A Nonlinear Model Predictive Control Based Virtual Driver for High Performance Driving

Mattia Bruschetta1, Enrico Picotti2, Enrico Mion3, Yutao Chen4, Alessandro Beghi5, Diego Minen6

  • 1University of Padova
  • 2University of Padua
  • 3Università di Padova
  • 4Eindhoven University of Technology
  • 5Universita di Padova
  • 6VI-Grade srl

Details

10:50 - 11:10 | Mon 19 Aug | Lau, 5-203 | MoA1.2

Session: Control for Connected and Automated Vehicles

Abstract

Virtual prototyping is currently a widely used tool for the development of new cars. In this paper, the development of an effective virtual driver (VD) is described, that aims at reproducing real-time driver's behaviour, also at the limit of performance. The proposed VD model, a four-wheel vehicle with longitudinal load transfer and Pacejka's lateral tires forces model, has been implemented in the nonlinear model predictive control framework. The implementation is developed in MATMPC, a Matlab-based open-source toolbox, and tested in co-simulation with commercial software VI-CarRealTime (VI-CRT), specifically designed to reproduce vehicles behaviour. A challenging Double Lane Change (DLC) maneuver has been used to evaluate performance, showing great abilities of the proposed VD in handling track boundaries during high speed manoeuvring.