Mutual understanding between driver and vehicle is critically important to the design of intelligent vehicles and customized interaction interface. In this study, a unified driver behavior reasoning system toward multi-scale and multi-tasks behavior recognition is proposed. Specifically, a multi-scale driver behavior recognition system is designed to recognize both the driver's physical and mental states based on a deep encoder-decoder framework. This system can jointly recognize three driver behaviors with different time scales based on the shared encoder network. Driver body postures and mental behaviors include intention and emotion are studied and identified. The encoder network is designed based on a deep convolutional neural network (CNN), and several decoders for different driver states estimation are proposed with fully connected (FC) and long short-term memory (LSTM) based recurrent neural networks (RNN). The joint feature learning with the CNN encoder increases the computational efficiency and feature diversity, while the customized decoders enable an efficient multi-tasks inference. The proposed framework can be used as a solution to exploit the relationship between different driver states, and it is found that when drivers generate lane change intentions, their emotions usually keep neutral state and more focus on the task. Two naturalistic datasets are used to investigate the model performance, which is a local highway dataset, namely, CranData and one public dataset from Brain4Cars. The testing results on these two datasets show accurate performance and outperform existing methods on driver postures, intention, and emotion recognition.
Drivers have unique and rich driving behaviors when operating vehicles in traffic. This paper presents a novel driver behavior learning approach that captures the uniqueness and richness of human driver behavior in realistic driving scenarios. A stochastic inverse reinforcement learning (SIRL) approach is proposed to learn a distribution of cost function, which represents the richness of the human driver behavior with a given set of driver-specific demonstrations. Evaluations are conducted on the realistic driving data collected from the 3D driver-in-the-loop driving simulation. The results sh…
Georgia Institute of Technology（佐治亚理工学院）；Bilkent University（比尔肯特大学）
This paper proposes a method for modeling human driver interactions that relies on multi-output gaussian processes. The proposed method is developed as a refinement of the game theoretical hierarchical reasoning approach called "level-k reasoning" which conventionally assigns discrete levels of behaviors to agents. Although it is shown to be an effective modeling tool, the level-k reasoning approach may pose undesired constraints for predicting human decision making due to a limited number (usually 2 or 3) of driver policies it extracts. The proposed approach is put forward to fill this gap in…
Human drivers have limited and time-varying cognitive resources when making decisions in real-world traffic scenarios, which often leads to unique and stochastic behaviors that can not be explained by perfect rationality assumption, a widely accepted premise in modeling driving behaviors that presume drivers rationally make decisions to maximize their own rewards under all circumstances. To explicitly address this disadvantage, this study presents a novel driver behavior model that aims to capture the resource rationality and stochasticity of the human driver's behaviors in realistic longitudi…
Institute of Industrial Science, The University of Tokyo（东京大学生产技术研究所）
For the optimum design of a driver-automation shared control system, an understanding of driver behavior based on measurements and modeling is crucial early in the development process. This paper presents a driver model through a weighting process of visual guidance from the road ahead and haptic guidance from a steering system for a lane-following task. The proposed weighting process describes the interaction of a driver with the haptic guidance steering and the driver reliance on it. A driving simulator experiment is conducted to identify the model parameters for driving manually and with ha…
Several factors influence traffic congestion and overall traffic dynamics. Simulation modeling has been utilized to understand the traffic performance parameters during traffic congestions. This paper focuses on driver behavior of route selection by differentiating three distinguishable decisions, which are shortest distance routing, shortest time routing and less crowded road routing. This research generated 864 different scenarios to capture various traffic dynamics under collective driving behavior of route selection. Factors such as vehicle arrival rate, behaviors at system boundary and tr…
National Institute of Information and Communications Technology (NICT)（日本信息通信研究机构）
Driver reaction is of vital importance in risk scenarios. Drivers can take correct evasive maneuver at proper cushion time to avoid the potential traffic crashes, but this reaction process is highly experience-dependent and requires various levels of driving skills. To improve driving safety and avoid the traffic accidents, it is necessary to provide all road drivers with on-board driving assistance. This study explores the plausibility of case-based reasoning (CBR) as the inference paradigm underlying the choice of personalized crash evasive maneuvers and the cushion time, by leveraging the w…
Corpy & Co., Inc.（Corpy公司）；The University of Tokyo（东京大学）
Driver distraction is a well-known cause for traffic collisions worldwide. Studies have indicated that shared steering control, which actively provides haptic guidance torque on the steering wheel, effectively improves the performance of distracted drivers. Recently, adaptive shared steering control based on the physiological status of the driver has been developed, although its effect on distracted driver behavior remains unclear. To this end, a high-fidelity driving simulator experiment was conducted involving 18 participants performing double lane changes. The experimental conditions compri…
The thesis presents contributions made to the evaluation and design of a haptic guidance system on improving driving performance in cases of normal and degraded visual information, which are based on behavior experiments, modeling and numerical simulations. The effect of shared control on driver behavior in cases of normal and degraded visual information has been successfully evaluated experimentally and numerically. The evaluation results indicate that the proposed haptic guidance system is capable of providing reliable haptic information, and is effective on improving lane following performa…
