EAGER: Core Capabilities for Smart Cars

Carnegie Mellon University, Pittsburgh PA

Investigators

Abstract

The fact that 6 vehicles completed a recent Urban Challenge competition, autonomous driving on city streets, clearly demonstrated that autonomous vehicles that can be built to deal with the complexities of driving. The technologies showcased, impressive as they were, need to be significantly enhanced to work dependably in the real world. Vehicles controlled by ever-vigilant cyber-physical systems can lead to significant declines in accidents and resulting deaths/injuries; compensate for human error, distraction, or reduced reaction rates; reduce hours spent in traffic congestion and improve traffic flow; and thereby increase productivity and independence for many segments of the population. However, these platforms and technologies must be deemed robust before widespread adoption can even begin to take place. This EAGER research grant enables a direct and timely interaction between the Carnegie Mellon University research team and the U.S. automotive industry to explore challenges that must be overcome before autonomous and highly-automated vehicles can be safely deployed in real-world driving situations. Challenges include (a) cyber-physical interactions inside the automobile such as ensuring the timely, safe and reliable operations of sensors, actuators, computers, communication buses/networks, electrical power, and cabling interfaces, necessitating a dependable, real-time cyber-physical platform, (b) real-time communications to/from the automobile, using vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) networks to coordinate behaviors with other vehicles when necessary and to obtain/react to information about traffic or accident conditions, traffic lights, and signs, (c) safe behaviors with built-in recovery and safety-enhancement mechanisms to deal with unforeseen road conditions. The research pursues contributions in sensor fusion with dynamic object detection/recognition, safe dynamic behaviors under ever-changing operating conditions, vehicular networks, joint cyber-physical system modeling, analysis/simulation tools, dynamic map-building, end-to-end resource management, and safety-critical real-time fault-tolerant distributed cyber-physical platforms.