arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2403.00748math.OCcs.RO
收藏

原始-对偶 iLQR

Primal-Dual iLQR

Apple(苹果公司);Polytechnique Montréal(蒙特利尔理工学院)

João Sousa-Pinto, Dominique Orban

AI总结:

本文提出原始-对偶 iLQR 算法,将 SQP 与增广拉格朗日原始-对偶评价函数结合,利用 LQR 求解牛顿-KKT 系统,实现全局收敛且高度可并行化的非线性离散时间最优控制求解。

摘要:

我们提出了一种用于求解无约束离散时间最优控制问题的新算法。我们的方法采用直接多重打靶法,并将 SQP 方法与 ℓ2 增广拉格朗日原始-对偶评价函数相结合。我们使用 LQR 算法高效求解原始-对偶牛顿-KKT 系统。由于我们的算法是 NPSQP 的一个特化版本,它继承了其通用性质,包括全局收敛性、快速局部收敛性,并且无需二阶校正或维度扩展,从而改进了现有的直接多重打靶方法,如 acados、ALTRO、GNMS、FATROP 和 FDDP。我们的算法所提出的 LQR 形式子问题的解可以并行化,其运行时间与阶段数、状态数和控制数的对数成正比。此外,由于我们的方法避免了非线性动力学的顺序前向模拟,因此每次线搜索迭代可以在 O(1) 并行时间内完成。因此,本文提供了一种实用、理论可靠且高度可并行化(例如使用 GPU)的方法,用于求解非线性离散时间最优控制问题。该算法的开源 JAX 实现可在 GitHub(joaospinto/primal_dual_ilqr)上找到。

英文摘要:

We introduce a new algorithm for solving unconstrained discrete-time optimal control problems. Our method follows a direct multiple shooting approach, and consists of applying the SQP method together with an $\ell_2$ augmented Lagrangian primal-dual merit function. We use the LQR algorithm to efficiently solve the primal-dual Newton-KKT system. As our algorithm is a specialization of NPSQP, it inherits its generic properties, including global convergence, fast local convergence, and the lack of need for second order corrections or dimension expansions, improving on existing direct multiple shooting approaches such as acados, ALTRO, GNMS, FATROP, and FDDP. The solutions of the LQR-shaped subproblems posed by our algorithm can be be parallelized to run in time logarithmic in the number of stages, states, and controls. Moreover, as our method avoids sequential rollouts of the nonlinear dynamics, it can run in $O(1)$ parallel time per line search iteration. Therefore, this paper provides a practical, theoretically sound, and highly parallelizable (for example, with a GPU) method for solving nonlinear discrete-time optimal control problems. An open-source JAX implementation of this algorithm can be found on GitHub (joaospinto/primal_dual_ilqr).

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