Robot learns missing physics to improve task planning
EMPIRIC: Experiment-Driven Learning of Residual World Models for Robot Planning
RoboticsArtificial IntelligenceMachine Learning
Summary
Robots often use physics engines to predict how things will move, but these engines can miss complex effects like glue drying or wind. The authors created EMPIRIC, a system that adds extra learned rules to the physics engine to fill in what’s missing. EMPIRIC experiments to understand unknown object behaviors, adjusts its models from noisy measurements, and uses these models to plan better actions. This approach helps robots solve tasks more efficiently with fewer tries, both in simulations and on a real robot.
What this means in practice
- •For robot developers: Enable robots to learn and model unknown physical effects for improved manipulation and planning in diverse environments.
- •For industrial automation teams: Enhance robotic systems used in manufacturing by integrating learned residual models of object interactions to reduce trial-and-error actions.
Authors
Yichao Liang, Amber Li, Dat Nguyen, Emily Bunnapradist, Michelangelo Naim, Sreela Kodali, Matteo Merler, Bowen Li, Kiran Gopinathan, Yiyun Liu, Nikhil Pimpalkhare, Joshua B. Tenenbaum, Adrian Weller, Zenna Tavares, Tom Silver, Kevin Ellis
Abstract
A robot should be able to learn through experiments how unfamiliar objects behave and interact, then plan with that knowledge. It need not start from scratch: physics engines supply knowledge of motion and contact, but can omit entire mechanisms, such as glue curing, water heating, or wind. We present EMPIRIC, an agent that learns a residual world model: a physics engine extended with code for the missing mechanisms. The learned programs can introduce new forces, constraints, and hidden state, and Bayesian inference estimates their parameters and states from noisy observations. The resulting model lets the agent predict the outcomes of actions, choose informative experiments, and revise its hypotheses when predictions fail. Across five simulated domains, EMPIRIC learns interpretable, reusable models, and solves more tasks with fewer environment interactions than all three baselines. On a physical robot, it learns wind forces and domino masses to solve a manipulation task. Website and code: https://yichao-liang.github.io/empiric