Needs for Physical Models and Related Methods for Development of Automated Road Vehicles

Other conference contribution, 2019

Road vehicle automation is a trendy topic in research and media. Some automation such as adaptive cruise control and lane keeping is already on market, while totally driverless vehicles in real everyday transportation tasks is still only a vision. How far the products will reach and when can be debated, but any step forward will benefit of, or need, support of modelling formats and related methods.

The presentation will explain typical functional/control architecture for vehicles and then focus on development of the partitions Traffic Situation Management and Vehicle Motion Management, assuming that the partition Environment Observation (image processing etc.) comes in place to a certain known maturity. Models will be used with methods for on-line optimization, on-line predictive simulation, virtual off-line verification for right and faulty function, system safety, simulations for legal assessment of long combination vehicles, etc. A goal for novel methods is that, with physical non-causal models, one can avoid some of the enormous amount of verification simulations needed if only input/output (fixed causality) models are used.

Examples from heavy vehicles and passenger cars will be given; what is achieved and what are the challenges for next steps.