Model Predictive Path Research Articles

Risk Analysis for Cooperation Between the Driver and the Control System of an Autonomous Vehicle

Autonomous vehicles will be a reality in our society shortly, first coming vehicle will come with level 3 of automation where the driver is expected to be available for occasional control either for pleasure or during critical situations. This article proposes a strategy that aims to predict the path desired by the driver through steering actions and the algorithms used by of the autonomous vehicle to utilize risk indicators. These risk indicators can be used to condition the cooperation and the transition from automatic to manual driving. This technique uses the data of a path model predictive controller and local path planning data based on a curvilinear space to predict the driver interaction and intention when he engages the steering wheel. Some results are presented in a processor-in-the-loop environment, which let to verify that driver intention can be integrated with the risk analysis of the autonomous vehicle with minimal computational cost. Some conclusions and future improvements to the system are proposed.

Model Predictive Path Integral Control: From Theory to Parallel Computation

In this paper, a model predictive path integral control algorithm based on a generalized importance sampling scheme is developed and parallel optimization via sampling is performed using a graphics processing unit. The proposed generalized importance sampling scheme allows for changes in the drift and diffusion terms of stochastic diffusion processes and plays a significant role in the performance of the model predictive control algorithm. The proposed algorithm is compared in simulation with a model predictive control version of differential dynamic programming on nonlinear systems. Finally, the proposed algorithm is applied on multiple vehicles for the task of navigating through a cluttered environment. The current simulations illustrate the efficiency and robustness of the proposed approach and demonstrate the advantages of computational frameworks that incorporate concepts from statistical physics, control theory, and parallelization against more traditional approaches of optimal control theory.

A306 小型前輪操舵車両を用いた加速度の制約を考慮したモデル予測経路追従制御(OS10 ビークルの運動と制御1)

A306 小型前輪操舵車両を用いた加速度の制約を考慮したモデル予測経路追従制御(OS10 ビークルの運動と制御1)

Model Predictive Sliding Mode Control for Four Wheel Steering and Driving Vehicles

In this paper, for four wheel steering and driving vehicles, a model predictive path tracking control with robustness with respect to external unknown disturbance is proposed. Vehicle dynamics are constrained onto the sliding mode so that unknown external disturbance does not affect the performance of the system. Then nonlinear vehicle dynamics are transformed into time-state control form, and linearized so that the model predictive control is computed faster. The proposed method is applied to the tasks of the JSAE-SICE benchmark problem and its advantage is numerically proved.

A Novel Method for SUAV Path Planning Based on MPC

Path planning has always being one of the most significant study fields in Small UAV researching. And Model Predictive Control (MPC) is a special strategy in obtaining the control actions which were achieved by solving a finite horizon optimal control problem at each instant. The paper advanced a novel method which was called Model Predictive Path Planning Strategy (MPPS) based on MPC to deal with the SUAV path planning problem and a responding predictive planner was put forward to generate an effective path for SUAV in simulative urban environment. The results of the simulation show that the advanced method can be used to plan path for SUAV.

Navigation system for agricultural machines: Nonlinear Model Predictive path tracking

This article presents a new kind of navigation system for agricultural machines. The focus is on trajectory control where a Nonlinear Model Predictive path tracking for tractor and trailer system is presented. The experiments of the proposed method are carried out by using real agricultural machines in real environments.The agricultural objective is to drive so that swaths are exactly side-by-side, without overlapping or gaps. Hence, the objective of this research was to control the lateral position of the towed implement and to keep it close to the adjacent driving line. The adjacent driving line was recognized locally by using a 2D laser scanner. The local measurement and global position information was merged with the help of an Extended Kalman Filter (EKF). The measurement of the heading by GPS was improved by using an inertial measurement unit and a separate EKF filter. The position of the implement was controlled by steering the tractor and by the use of a hydraulically controlled joint. Because there were two actuators which affected the position of the implement, the problem was a multivariable control problem. Nonlinear Model Predictive Control (NMPC) was used to accomplish the navigation task.The goal was to build a system, which is able to have at least the same accuracy as a human driver. The sufficient accuracy requirement was at most 10cm lateral error at a speed of 12km/h. The results presented in the article show that the goal was met and NMPC is a feasible method for accurate path tracking. Further investigation is, however, needed to adapt the method to other kinds of agricultural machines.