Driver lane change intention recognition in the connected environment

The connected environment provides information on surrounding traffic and areas beyond the visual range to improve driving behavior and avoid dangerous incidents. However, due to the novelty of the connected environment and the scarcity of connected data, current research on driver lane change intention in this field has received little attention. In this work, we designed a typical lane change scenario in the connected environment based on a driving simulator and real-time collection of multi-modal data from eye trackers, driving simulators, and a connected platform. The driver's eye movement, head rotation, vehicle movement, and the driver's maneuver parameters were analyzed, revealing a significant difference between the lane change intention and lane keep stages in the connected environment. In addition, the length of the intention time window with connected information (6.5 s) was longer than that without connected information (4 s). The bi-directional long and short-term memory network based on the attention mechanism (AT-BiLSTM) was used to establish a lane change intention model. The accuracy of the lane change intention model based on the proposed AT-BiLSTM algorithm surpassed that of existing machine learning algorithms. The recognition accuracy of the lane change intention model was 93.33% at 3 s prior to the lane change. The conclusions of this study are of great significance for the development of a side warning assist system in future connected environments. (C) 2021 Elsevier B.V. All rights reserved.