Vehicle maneuver recognition and correction algorithm for road quality measurement system optimization

Abstract

Vibrations in road vehicles related to road surface damage have a number of harmful consequences for the health of the occupants and for the components of the vehicle. To mitigate these effects and support timely pavement repairs, continuous road condition monitoring is essential. Vibration-based measurement systems have gained prominence in recent years, but their accuracy can be significantly compromised by vehicle maneuvers, particularly on urban or curvy roads. Despite this, the influence of aggressive maneuvers has largely been overlooked in previous studies. In this paper, we address this gap by presenting a comprehensive investigation into the impact of abrupt maneuvers on vibration-based road quality measurement. We introduce a novel, computationally efficient soft-sensor algorithm that detects and isolates aggressive maneuvers using sensor data from existing road quality measurement systems, classifying them into four categories. This algorithm combines rule-based methods with machine learning, offering enhanced accuracy and lower computational costs compared to alternative approaches. In this way, the overall maneuver classification achieves an accuracy of 93%. By applying the introduced approach to identify and correct the influence of maneuvers, we achieved a 7% increase in accuracy of pavement quality classification in a suburban environment and a 10% increase in an urban environment. The proposed solution can be easily integrated into current vibration-based road quality measurement frameworks, enhancing their performance while maintaining scalability and low operational cost.