OPTEC - K.U.Leuven Center of Excellence: Optimization in Engineering

"Time-Optimal Control of Race Cars with Gear Shifts"

Janick Frasch
Interdisciplinary Center for Scientific Computing (IWR), Heidelberg University


Abstract:
Control of autonomous vehicles and providing recommendations to drivers in real time are challenging tasks from an algorithmic point of view. To include realistic effects, such as nonlinear tire dynamics and integer gear shifts, at least medium-sized mathematical models need to be considered. Yet, fast feedback is of utmost importance. Existing Nonlinear Model Predictive Control (NMPC) algorithms need to be enhanced to comply with these two contradictory requirements. As the testing of algorithms in an automatic driving context is cumbersome and expensive, we propose a virtual testbed for NMPC of driving cars. We use the open source race simulator VDrift as virtual real world, in which algorithms need to cope with the mismatch between the detailed physical model in the simulator and a coarser approximative model used for NMPC. We present the general framework of this virtual environment and an optimal control problem based on a medium-sized ordinary differential equation model and a generic and flexible parameterization of the track constraint. We discuss one possible algorithmic approach to the task of minimum time driving including gear shifts.


"Air/Fuel Ratio Model Predictive Control of a Spark Ignition Engine"

Matus Kopacka
Slovak University of Technology, Bratislava, Slovakia


Abstract:
The topic of the presentation will be a control of an air/fuel ratio (AFR) of a spark ignition engine utilizing the model predictive controller based on the multi-model approach. The multi-model approach employs the autoregressive model (ARX) network, using the weighting of local models. The weighted ARX models are identified in the particular working points and are creating a global engine model, covering its nonlinearity. Awaited improvement of a proper air/fuel mixture combusted in a cylinder is mostly gained in the transient working regimes of an engine. In these regimes, the traditional control approach looses its quality, compared to steady state working regimes of an engine. This leads to higher fuel consumption and level of emissions from an engine.
There will be presented results of the real-time control of the VW Polo 1390cm3 engine, at which the original electronic control unit (ECU) has been replaced by a dSpace system executing the model predictive controller.