Comparative Study of Prediction Models for Model Predictive Path- Tracking Control in Wide Driving Speed Range

Abstract

Four-wheel independent drive and steering vehicle (4WIDS Vehicle, Swerve Drive Robot) has the ability to move in any direction by its eight degrees of freedom (DoF) control inputs. Although the high maneuverability enables efficient navigation in narrow spaces, obtaining the optimal command is challenging due to the high dimension of the solution space. This paper presents a navigation architecture using the Model Predictive Path Integral (MPPI) control algorithm to avoid collisions with obstacles of any shape and reach a goal point.The key idea to make the problem easier is to explore the optimal control input in a reasonably reduced dimension that is adequate for navigation. Through evaluation in simulation, we found that the selecting sampling space of MPPI greatly affects navigation performance. In addition, our proposed controller which switches multiple sampling spaces according to the realtime situation can achieve balanced behavior between efficiency and safety. Source code is available at https://github. com/MizuhoAOKI/mppi_swerve_drive_ros.

Connection of nonlinear model predictive controllers for smooth task switching in autonomous driving

“…It is also useful when the vehicle dynamics model in the MPC is switched depending on the situation. As described in [39], the appropriate vehicle dynamics model for path following differs depending on vehicle speed. In addition, these models have different state spaces.…”

“…In (), the parameters $$$ {a}_{ij} $$$ and $$$ {b}_k $$$ $$$ \left(i;j;k=1,2\right) $$$ are constant parameters in the DBM calculated from the vehicle physical parameters. For details, please refer to [39]. In addition, $$$ {V}_{o1} $$$ and $$$ {V}_{o2} $$$ are the vehicle speeds of cars 1 and 2, respectively, and the constant velocity motion is assumed for the longitudinal movement; $$$ {\dot{p}}_x^1\simeq {V}_{o1} $$$ and $$$ {\dot{p}}_x^2\simeq {V}_{o2} $$$.…”

“…The objective of MPC (AB) is to guide the state to the feasible set of MPC (B) while continuing the execution of MPC (A) before executing the task of MPC (B). The cost function (39) reflects this objective. The cost function of MPC (AB) is defined as the summation of the cost function of MPC (A) and penalty functions [25].…”

“…The state is inside C AB in (54). In MPC (AB), the penalty function in (39) smoothly guides the state to A AB in (52), represented by the orange box in Figure 6. (iv) The mode switches to MPC (B) at the time it enters A AB in (52).…”

MPC Builder for Autonomous Drive: Automatic Generation of MPCs for Motion Planning and Control

“…Aggregator selects the lateral and longitudinal task primitives corresponding to the given driving task commands. Next, the Aggregator selects the appropriate ego dynamics primitive, KBM primitive for slow speeds (≤ 5.0 m/s), and the dynamic bicycle model primitive otherwise, based on the measured vehicle-speed range [26]. Finally, Aggregator adds multiple safety primitives according to the number of observations surrounding the others.…”