On-the-Fly Path Replanning Using Real-Time Road Condition Data
Abstract
The fast advancements in intelligent transportation call for quick and responsive solutions as roads become unpredictable. It proposes RoadAware-DIRL, which applies decentralised edge sensing, real-time data merging, and IRL to create a more effective framework for passing information between edge computing devices, thereby improving vehicular navigation. Static maps and re-updated traffic feeds are the mainstays of traditional routing, making it hard for them to deal with quick changes on the roads. With the help of decentralised sensors and an effective noise filter, RoadAware-DIRL fills this gap by keeping a constant track of the road and making sure the data is reliable. This Decision Engine behaves like a human does by using information from stressful navigation events from the past. In addition, a Predictive Obstruction Forecaster depends on recurring road changes to predict where trouble will occur and steer traffic in another direction. Hazardous routes are given higher penalties in a Contextual Cost Map, and MPC keeps making split-second adjustments to the planned path. Extra support for communication is available thanks to V2V and V2I layers, which set up a network of moving road scouts. This framework is better than existing ones since it can adapt flexibly, make decisions locally, and learn like a person. Improved performance, swifter handling of hazards, and less travel disruption support the usage of the system in both autonomous and semi-autonomous cars. The platform makes today’s urban roadways more resilient and aware of the situations they encounter.