This product began with TEM Software providing PennDOT with a web application that plans road treatments in response to weather events. A centralized database stores the average speed for every road segment within a Department of Transportation district. Reductions in the average speed signify “events” requiring a response. During snow storms, actions include snow plowing and applying salt.
The application correlates real-time micro cell weather information with weather forecasts in an effort to prescribe road treatments to avoid icing. Our solution effectively processes structured live data streams from traffic sensors, remote weather stations, and service vehicles with unstructured data from voice-to-text communication among emergency vehicles and phone calls and texts from the public.
The centralized database exploits a multi-tenant architecture containerizing information local to each district. The product is delivered as a platform as a service on a monthly payment schedule.
Our solution includes a maintenance management system that gives administrators the details they need to forecast budgets, predict resource usage, and evaluate performance on standardized key performance metrics. Immediately upon executing this solution, a department of transportation can earn cost savings, provide more effective road treatment, improve safety, and attain more consistent operations.
Internal to our product is a “learning engine” associating current information with previous efforts that were scored for their process effectiveness. A threshold is established to send inappropriate information to alternate processes or human involvement.
Product learning is sensitive to contradictions and ambiguities in weather reports and traffic speeds in adjoining road segments. Users are provided with a manual human interface to adjudicate conflicts.
Our product handles a manifold of live streaming data from sensors. Rather than collecting all raw data over time, our data input engine stores only changes in the stream. Our solution scores data quality based on multiple criteria including source reliability, verification among sensors, and heuristic models.