Reducing Warning Errors in Driver Support with Personalized Risk Maps

TL;DR

A warning system that estimates a personalized risk factor for the given driver based on the driver's behavior and is able to adapt the warning signal with personalized Risk Maps is proposed.

Abstract

We consider the problem of human-focused driver support. State-of-the-art personalization concepts allow to estimate parameters for vehicle control systems or driver models. However, there are currently few approaches proposed that use personalized models and evaluate the effectiveness in the form of general risk warning. In this paper, we therefore propose a warning system that estimates a personalized risk factor for the given driver based on the driver's behavior. The system afterwards is able to adapt the warning signal with personalized Risk Maps. In experiments, we show examples for longitudinal following and intersection scenarios in which the novel warning system can effectively reduce false negative errors and false positive errors compared to a baseline approach which does not use personalized driver considerations. This underlines the potential of personalization for reducing warning errors in risk warning and driver support.

Figures (8)

• Figure 1: The image shows a driving situation example of a dynamic car following in which the driver in the green car can react differently depending on the driver type. In this paper, we propose a driver support in the form of warning that can reduce driver warning errors by using these driver types in personalized Risk Maps.
• Figure 2: For the risk factor estimation, we plan driver behaviors by using the behavior planner Risk Maps with a defensive risk factor parametrization and a confident risk factor parametrization. By comparing the current driver behavior with the behavior plans, we can estimate the risk factor.
• Figure 3: The warning system consists of Risk Maps that uses the risk factor to create a personalized Risk Maps and a warning adaptation module that uses a personalized weight to the risk value to obtain the warning signal.
• Figure 4: The image shows the risk factor estimation module applied on the car following scenario example with a defensive driver type. The driver is braking and the module thus correctly estimates the driver to be a defensive driver type. The personalized risk factor is high for the driver. The dashed line in the risk factor plot (top right) represents the average of the estimations over the whole simulation time.
• Figure 5: The image shows examples of the risk factor estimation results for a normal driver and a confident driver for the car-following scenario. The estimation module allows to also estimate here the risk factor.