernels and CPU memory access in the critical path. We present InstantMimic, a system that addresses these inefficiencies by making the entire training loop GPU-native. Built on a GPU-native physics backend, our unified pipeline integrates simulation, environment computation, policy inference, and policy updates within a single execution flow. As a result, InstantMimic reduces training time for diverse physics-based skills to a few seconds and makes LLM-agent-driven hyperparameter search practical.