Application of hierarchical Bayesian estimation to calibrating a car-following model with time-varying parameters

Kasai, Makoto; Shibagaki, Shun; Terabe, Shintaro

doi:10.1109/ivs.2013.6629576

Cited by 7 publications

(5 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…CFMs have been extensively calibrated using a maximum likelihood approach (Hoogendoorn and Hoogendoorn, 2010), Bayesian estimation (van Hinsbergen et al, 2009;Kasai et al, 2013;Rahman et al, 2015;Lee and Ozbay, 2009a;Davis, 2017), fundamental diagram regression (Qu et al, 2015;Phegley et al, 2014), or heuristics (Lee and Ozbay, 2009b;Ma and Abdulhai, 2002). Most of them are calibrated using a pair of leading and following vehicle trajectories.…”

Section: Physics-based Model Parameter Calibrationmentioning

confidence: 99%

A Survey on Autonomous Vehicle Control in the Era of Mixed-Autonomy: From Physics-Based to AI-Guided Driving Policy Learning

Shi

2020

Preprint

View full text Add to dashboard Cite

This paper serves as an introduction and overview of the potentially useful models and methodologies from artificial intelligence (AI) into the field of transportation engineering for autonomous vehicle (AV) control in the era of mixed autonomy. It is the first-of-its-kind survey paper to comprehensively review literature in both transportation engineering and AI for mixed traffic modeling. We will discuss state-ofthe-art applications of AI-guided methods, identify opportunities and obstacles, raise open questions, and help suggest the building blocks and areas where AI could play a role in mixed autonomy. We divide the stage of autonomous vehicle (AV) deployment into four phases: the pure HVs, the HV-dominated, the AV-dominated, and the pure AVs. This paper is primarily focused on the latter three phases. Models used for each phase are summarized, encompassing game theory, deep (reinforcement) learning, and imitation learning. While reviewing the methodologies, we primarily focus on the following research questions: (1) What scalable driving policies are to control a large number of AVs in mixed traffic comprised of human drivers and uncontrollable AVs? (2) How do we estimate human driver behaviors? (3) How should the driving behavior of uncontrollable AVs be modeled in the environment? (4) How are the interactions between human drivers and autonomous vehicles characterized? Hopefully this paper will not only inspire our transportation community to rethink the conventional models that are developed in the data-shortage era, but also reach out to other disciplines, in particular robotics and machine learning, to join forces towards creating a safe and efficient mixed traffic ecosystem.

show abstract

Section: Physics-based Model Parameter Calibrationmentioning

confidence: 99%

A Survey on Autonomous Vehicle Control in the Era of Mixed-Autonomy: From Physics-Based to AI-Guided Driving Policy Learning

Shi

2020

Preprint

View full text Add to dashboard Cite

show abstract

“…[4] quantified the extent of heterogeneity by analyzing a trajectory data collected in real-world traffic, which turned out that different drivers react differently to the stimuli from a leading car, [5] extended the work by relating such observed heterogeneity to vehicle types in the composition of follower-leader pairs, [8] calibrated a full set of random coefficients that account for the heterogeneity across drivers, and [9] proposed a stochastic framework that takes into account both individual and general driving characteristics as one aggregate model. Intra-driver heterogeneity has been studied as well, where [10] divided the procedure in different regimes and modeled acceleration control at each particular situation, [11] incorporates stochastic Markov regime switching model to address the driving features at different regimes, [12] developed a hierarchical Bayesian model with time-varying parameters to account for the gradual changes in car-following behaviors, [13] use dynamic time warping (DTW) to calibrate time-varying response times and critical jam spacing of a car following model, which are further used to analyze the intra-driver heterogeneity and situation-dependent behavior within a trip, and [14] segmented a continuous stream into clusters by evaluating similarities on driving features. In addition, heterogeneity related with roadway categories has also been reported [15].…”

