Accounting for localization errors in a mixed-vehicle centralized control system

Future autonomous vehicles are expected to coordinate controls to improve trac eciency, fuel consumption and most importantly safety. A centralized vehicle control system is a system where a centralized entity computes control inputs considering variables like vehicles' localization, velocity, etc. and transmits it to the vehicles to implement them. Due to the presence of localization errors, perceived localization of a vehicle is dierent from its true localization. Control inputs calculated using erroneous localization, when implemented on vehicles result into collisions. In this paper we propose a centralized control approach which takes into consideration localization errors for collision free braking in a mixed trac scenario. The performance of the proposed approach is evaluated under three dierent trac scenarios with vehicles having erroneous localization. Results show significant increase in the number of collisions avoided, while using proposed approach, as compared to scenarios where localization errors were not accounted for. Moreover, despite localization errors, the proposed approach can ensure upto the same number of collision avoidance as in a scenario where true localization was known.