English

DiffTune: Auto-Tuning through Auto-Differentiation

Robotics 2024-07-12 v3

Abstract

The performance of robots in high-level tasks depends on the quality of their lower-level controller, which requires fine-tuning. However, the intrinsically nonlinear dynamics and controllers make tuning a challenging task when it is done by hand. In this paper, we present DiffTune, a novel, gradient-based automatic tuning framework. We formulate the controller tuning as a parameter optimization problem. Our method unrolls the dynamical system and controller as a computational graph and updates the controller parameters through gradient-based optimization. The gradient is obtained using sensitivity propagation, which is the only method for gradient computation when tuning for a physical system instead of its simulated counterpart. Furthermore, we use L1\mathcal{L}_1 adaptive control to compensate for the uncertainties (that unavoidably exist in a physical system) such that the gradient is not biased by the unmodelled uncertainties. We validate the DiffTune on a Dubin's car and a quadrotor in challenging simulation environments. In comparison with state-of-the-art auto-tuning methods, DiffTune achieves the best performance in a more efficient manner owing to its effective usage of the first-order information of the system. Experiments on tuning a nonlinear controller for quadrotor show promising results, where DiffTune achieves 3.5x tracking error reduction on an aggressive trajectory in only 10 trials over a 12-dimensional controller parameter space.

Keywords

Cite

@article{arxiv.2209.10021,
  title  = {DiffTune: Auto-Tuning through Auto-Differentiation},
  author = {Sheng Cheng and Minkyung Kim and Lin Song and Chengyu Yang and Yiquan Jin and Shenlong Wang and Naira Hovakimyan},
  journal= {arXiv preprint arXiv:2209.10021},
  year   = {2024}
}

Comments

Minkyung Kim and Lin Song contributed equally to this work. Accepted for publication by IEEE Transactions on Robotics in July 2024

R2 v1 2026-06-28T01:46:40.093Z