Neural Solvers for Fast and Accurate Numerical Optimal Control
Neural Solvers for Fast and Accurate Numerical Optimal Control
复制标题
DOI：
10.48550/arxiv.2203.08072
复制
发表时间：
2022-03
期刊：
影响因子：
--
通讯作者：
Federico Berto;Stefano Massaroli;Michael Poli;Jinkyoo Park
中科院分区：
文献类型：
--
作者：
Federico Berto;Stefano Massaroli;Michael Poli;Jinkyoo Park
为动态系统综合最优控制器通常涉及解决具有硬实时约束的优化问题。这些限制决定了可以应用的数值方法的类别：计算昂贵但准确的数值例程被快速和不准确的方法所取代，以推理时间换取解决方案的准确性。本文提供了技术，以提高质量的优化控制策略给定一个固定的计算预算。我们通过一个超求解器的方法，它混合了微分方程求解器和神经网络实现上述目标。在低维度和高维度的直接和滚动时域最优控制任务中评估了性能，其中所提出的方法在解决方案精度和控制性能方面表现出一致的帕累托改进。
Synthesizing optimal controllers for dynamical systems often involves solving optimization problems with hard real-time constraints. These constraints determine the class of numerical methods that can be applied: computationally expensive but accurate numerical routines are replaced by fast and inaccurate methods, trading inference time for solution accuracy. This paper provides techniques to improve the quality of optimized control policies given a fixed computational budget. We achieve the above via a hypersolvers approach, which hybridizes a differential equation solver and a neural network. The performance is evaluated in direct and receding-horizon optimal control tasks in both low and high dimensions, where the proposed approach shows consistent Pareto improvements in solution accuracy and control performance.