Learning on the Fly: Rapid Policy Adaptation via Differentiable Simulation

Jiahe Pan, Jiaxu Xing, Rudolf Reiter, Yifan Zhai, Elie Aljalbout, Davide Scaramuzza

Reviewer: ~Jiahe_Pan1

Keywords: Robot Learning, Differentiable Simulation, Continual Learning, Aerial Robots

TL;DR: We introduce an online, real-world policy learning framework based on residual dynamics learning and differentiable simulation that enables rapid continual adaptation of control policies to unknown, varying real-world dynamics.

Abstract: Learning control policies in simulation enables rapid, safe, and cost-effective development of advanced robotic capabilities. However, transferring these policies to the real world remains difficult due to the sim-to-real gap, where unmodeled dynamics and environmental disturbances can degrade policy performance. Existing approaches, such as domain randomization and Real2Sim2Real pipelines, can improve policy robustness, but either struggle under out-of-distribution conditions or require costly offline retraining. In this work, we approach these problems from a different perspective. Instead of relying on diverse training conditions before deployment, we focus on rapidly adapting the learned policy in the real world in an online fashion. To achieve this, we propose a novel online adaptive learning framework that unifies residual dynamics learning with real-time policy adaptation inside a differentiable simulation. Starting from a simple dynamics model, our framework continuously refines the model using real-world data to capture unmodeled effects and disturbances, such as payload changes and wind. The refined dynamics model is embedded in a differentiable simulation framework, enabling gradient backpropagation through the dynamics and thus rapid, sample-efficient policy updates beyond the reach of classical RL methods like PPO. All components of our system are designed for rapid adaptation, enabling the policy to adjust to unseen disturbances within 5 seconds of training. We validate the approach on agile quadrotor control under various disturbances in both simulation and the real world. Our framework reduces hovering error by up to 81% compared to L1-MPC and 55% compared to DATT, while also demonstrating robustness in vision-based control without explicit state estimation.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 14