There is a desperate need to efficiently and effectively monitor and maintain the world's civil infrastructure. RoadBotics allows for the use of readily available sensing technology, in the form of smartphones and other devices, to do just that. Inexpensive and more frequent means improved streets, roads, highways, sidewalks and bridges and far lower maintenance costs.
Moving Beyond the Romans: Deep Learning and Roadway Maintenance
According the Transportation Research Group, bad roads cost each US driver over $2,000 annually in unnecessary costs with an annual US cost of well over $100B and the costs of bad globally is estimated to be 5X that amount. A central challenge to maintaining good roads – that has existed since the Romans built the Appian Way over 2,000 years ago – is regular, thorough inspection of those roads. Two millennia ago a ‘liktor’ or road inspector sat on the back of a chariot looking for and making notes of imperfections and damage on the surface and along the edges of the road. He would then share those notes with the local road crew to fix. That process was tedious, expensive, dangerous and highly subjective. Regrettably little has changed in two thousand years, save for the fact that the chariot is now a Ford150 or Toyota truck.
Visual inspection is still the most popular method of road inspection, and not just for roads but for the inspection of large infrastructure. Fortunately, rapid advances in both AI/deep learning and the availability inexpensive, yet precise, sensors is transforming infrastructure monitoring and maintenance, with the result of lower costs and far greater transparency. My presentation will focus on how this revolution in 'asset transparency' is happening now by, among other things, using the presenter’s company, RoadBotics, as a prime example of this transformation. The presenter will highlight the opportunities and challenges of deploying this technology not only through the example of his own company but other similar companies assessing other infrastructure. We use deep learning and standard smartphones to assess road surfaces and roadways. We were spun out of Carnegie Mellon Robotics Institute in 2016 and we serve over 100 cities in 16 US states and 4 countries.