Projects Automates Pothole Detection and Management
By Mark Dwortzan
The impact of a long and stormy winter continues to be felt on the roads. According to the Boston Globe, this year the City of Boston has already filled more than 8,800 potholes, primarily reported by drivers, including one in Cleveland Circle that sent a man to the hospital. Taking a more proactive approach could prevent vehicular damage, injuries and claims against the city while saving time and money for all concerned.
Now a vehicle mounted pothole detection system developed by Electrical Engineering seniors as part of their senior design project aims to do just that. Instead of relying on citizens to report potholes or paying crews to look for them, the system, known as AutoScan, could enable city vehicles to detect them automatically as they go about their daily routes. Coupled with tracking and scheduling software and incorporating a low-cost, embedded technology development platform called a Gizmo board, the system could provide a comprehensive and economical road repair solution.
“The low cost, achieved through the extensive use of open source solutions, made it compelling to the Gizmo community,” said Scott Hoot, president of GizmoSphere. “But the idea of how seamlessly this idea fit into the Internet of Things, made the BU project a winner. Clearly this is a project that takes close to real-time measurements in the physical world, and utilizes those measurements through the open standards available in the Internet.”
The AutoScan senior design team consists of Austen Schmidt (systems integration), Vinny DeGenova (image processing), Nandheesh Prasad (power engineering), Charlie Vincent (networking and GPS) and Stuart Minshull (Web application). The EE seniors developed their prototype under the supervision of ECE Adjunct Professor Babak Kia, who often assumed the role of prospective customer.
While there are several solutions available that can quickly measure potholes on a mobile platform, ranging from lasers to accelerometers, the EE team focused on a “time-of-flight” infrared camera that determines distance between the camera and various points in its field of view.
“Our system is basically an onboard computer that mounts to the bottom of a city vehicle, such as a bus,” said Schmidt. “As the bus goes along, it uses the infrared camera to scan the road for potholes and computes their depth, and sends the data collected on each pothole—volume, GPS coordinates, time and date—over a cellular network to a database hosted by a website. The website interprets data coming in from multiple scanners, displays it on a Google map and updates a Web-accessible road repair schedule.”
Exploiting the Gizmo board and open source software, the team has advanced a prototype of a system that promises to cost a few thousand dollars, far cheaper than alternatives that can range from $10,000 to $100,000. The only sacrifice is a bit of accuracy.
“Our system is a little less accurate than our competitors, because they focus on applications where you really need high-fidelity detection, such as airport tarmacs or bridges,” said Minshull. “We wanted a cheaper way for potholes to be detected without having to worry about tracking millimeter-line cracks in the road.”
To put AutoScan to the test, the team used cardboard boxes to create an elevated road surface with cutouts of different volumes representing potholes. Tests showed that the system accurately measured the volume of each cutout and successfully relayed collected data to the website. Next steps include conducting high-speed tests beyond the lab environment, and finding a way to protect the unit against vibration and adverse weather conditions.