Dipnet: Driver Intention Prediction for a Safe Takeover Transition in Automated Vehicles

Processing of complex traffic scenes for effective steering and collision avoidance: a perspective, from research into human control, on the challenges for sensor-based autonomous vehicles on urban roads

“…In the latter case it is the most vulnerable road users such as pedestrians and cyclists that are most likely to be placed at risk by AVs with no “theory of road users.” Table 1 presents a categorization of some of the judgments that humans may make about other road users and the features or behaviors that may support those judgment, with that caveat that those assumptions may sometimes be erroneous. There have been neural network systems developed that attempt to predict driver behavior or intent, but a number of these are using vehicle kinematics to estimate trajectories ( Bonyani et al, 2015 ; Girma et al, 2020 ), but not yet attempting the broader process of detecting potential behavior of pedestrians and cyclists, which remains a significant challenge ( Gilpin, 2021 ). What would be valuable would be the identification of the most reliable dynamic features of road user behavior that can lead to computational models arriving at robust estimates of emergent behavior (e.g., Markkula et al (2023) ).…”

Driver intention prediction based on multi-dimensional cross-modality information interaction

“…Based on a report by the World Health Organization (WHO), fatal injuries resulting from road traffic accidents account for approximately 1.35 million deaths worldwide annually, with non-fatal incidents excluded [1]. Driver misconduct, which includes dangerous driving behaviors such as illegal lane changes or turns, speeding, and fatigue driving, has been found to be the primary cause of most traffic accidents according to research in the field of road safety [2].…”

“…In recent years, advanced driver assistance systems (ADAS) have been widely valued as a means to improve driving safety and prevent car accidents. Driver intention prediction, a key component of such systems, enables drivers to rapidly detect potential hazards and develop a range of solutions to enhance road safety [1,3,4].…”

“…Deep learning has recently gained extensive development and application in various fields. Due to the principle of using a large number of neurons to simulate human perception, thinking, and other activities, researchers have employed deep learning to address driver intention prediction, with promising outcomes [1,10,[16][17][18]. Generally, deep learning-based driver intention prediction approaches offer advantages such as automatic feature learning, end-to-end learning, and more comprehensive and superior performance of the learned features.…”

“…Generally, deep learning-based driver intention prediction approaches offer advantages such as automatic feature learning, end-to-end learning, and more comprehensive and superior performance of the learned features. However, most studies utilize 3D Conv [10,16,19], optical flow [1], or stacking of LSTM [3,[16][17][18]20] to model the temporal information of video sequences, leading to issues of large network parameters and high algorithm deployment costs. Moreover, it is noteworthy that most studies neglect or do not effectively utilize GPS information.…”