Fault Diagnostics for Intelligent Vehicle Applications
Rajesh Rajamani, Ankur Shrivastava, Chunyu Zhu, Lee Alexander
Report no. Mn/DOT 2001-20
Topics: Intelligent Vehicles
This project involved the development of a fault diagnostic system for Safetruck, an intelligent vehicle prototype. The fault diagnostic system continuously monitors the health of vehicle sensors, detects a failure when it happens, and identifies the source of the failure. The fault diagnostic system monitors several key components: the Global Positioning System, lateral accelerometer, and yaw-rate gyroscope, which constitute the set of lateral dynamic sensors, as well as the forward-looking radar that measures distance, relative velocity, and azimuth angle to other vehicles and objects on the highway. To design the project's lateral fault diagnostic system, researchers exploited the model-based dynamic relationships that exist between the three lateral sensors. They verified the system's performance through extensive experiments on the Safetruck. This project also explored a number of new approaches to creating a reliable fault detection system for radar. Monitoring the radar's health poses a special challenge because the radar measures the distance to another independent vehicle on the highway. In the absence of inter-vehicle communications, the fault diagnostic system has no way of knowing the other vehicle's motion, which means that model-based approaches cannot be used. Experimental results indicate that an inexpensive redundant sensor combined with a specially designed nonlinear filter would provide the most reliable method for radar health monitoring.