Search Results

Spoken language technologies, including speech recognition, natural language understanding, dialogue management, language generation and speech synthesis, can be important elements of driver interaction systems. They appear to offer the promise of a natural, hands-free, eyes-on-the-road way to interact with systems on and off the vehicle, especially systems that are rich in features and flexibility. This paper begins with a brief history of spoken language technologies in vehicles, examines some current applications and describes the special challenges of using speech technologies in the vehicle environment. Various approaches to meeting these challenges are examined. The paper describes the current state of spoken dialogue systems and provides some examples of their potential use in vehicles. The paper concludes with a description of the likely evolution of spoken language systems in vehicles.

TravTek is a joint public sector-private sector project to develop, test and evaluate an integrated advanced driver information system and supporting infrastructure. TravTek will provide drivers of 100 specially-equipped 1992 Oldsmobile Toronados with navigation, real-time traffic information, route guidance, and motorist information services. The system begins operations in Orlando, Florida in January 1992.

Over the next 20 years, advanced driver information systems for North America will evolve through three stages: (1) the information stage, (2) the advisory stage, and (3) the coordination stage. For each stage, this paper describes the goals, vehicle equipment and infrastructure, expected benefits, research and development required, deployment issues, obstacles and constraints. A series of operational field tests are defined for evaluating the systems at each stage. The paper concludes with an outline of standards which will be required. This information was developed at a series of workshops sponsored by Mobility 2000 under the leadership of the Federal Highway Administration.

The existence of the computer coupled with the rapid development of analytical engineering techniques has made it possible to design complex systems which are different and better, and to do so in a manner that is more efficient. Examples from the aircraft industry include hypersonic airplanes in which the fuel serves as a coolant to heat sensitive frame components, and the frame serves as part of the engine. Thus, design changes to any single part have cascading effects on the entire Bystem. This leads to extreme complexity in the design process and requires extensive computational resources and sophisticated computer techniques to proceed in a rational manner. For example, large scale systems theory and multi-level optimization are some of the approaches applicable to this class of design problems.

The axially stratified-charge (ASC) concept was demonstrated in a compact production car by modifying the engine and developing the required control system and calibration. Two production 1.8 L four-cylinder engines were modified to operate as ASC engines by adding shrouded inlet valves to produce swirl, and by providing timed-sequential port fuel injection. One of these engines was calibrated for minimum fuel consumption in the laboratory using a computer-controlled engine and dynamometer. The second engine was installed in a vehicle equipped with an oxidizing catalyst. A complete control system was developed for this engine to implement the minimum fuel consumption calibration in the vehicle. The fuel economy of the ASC vehicle was six percent better than that of the base vehicle. It had acceptable driveability, and had a 91 Research octane requirement on the fuel.

THIS PAPER describes the use of sensitivity functions to investigate the sensitivity of engine control systems to unknown variations in parameters. The definition of output sensitivity functions and the derivation of their defining differential equations are reviewed. A structural method for generating a sensitivity model is presented and applied to a second order example. Finally, the sensitivity model for EGR valve gain of an eighteen state linear perturbation engine model is generated and the resulting sensitivity functions used to predict how engine response changes for changes in the gain of the EGR valve. These predictions are then shown to be accurate by comparison with simulations of the engine model for various values of the EGR valve gain.