Development and Evaluation of Two Learning-Based Personalized Driver Models for Car-Following Behaviors
Beijing Institute of Technology（北京理工大学）；University of California Berkeley（加州大学伯克利分校）；University of Michigan Transportation Research Institute（密歇根大学交通研究所）
The work is devoted to ways of modeling street traffic on a street layout without traffic lights of an established topology. The behavior of traffic participants takes into account the individual inclinations of drivers to creatively interpret traffic rules. Participant interactions describe game theory models that provide information for simulation algorithms based on cellular automata. Driver diversification comes down to two types often considered in such research: DE(fective)-agent and CO(operative)-agent. Various ways of using the description of traffic participants to examine the impact …
We propose a novel framework to differentiate between vehicle trajectories originating from human and non-human drivers by constructing a data-driven boundary using parametric signal temporal logic (STL). Such construction allows us to evaluate the trajectories, detect rare-events, and reduce the uncertainty of driver behaviors when it assumes the form of a disturbance in control synthesis and evaluation problems. We train a classifier that separates admissible (i.e. human) examples - which arise from real-world demonstrations - and inadmissible (i.e. non-human) examples that are generated by …
Drivers' heterogeneity and the broad range of vehicle characteristics on public roads are primarily responsible for the stochasticity observed in road traffic dynamics. Understanding the behavioural differences in drivers (human or automated systems) and reproducing observed behaviours in microsimulation attracts significant attention lately. Calibration of car-following model parameters is the prevalent way to simulate different driving behaviors through randomly injected variation around average parameter values. An issue is that, as shown in the literature, most car-following model do not r…
Higher School of Communications of Tunis, University of Carthage（迦太基大学突尼斯高等通信学院）；CEMSE Division, King Abdullah University of Science and Technology (KAUST)（阿卜杜拉国王科技大学（KAUST）计算工程与科学学院（CEMSE Division））
Driver behavior profiling is one of the main issues in the insurance industries and fleet management, thus being able to classify the driver behavior with low-cost mobile applications remains in the spotlight of autonomous driving. However, using mobile sensors may face the challenge of security, privacy, and trust issues. To overcome those challenges, we propose to collect data sensors using Carla Simulator available in smartphones (Accelerometer, Gyroscope, GPS) in order to classify the driver behavior using speed, acceleration, direction, the 3-axis rotation angles (Yaw, Pitch, Roll) taking…
Institute of Industrial Science, The University of Tokyo（东京大学工业科学研究所）；Hino Motors, Ltd.（日野汽车公司）
Lane change is a very demanding driving task and number of traffic accidents are induced by mistaken maneuvers. An automated lane change system has the potential to reduce driver workload and to improve driving safety. One challenge is how to improve driver acceptance on the automated system. From the viewpoint of human factors, an automated system with different styles would improve user acceptance as the drivers can adapt the style to different driving situations. This paper proposes a method to design different lane change styles in automated driving by analysis and modeling of truck driver…
Naturalistic driving data were applied to study driver acceleration behaviour, and a probability model of the driver was proposed. First, the question of whether the database is large enough is resolved using kernel density estimation and Kullback-Liebler divergence. Next, the convergence database is utilised to achieve the bivariate acceleration distribution pattern. Subsequently, two probability models are proposed to explain the pattern. Finally, the statistical characteristics of the acceleration behaviours are studied to verify the probability models. The longitudinal and lateral accelera…
Drivers have distinctively diverse behaviors when operating vehicles in natural traffic flow, such as preferred pedal position, car-following distance, preview time headway, etc. These highly personalized behavioral variations are known to impact vehicle fuel economy qualitatively. Nevertheless, the quantitative relationship between driving behaviors and vehicle fuel consumption remains obscure. Addressing this critical missing link will contribute to the improvement of transportation sustainability, as well as understanding drivers' behavioral diversity. This study proposed an integrated micr…
SEIDM: A Safe and Efficient Intelligent Driver Model for Autonomous Driving Behavior
Department of Information and Communication Engineering, Tongji University（信息与通信工程系，同济大学）；Department of Electrical and Computer Engineering, Duke University（电气与计算机工程系，杜克大学）
This paper presents a personalized adaptive cruise control (PACC) design that can learn driver behavior and adaptively control the semi-autonomous vehicle (SAV) in the car-following scenario, and investigates its impacts on mixed traffic. In mixed traffic where the SAV and human-driven vehicles share the road, the SAV's driver can choose a PACC tuning that better fits the driver's preferred driving behaviors. The individual driver's preferences are learned through the inverse reinforcement learning (IRL) approach by recovering a unique cost function from the driver's demonstrated driving data …
We derive time and energy-optimal control policies for a Connected Autonomous Vehicle (CAV) to complete lane change maneuvers in mixed traffic. The interaction between CAVs and Human-Driven Vehicles (HDVs) requires designing the best possible response of a CAV to actions by its neighboring HDVs. This interaction is formulated using a bilevel optimization setting with an appropriate behavioral model for an HDV's. Then, an iterated best response (IBR) method is used to determine a Nash equilibrium. However, we also show that when a common and simple-to-detect condition applies, the optimal lane-…
The focus of this paper is to propose a driver model that incorporates human reasoning levels as actions during interactions with other drivers. Different from earlier work using game theoretical human reasoning levels, we propose a dynamic approach, where the actions are the levels themselves, instead of conventional driving actions such as accelerating or braking. This results in a dynamic behavior, where the agent adapts to its environment by exploiting different behavior models as available moves to choose from, depending on the requirements of the traffic situation. The bounded rationalit…
