Data-Driven Objective Road-Driving Classification For Durable Design Of Vehicles

Durability development in the automotive industry depends on a clear understanding of how road surfaces and driving characteristics affect structural road loads and fatigue. Traditionally, road surface classification has been subjective (e.g., city, highway, rural), and done through driving instrumented vehicles over a small selection of roads. The variations in driving characteristics that are often consequent to the road surface quality is rarely accounted for in designing vehicle level durability tests. This makes it difficult to establish targets for the durability testing that accurately match the wide variations in real-world roads and driving across the market of vehicles. This paper presents a data-driven approach to objectively classify road surface and driving characteristics using metrics derived from existing vehicle and road response metrics like Vibration Dose Value (VDV), International Roughness Index (IRI) and statistical estimates of vehicle speed and acceleration. Data collected at the proving grounds, on gravel roads, smooth roads, city-like roads, etc., is used to identify classifiers that categorize road-driving combinations into groups correlating with structural fatigue damage. This correlation between fatigue damage and road-driving classification is validated using Wheel Force Transducer (WFT) measurements in instrumented vehicles. This method shows promise to develop structural fatigue estimates directly from telemetry data. The method provides a path to replacing subjective road classification with a vehicle-sensor and signal-based, objective classification for developing durability targets and tests. This method is also a scalable application of vehicle fleet data in uncontrolled environments, to develop an accurate understanding of real-world use of vehicles by customers.