Pendulum and Cartpole Benchmark Domains

该数据集包含了一系列用于评估可微分的MPC求解器和模仿学习在摆动和倒立摆杆领域任务中的表现。此外，该数据集还用于展示所提出方法恢复控制器成本和动态的能力，并与神经网络策略以及传统的系统识别方法进行了比较。该数据集的任务是模仿学习。

This dataset includes a series of tasks for evaluating the performance of differentiable MPC solvers and imitation learning in the domain of swinging and inverted pendulum systems. Additionally, it is utilized to demonstrate the capability of the proposed method in recovering controller costs and system dynamics, as well as to make comparisons against neural network policies and traditional system identification methods. The core task of this dataset is imitation learning.