Acceleration of Gradient-based Path Integral Method for Efficient Optimal and Inverse Optimal Control

نویسندگان

  • Masashi Okada
  • Tadahiro Taniguchi
چکیده

This paper deals with a new accelerated path integral method, which iteratively searches optimal controls with a small number of iterations. This study is based on the recent observations that a path integral method for reinforcement learning can be interpreted as gradient descent. This observation also applies to an iterative path integral method for optimal control, which sets a convincing argument for utilizing various optimization methods for gradient descent, such as momentum-based acceleration, step-size adaptation and their combination. We introduce these types of methods to the path integral and demonstrate that momentum-based methods, like Nesterov Accelerated Gradient and Adam, can significantly improve the convergence rate to search for optimal controls in simulated control systems. We also demonstrate that the accelerated path integral could improve the performance on model predictive control for various vehicle navigation tasks. Finally, we represent this accelerated path integral method as a recurrent network, which is the accelerated version of the previously proposed path integral networks (PI-Net). We can train the accelerated PI-Net more efficiently for inverse optimal control with less RAM than the original PI-Net.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

کنترل بهینة شار حرارتی سطحی در یک جسم دوبعدی با خواص وابسته به دما

In this paper the optimal control of boundary heat flux in a 2-D solid body with an arbitrary shape is performed in order to achieve the desired temperature distribution at a given time interval. The boundary of the body is subdivided into a number of components. On each component a time-dependent heat flux is applied which is independent of the others. Since the thermophysical properties are t...

متن کامل

Near-Minimum-Time Motion Planning of Manipulators along Specified Path

The large amount of computation necessary for obtaining time optimal solution for moving a manipulator on specified path has made it impossible to introduce an on line time optimal control algorithm. Most of this computational burden is due to calculation of switching points. In this paper a learning algorithm is proposed for finding the switching points. The method, which can be used for both ...

متن کامل

A meshless method for optimal control problem of Volterra-Fredholm integral equations using multiquadratic radial basis functions

In this paper, a numerical method is proposed for solving optimal control problem of Volterra integral equations using radial basis functions (RBFs) for approximating unknown function. Actually, the method is based on interpolation by radial basis functions including multiquadrics (MQs), to determine the control vector and the corresponding state vector in linear dynamic system while minimizing...

متن کامل

Solving optimal control problems with integral equations or integral equations - differential with the help of cubic B-spline scaling functions and wavelets

In this paper, a numerical method based on cubic B-spline scaling functions and wavelets for solving optimal control problems with the dynamical system of the integral equation or the differential-integral equation is discussed. The Operational matrices of derivative and integration of the product of two cubic B-spline wavelet vectors, collocation method and Gauss-Legendre integration rule for ...

متن کامل

Inverse Kinematics Resolution of Redundant Cooperative Manipulators Using Optimal Control Theory

The optimal path planning of cooperative manipulators is studied in the present research. Optimal Control Theory is employed to calculate the optimal path of each joint choosing an appropriate index of the system to be minimized and taking the kinematics equations as the constraints. The formulation has been derived using Pontryagin Minimum Principle and results in a Two Point Boundary Value Pr...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:
  • CoRR

دوره abs/1710.06578  شماره 

صفحات  -

تاریخ انتشار 2017