Centralized Model Predictive CACC Control Robust to Burst Communication Errors

摘要

A centralized controller computes controls and transmits them to automated vehicles. In case of a communication failure, control values will be missing and vehicles might take incorrect driving decisions. Although modern controllers may resist spurious communication errors, burst errors are expected to be more challenging. This paper introduces a centralized controller robust to burst errors. We propose a model predictive controller (MPC) computing intended future controls and sharing them with vehicles to be stored in a buffer. Through such buffer, vehicles may still retrieve control values even under burst errors. The proposed concept is evaluated in a mixed multi vehicle braking scenario and its superiority over other approaches in terms of distance and acceleration errors compared is shown.