Robust TS-ANFIS MPC of an Autonomous Racing Electrical Vehicle Considering the Battery State of Charge

In this work, the trajectory tracking problem of an autonomous racing electrical vehicle is addressed. Accordingly, a two-layer control scheme is developed, such that stability, recursive feasibility, and constraint satisfaction are guaranteed. The outer layer includes a zonotopic tube-based predictive control to ensure trajectory tracking while minimizing energy consumption considering the state of charge of the vehicle's battery. Meanwhile, the inner layer combines a linear quadratic zonotopic controller with a zonotopic Kalman filter to reduce the effect of exogenous disturbances and modeling errors. Moreover, for control and estimation purposes, a data-driven Takagi-Sugeno (TS) model trained by an adaptive neuro-fuzzy inference system (ANFIS) is employed. To illustrate the performance of the proposed control scheme, a simulated 1/10 Scale RC car is used.

Automated summary

This work addresses the trajectory tracking problem of an autonomous racing electrical vehicle and proposes a two-layer control scheme to guarantee stability, recursive feasibility, and constraint satisfaction. The outer layer uses a zonotopic tube-based predictive control to ensure trajectory tracking while minimizing energy consumption. The inner layer combines a linear quadratic zonotopic controller with a zonotopic Kalman filter to reduce the effect of exogenous disturbances and modeling errors. Additionally, a data-driven Takagi-Sugeno (TS) model trained by an adaptive neuro-fuzzy inference system (ANFIS) is employed for control and estimation purposes. The performance of the proposed control scheme is illustrated using a simulated 1/10 Scale RC car.