Mitigation Strategy for Navigation Errors in Strict Route Plans

Localization errors can be caused by unintentional malfunctions or intentional attacks on localization processes and sensory devices. These errors can occur at critical driving situations and at high speeds, resulting in false navigation for Autonomous Vehicles (AV). Those situations can be at a road fork that is branching to an adjacent parallel road or to a different level of a multi-level road. The AV can merge to the wrong road and continue following the original route plan until the correct current location is realized. This can lead to hazardous driving situations or deviations from a strict plan. This work considers special cases where the AV must adhere to a strictly specified route, i.e. large vehicles that must be driven on highways. An algorithm that we call Road-Class Aware Rerouting (RCAR) is developed to identify the point of deviation within a directed graph representing a road map and find the optimal way to return to the original route or reach the destination while maintaining road network constraints. Simulated examples are included to illustrate the proposed algorithm.