Sensors on Vehicles (SENSOVO) – Proof-of-concept for Road Surface Distress Detection with Wheel Accelerations and ToF Camera Data Collected by a Fleet of Ordinary Vehicles

Abstract

This contribution presents the results of the "SENSOVO" project initiated by the Flanders Institute for Mobility (VIM), executed by the University of Antwerp (UAntwerp), the Flemish Institute for Technological Research (VITO) and the Belgian Road Research Centre (BRRC), and supported by several other parties. Both road users and road managers could benefit from massively, continuously, automatized collecting of information on road surface distress (potholes, cracking, subsidence,…) by a fleet of vehicles equipped with low-cost sensors. Road users could have immediate information on road conditions while road managers could get year-round insight on the general performance of the road network in addition to the data they obtain from annual inspections with specialized monitoring devices. The project's objective was to investigate possibilities of road surface distress detection using data collected by a fleet of vehicles. Two scenarios were considered: a large fleet of ordinary cars and trucks transmitting collected relevant sensor data already available on the CAN-bus in such vehicles; a vehicle, equipped with one or more Time-of-Flight cameras (ToF). Data on the CAN-bus were collected using either a CAN-logger that sends its data to a central server using GPRS where it is processed for road surface distress detection, or using an OBD scan-tool that sends its data using Bluetooth to a smartphone that already processes the data into detection events, forwarding only these to a central server. The performances were tested by UAntwerp and VITO. A simple computation developed by UAntwerp on the speed of all four wheels and on the vertical accelerations often allows indicating road distress. Either the CAN data-logger or the smartphone delivers the GPS location of the event. Several cars equipped with this technology sent their observations to a central database. Several ToF cameras were benchmarked by UAntwerp. Road data were collected at 40km/h and 40frames/s with ToF-cameras of brands Mesa and Fotonic. Several image processing algorithms dedicated to the identification of road distress on a flow of images were developed by UAntwerp. It has been demonstrated that several types of "unevenness" of the road surface (including potholes) can be detected. Some ToF-camera observations were added to the central database. The BRRC developed and implemented an algorithm for treatment and interpretation of the collected events in the central database using the ArcGIS software, simulating both sending out alerts to road users and providing daily "quality scores" for each road section in the network to the road manager. For this, the network was defined from a geographical map provided by the Flanders Geographical Information Agency (FGIA/AGIV). As soon as "enough" observations are made taking into account the frequency of observations in time, the defect is considered as real and will be reported. When the defect is no longer observed, it is considered as being repaired. A "score" for each road section was computed daily, from ...

Opportunistic sensing for road pavement monitoring

“…Introducing the noise and vibration equipment in cars is on the other hand less intrusive and requires less maintenance than previously developed systems based on visual inspection (Van Geem et al, 2016). When compared to smartphone-based approaches, observations are obtained with a significantly higher continuity and reliability.…”

“…An alternative approach is based on the analysis of camera images performed while driving. An example of this approach is Sensovo (Van Geem et al, 2016), which uses dashboard cameras and accelerometers placed in regular vehicles.…”

A Modular In-Vehicle C-ITS Architecture for Sensor Data Collection, Vehicular Communications and Cloud Connectivity

“…Inferring events or statistics from the road environment from vehicle-integrated sensor systems has a been topic of study by various authors. The comparison of each wheel speed, the position of the pedals, and the speed of the vehicle can be used to detect slippery road conditions in real-time [ 7 , 8 ], as well as pavement damage [ 9 ]. Adding external sensors such as the ones present in smartphones can also be combined with the vehicle’s own sensors to add redundancy and improve the accuracy of these detection mechanisms.…”

NORA: Towards Large-Scale Vehicular Analytics for Driving Environment Monitoring/Assessment

“…The most relevant studies to our paper are the prior works on pothole detection. The Pothole Patrol [15] [26] project leverages a high-pass filter to process vehicle information (such as speed, vertical and lateral acceleration) to identify potholes in signals [27]. In a similar fashion, road bumps are detected in [5] by carefully studying the vertical acceleration and other supplementary information.…”