Publication

We present a threat assessment algorithm for driver assistance systems in which mathematical vehicle and driver models are used to explicitly account for the vehicle and driver behavior. For the application considered in this paper, requirements that the vehicle stays on the road while operating in a stable operating region are expressed as constraints on the vehicle states which need to be satisfied over a finite time horizon. The threat assessment problem is then formulated as a constraint satisfaction problem, solved through interval-based techniques considering bounded uncertainty in the measurements. Experimental results demonstrate the algorithms ability to detect unsafe operation in advance.