A distributed virtual driving simulator

Abstract

The authors' goal is to develop techniques for distributed driving simulation on low cost computers. Successful distributed environments have already been implemented for military and commercial applications (Macedonia et al., 1994, Stytz, 1996). These virtual environments are scalable and often use dead-reckoning algorithms to improve network performance. However, a driving simulator with multiple human controlled actors may require near or absolute synchronization. For example, when the lead driver in a car-following situation suddenly brakes, the following car driver needs to respond as quickly as possible to avoid a collision. Such driving paradigms suggest that broadcasting and dead-reckoning may be applicable only if the human controlled actors are further apart than some delta time value. Their multi-driver virtual driving simulator is an extension of the virtual environments driving simulator developed by Levine and Mourant (1995). The study compares two configurations. The first is a typical distributed virtual environment in that it will use standard networking. The second configuration utilizes cloned data acquisition. This is where the analog signals of each human controlled vehicle (gas pedal, brake pedal, and steering) are sent to every node. Since they currently have only two nodes that are located in close physical proximity, cloned data acquisition can be easily accomplished. Duplicate databases for the 3D environment and vehicles reside on each computer.