Development of Automated Driveability Rating System

Abstract

<div class="section abstract"><div class="htmlview paragraph">Trained human raters have been used by organizations such as the Coordinating Research Council (CRC) to assess the vehicle driveability performance effect of fuel volatility. CRC conducts workshops to test fuel effects and their impact on vehicle driveability. CRC commissioned Southwest Research Institute (SwRI) to develop a “Trick Car” vehicle that could trigger malfunctions on-demand that mimic driveability events. This vehicle has been used to train novice personnel on the CRC Driveability Procedure E-28-94. While largely effective, even well-trained human raters can be inconsistent with other raters. Further, CRC rater workshop programs used to train and calibrate raters are infrequent, and there are a limited number of available trained raters. The goal of this program was to augment or substitute human raters with an electronic driveability sensing system. The Automated Driveability Rating System (ADRS) was developed for Light Duty (LD) vehicles and can identify and rate fuel-related driveability events including hesitation, stumble, surge, stall, and idle quality at trace, moderate, and heavy severities. The portable system uses sensors such as accelerometers, and interfaces with a vehicle to gather and process an array of information. Overall, ADRS performance ranged from somewhat less accurate to significantly better than trained human raters depending on the event type and severity. For light and moderate vehicle throttle tests, detection of stumble, surge, and hesitation events by the ADRS was close to or better than 90%, while idle quality accuracy was 80%. These results are better when compared to the performance of trained raters. Additional effort in refining the calibration and improving event identification could enhance performance even further, and the system could be applied more broadly in rating ride quality and vehicle behavior.</div></div>