Dr. Cody Stolle
Date of this Version
Sweigard, M.E., DEVELOPMENT OF A NOVEL VEHICLE GUIDANCE SYSTEM: VEHICLE RISK MITIGATION AND CONTROL, Thesis, University of Nebraska, April 2020
Over a half of fatal vehicular crashes occur due to vehicles leaving their designated travel lane and entering other lanes or leaving the roadway. Lane departure accidents also result in billions of dollars in cost to society. Recent vehicle technology research into driver assistance and vehicle autonomy has developed to assume various driving tasks. However, these systems are do not work for all roads and travel conditions. The purpose of this research study was to begin the development a novel vehicle guidance approach, specifically studying how the vehicle interacts with the system to detect departures and control the vehicle
A literature review was conducted, covering topics such as vehicle sensors, control methods, environment recognition, driver assistance methods, vehicle autonomy methods, communication, positioning, and regulations. Researchers identified environment independence, recognition accuracy, computational load, and industry collaboration as areas of need in intelligent transportation.
A novel method of vehicle guidance was conceptualized known as the MwRSF Smart Barrier. The vision of this method is to send verified road path data, based AASHTO design and vehicle dynamic aspects, to guide the vehicle.
To further development research was done to determine various aspects of vehicle dynamics and trajectory trends can be used to predict departures and control the vehicle. Tire-to-road friction capacity and roll stability were identified as traits that can be prevented with future road path knowledge. Road departure characteristics were mathematically developed. It was shown that lateral departure, orientation error, and curvature error are parametrically linked, and discussion was given for these metrics as the basis for of departure prediction.
A three parallel PID controller for modulating vehicle steering inputs to a virtual vehicle to remain on the path was developed. The controller was informed by a matrix of XY road coordinates, road curvature and future road curvature and was able to keep the simulated vehicle to within 1 in of the centerline target path. Recommendations were made for the creation of warning modules, threshold levels, improvements to be applied to vehicle controller, and ultimately full-scale testing.
Advisor: Cody S. Stolle