Implicit Contact Dynamics Modeling With Explicit Inertia Matrix Representation for Real-Time, Model-Based Control in Physical Environment
Implicit Contact Dynamics Modeling With Explicit Inertia Matrix Representation for Real-Time, Model-Based Control in Physical Environment
复制标题
DOI：
10.1162/neco_a_01465
复制
发表时间：
2021-11
影响因子：
2.9
通讯作者：
Takeshi D. Itoh;K. Ishihara;J. Morimoto
中科院分区：
文献类型：
--
作者：
Takeshi D. Itoh;K. Ishihara;J. Morimoto
摘要基于模型的控制具有很高的采样效率，在实际机器人中具有很大的应用潜力。然而，对于实际的机器人控制任务，如精确操作，处理物理接触和产生准确的运动是不可避免的。对于基于模型的实时方法，需要精确移动的接触丰富任务的困难在于，模型需要在有限的时间长度内准确预测即将发生的接触事件，而不是在之后使用传感器检测它们。因此，在这项研究中，我们研究神经网络模型是否以及如何学习一个对基于模型的控制足够有用的任务相关模型，即预测包括接触事件在内的未来状态的模型。为此，我们提出了一种结构化神经网络模型预测控制(SNN-MPC)方法，其神经网络结构采用显式惯性矩阵表示。为了训练所提出的网络，我们从有限数量的样本中开发了一个两阶段的接触丰富动态建模过程。作为一项接触丰富的任务，我们使用物理的3-DOF手指机器人来执行轨迹球操作任务。结果表明，SNN-MPC在操作任务上优于传统的全连通网络模型。
Abstract Model-based control has great potential for use in real robots due to its high sampling efficiency. Nevertheless, dealing with physical contacts and generating accurate motions are inevitable for practical robot control tasks, such as precise manipulation. For a real-time, model-based approach, the difficulty of contact-rich tasks that requires precise movement lies in the fact that a model needs to accurately predict forthcoming contact events within a limited length of time rather than detect them afterward with sensors. Therefore, in this study, we investigate whether and how neural network models can learn a task-related model useful enough for model-based control, that is, a model predicting future states, including contact events. To this end, we propose a structured neural network model predictive control (SNN-MPC) method, whose neural network architecture is designed with explicit inertia matrix representation. To train the proposed network, we develop a two-stage modeling procedure for contact-rich dynamics from a limited number of samples. As a contact-rich task, we take up a trackball manipulation task using a physical 3-DoF finger robot. The results showed that the SNN-MPC outperformed MPC with a conventional fully connected network model on the manipulation task.