Section: Literature Review a Heterogeneity In Car-following Behaviorsmentioning

confidence: 99%

“…II, we can find that there is roughly an accuracy peak region in where the average accuracy is higher than surrounding (M, Q) combinations, and accuracy tends to decrease when (M, Q) go far from the region. Similarly, in testing results, there is also such a region: roughly the upper triangle part of M ∈ [2,4], Q ∈ [12,28].…”

Section: B 3-driver Experiment-model and Algorithm Visualizationmentioning

confidence: 99%

Driver Identification through Stochastic Multi-State Car-Following Modeling

Xu¹,

Ding²,

Tu³

et al. 2020

Preprint

View full text Add to dashboard Cite

Intra-driver and inter-driver heterogeneity has been confirmed to exist in human driving behaviors by many studies. In this study, a joint model of the two types of heterogeneity in car-following behavior is proposed as an approach of driver profiling and identification. It is assumed that all drivers share a pool of driver states; under each state a car-following data sequence obeys a specific probability distribution in feature space; each driver has his/her own probability distribution over the states, called driver profile, which characterize the intradriver heterogeneity, while the difference between the driver profile of different drivers depict the inter-driver heterogeneity. Thus, the driver profile can be used to distinguish a driver from others. Based on the assumption, a stochastic car-following model is proposed to take both intra-driver and inter-driver heterogeneity into consideration, and a method is proposed to jointly learn parameters in behavioral feature extractor, driver states and driver profiles. Experiments demonstrate the performance of the proposed method in driver identification on naturalistic car-following data: accuracy of 82.3% is achieved in an 8-driver experiment using 10 car-following sequences of duration 15 seconds for online inference. The potential of fast registration of new drivers are demonstrated and discussed.This work is partially supported by Groupe PSA's OpenLab program (Multimodal Perception and Reasoning for Intelligent Vehicles) and the NSFC Grants 61973004. D.Xu and Z.Ding are both the first authors of this paper.

show abstract

“…To do this, autonomous vehicle receives messages from the ones nearby, and make an active decision to increase or decrease its acceleration to maintain the traffic optimally based on the method presented in the previous work [9]. In detail, the very front car transmits an information of a speed with the position to surrounding cars.…”

Section: B Optimal Speed Calculation Of the Vehicle For Pilot Drivingmentioning

confidence: 99%

“…After certain information are acquired from the surrounding area, an average velocity is calculated to get an optimal acceleration speed to maintain the most efficient traffic. In theory, it is most efficient once the distances between the vehicles are identical [9]. Thus, the vehicle controls its acceleration to be aligned distance with surrounding vehicles.…”

Section: B Optimal Speed Calculation Of the Vehicle For Pilot Drivingmentioning

confidence: 99%

A 3-D real-time simulation for autonomous driving with V2V communications

Lee

Janghee

Kim

2013

2013 International Conference on Connected Vehicles and Expo (ICCVE)

View full text Add to dashboard Cite

We propose a real-time traffic simulation for autonomous vehicles with V2V communications. In order to execute a simulation, imitated generic traffic signals are developed as a component of the simulator, which is similar as ones in the real world. In addition, autonomous vehicle driving module supported by V2V communication is integrated into the simulator. Then, we generate robust numbers of the vehicle in the simulation software to conduct experiments. The results can be shown in real-time 3D rendering with ongoing status of surrounding vehicles.

show abstract

Application of hierarchical Bayesian estimation to calibrating a car-following model with time-varying parameters

Cited by 7 publications

References 9 publications

A Survey on Autonomous Vehicle Control in the Era of Mixed-Autonomy: From Physics-Based to AI-Guided Driving Policy Learning

A Survey on Autonomous Vehicle Control in the Era of Mixed-Autonomy: From Physics-Based to AI-Guided Driving Policy Learning

Driver Identification through Stochastic Multi-State Car-Following Modeling

A 3-D real-time simulation for autonomous driving with V2V communications

Contact Info

Product

Resources

About