VEH-Sim: An Integrated Vehicle, Engine, and Human Driver Simulation Platform

Principal Investigators Hosam K. Fathy (U. of Michigan) Omer Tsimhoni (UMTRI) University Researcher Jeffrey L. Stein, (U. of Michigan) Matthew Reed (UMTRI) Government David Gunter (TARDEC) Student Tulga Ersal (U. of Michigan)

Figure 1: Multibody Model of HMMWV Dynamics Figure 2: Snapshot of Integrated Vehicle/Driver Simulation

This project is motivated by the need for a modeling and simulation platform that can capture the dynamic interactions between human drivers, external stimuli, vehicle safety. Such a platform is envisioned as a cornerstone of a driver training system that can enhance the skill level with which drivers can execute sharp maneuvers with different vehicles at different load levels without incurring dynamic instabilities such as rollover. Such a platform can also enable in-depth analyses of the interactions between vehicle motion, safety systems, and human cognitive processes/mental load. This will make it possible to examine the viability of different vehicle safety systems for different levels of vehicle loading, driver warning, and driver training.

Previous research in Thrust Areas 1 and 4 furnished VESim: an integrated vehicle and engine simulation package capable of simulating a variety of powertrain systems at different levels of fidelity and complexity. Previous research in both Thrust Area 2 and UMTRI furnished virtual human driver models capable of modeling human driver cognitive processes and mental load. A 2007 ARC conference case study led by the investigators examined, for the first time, the combined use of VESim and these virtual driver models for assessing the viability of a vehicle safety systems (specifically, a load-adapting rollover warning system for HMMWVs). In this case study, we built a high-fidelity multibody dynamics model for the HMMWV (Figure 1) and combined it with the virtual driver to model driver reactions to load-adapting rollover warnings, and therefore assess the viability of such warnings (Figure 2).

The above case study demonstrated the power of combined driver/vehicle simulation, but it had three main limitations. First, it did not integrate the VESim and virtual driver models fully. Instead, it allowed them to exchange data offline, in an iterative fashion. Secondly, it only utilized VESim as a vehicle dynamics simulator, and thus did not capitalize on VESim’s full capabilities as both a vehicle and powertrain system simulator. Finally, while it modeled the human cognitive decision-making process using a high-fidelity platform, it used fairly simple stochastic dynamic models for the execution of such decisions (i.e., the continuous steering commands from the driver). Other research in the ARC, in UMTRI, and at other institutions (e.g., by Peng, Gordon, MacAdam, Ulsoy, etc.) has furnished very sophisticated models of how human drivers execute their decisions (e.g., steer continuously after making a cognitive decision to change lanes). Such models were not fully utilized in the above preliminary case study.

To address the above challenges, we propose a new platform, VEH-Sim, for integrated vehicle, powertrain, and human driver simulation. VEH-Sim will provide both the army and industry with a powerful platform for analyzing the dynamic interplay between human drivers, external stimuli, vehicle safety systems, and vehicle dynamics. Such a platform will then make it possible to thoroughly analyze the viability of different vehicle safety systems, especially those that involve driver warning. Such a platform is also envisioned as the cornerstone of a driver training system that can enhance the skill level with which drivers can execute sharp maneuvers with different vehicles at different load levels without incurring dynamic instabilities such as rollover.