Human-Inspired Anticipative Cruise Control for Enhancing Mixed Traffic Flow

Introducing the human driver’s intelligence of interpreting and anticipating preceding traffic conditions into automated driving control is crucial for improved driving experience. Existing human-like automated driving systems imitate human driver’s reactive control patterns without reflecting anticipative behaviors. These systems involve either online solving predictive optimization or offline training machine learning techniques, which are computationally expensive and impractical for current production vehicles. Additionally, these systems are less effective in traffic environments with lean vehicular connectivity. This paper proposes a practical human-inspired anticipative cruise control for connected automated vehicles (CAVs) that not only respects human drivers’ reactive driving preferences but exploits their anticipative capabilities to enhance mixed traffic flow. Two major contributions are (i) learning the reactive control objective from a segment of the human driving trajectory based on a recovery matrix; and (ii) data-enabled preceding traffic anticipation leveraging the behavioral theory of linear systems, which provides the estimated motion of immediate preceding vehicle (IPV) at a personalized anticipation horizon, considering the downstream traffic trends from a distant connected human-driven vehicle (CHV). The proposed control strategy is evaluated on a human-in-the-loop (HiL) driving simulator. Results validate its real-time implementation potential in (a) learning human-like reactive driving control; (b) improving traffic flow energy economy by 4.58%; (c) enhancing traffic efficiency by attenuating latent traffic fluctuations. Moreover, result reveals that the capability of mitigating traffic fluctuations is affected by the anticipation horizon, which can be adaptably learned from human driver’s timely guidance by virtue of their intelligence of adaptability to dynamic traffic conditions.