Fast Model Predictive Control using the Primal Barrier Method
One of the major areas of focus in the field of autonomous vehicles is safe navigation and control. While there are a plethora of techniques available for control, one of the most studied technique is Model Predictive Control (MPC) of vehicle. An even faster way is the Fast MPC implementation which improves the speed of standard MPC by exploiting the structure of the problem that suits the particular problem. In this paper, we discuss the different aspects of design and implementation of a fast MPC model on the 1/10th scale F1 race car. A nonlinear bicycle model is linearized and used. We used some techniques that can be incorporated on MPC to reach real-time constraints by reducing the computation cost of optimization. Fast MPC provides the suboptimal solution yet satisfies the inequality constrains. This way we can be sure that the control strategy will never produce the control input to the system which can’t be implementable.
The error in x and y was :
