Surrogate Safety Measures Prediction at Multiple Timescales in V2P Conflicts Based on Gated Recurrent Unit

Improving pedestrian safety at urban intersections requires intelligent systems that should not only understand the actual vehicle–pedestrian (V2P) interaction state but also proactively anticipate the event’s future severity pattern. This paper presents a Gated Recurrent Unit-based system that aims to predict, up to 3 s ahead in time, the severity level of V2P encounters, depending on the current scene representation drawn from on-board radars’ data. A car-driving simulator experiment has been designed to collect sequential mobility features on a cohort of 65 licensed university students who faced different V2P conflicts on a planned urban route. To accurately describe the pedestrian safety condition during the encounter process, a combination of surrogate safety indicators, namely TAdv (Time Advantage) and T2 (Nearness of the Encroachment), are considered for modeling. Due to the nature of these indicators, multiple recurrent neural networks are trained to separately predict T2 continuous values and TAdv categories. Afterwards, their predictions are exploited to label serious conflict interactions. As a comparison, an additional Gated Recurrent Unit (GRU) neural network is developed to directly predict the severity level of inner-city encounters. The latter neural model reaches the best performance on the test set, scoring a recall value of 0.899. Based on selected threshold values, the presented models can be used to label pedestrians near accident events and to enhance existing intelligent driving systems.

[1]  Li Li,et al.  Using LSTM and GRU neural network methods for traffic flow prediction , 2016, 2016 31st Youth Academic Annual Conference of Chinese Association of Automation (YAC).

[2]  Omar Y. Al-Jarrah,et al.  Deep Learning-based Vehicle Behaviour Prediction For Autonomous Driving Applications: A Review , 2019, ArXiv.

[3]  Zahra Karevan,et al.  Transductive LSTM for time-series prediction: An application to weather forecasting , 2020, Neural Networks.

[4]  Jian Cao,et al.  Financial time series forecasting model based on CEEMDAN and LSTM , 2019, Physica A: Statistical Mechanics and its Applications.

[5]  Florent Altché,et al.  An LSTM network for highway trajectory prediction , 2017, 2017 IEEE 20th International Conference on Intelligent Transportation Systems (ITSC).

[6]  Demetris Koutsoyiannis,et al.  Comparison of stochastic and machine learning methods for multi-step ahead forecasting of hydrological processes , 2019, Stochastic Environmental Research and Risk Assessment.

[7]  F. Bella,et al.  VEHICLE–PEDESTRIAN INTERACTIONS INTO AND OUTSIDE OF CROSSWALKS: EFFECTS OF DRIVER ASSISTANCE SYSTEMS , 2021, Transport.

[8]  A Várhelyi,et al.  Drivers' speed behaviour at a zebra crossing: a case study. , 1998, Accident; analysis and prevention.

[9]  Michael Bauer,et al.  Real-Time Driver Maneuver Prediction Using LSTM , 2020, IEEE Transactions on Intelligent Vehicles.

[10]  Niki Martinel,et al.  Weakly-Supervised Domain Adaptation of Deep Regression Trackers via Reinforced Knowledge Distillation , 2021, IEEE Robotics and Automation Letters.

[11]  C. Micheloni,et al.  Tracking-by-Trackers with a Distilled and Reinforced Model , 2020, ACCV.

[12]  Qian Lei,et al.  Research on Longitudinal Active Collision Avoidance of Autonomous Emergency Braking Pedestrian System (AEB-P) , 2019, Sensors.

[13]  Keping Li,et al.  Evaluation of pedestrian safety at intersections: A theoretical framework based on pedestrian-vehicle interaction patterns. , 2016, Accident; analysis and prevention.

[14]  Bo Cheng,et al.  Drivers’ Braking Behaviors in Different Motion Patterns of Vehicle-Bicycle Conflicts , 2019, Journal of Advanced Transportation.

[15]  Yoshua Bengio,et al.  Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling , 2014, ArXiv.

[16]  Haneen Farah,et al.  Are collision and crossing course surrogate safety indicators transferable? A probability based approach using extreme value theory. , 2020, Accident; analysis and prevention.

[17]  Guo Zhongyin,et al.  Driving Simulator Validity of Driving Behavior in Work Zones , 2020 .

[18]  Franz Kummert,et al.  Behavior prediction at multiple time-scales in inner-city scenarios , 2011, 2011 IEEE Intelligent Vehicles Symposium (IV).

[19]  Zhi Huang,et al.  LSTM based trajectory prediction model for cyclist utilizing multiple interactions with environment , 2021, Pattern Recognit..

[20]  Ross B. Girshick,et al.  Mask R-CNN , 2017, 1703.06870.

[21]  Zhaoxin Li,et al.  STGAT: Modeling Spatial-Temporal Interactions for Human Trajectory Prediction , 2019, 2019 IEEE/CVF International Conference on Computer Vision (ICCV).

[22]  Pongsathorn Raksincharoensak,et al.  Shared Control in Risk Predictive Braking Maneuver for Preventing Collisions With Pedestrians , 2016, IEEE Transactions on Intelligent Vehicles.

[23]  Qing Xu,et al.  Modified Driving Safety Field Based on Trajectory Prediction Model for Pedestrian–Vehicle Collision , 2019, Sustainability.

[24]  Silvio Savarese,et al.  Social LSTM: Human Trajectory Prediction in Crowded Spaces , 2016, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[25]  Ankit Kathuria,et al.  Evaluating pedestrian vehicle interaction dynamics at un-signalized intersections: A proactive approach for safety analysis. , 2019, Accident; analysis and prevention.

[26]  Karim Ismail,et al.  Traffic conflict techniques for road safety analysis: open questions and some insights , 2014 .

[27]  C. Hydén,et al.  Evaluation of traffic safety, based on micro-level behavioural data: theoretical framework and first implementation. , 2010, Accident; analysis and prevention.

[28]  Mohamed Abdel-Aty,et al.  Modeling pedestrians' near-accident events at signalized intersections using gated recurrent unit (GRU). , 2020, Accident; analysis and prevention.

[29]  Francesco Bella,et al.  Effects on Driver’s Behavior of Illegal Pedestrian Crossings , 2018 .

[30]  Kip Smith,et al.  Pedestrian injury mitigation by autonomous braking. , 2010, Accident; analysis and prevention.

[31]  Zachary Chase Lipton.  A Critical Review of Recurrent Neural Networks for Sequence Learning , 2015, ArXiv.

[32]  N. Baldo,et al.  Drivers’ Braking Behavior Affected by Cognitive Distractions: An Experimental Investigation with a Virtual Car Simulator , 2020, Behavioral sciences.

[33]  Tarek Sayed,et al.  Bivariate extreme value modeling for road safety estimation. , 2018, Accident; analysis and prevention.

[34]  Kaiming He,et al.  Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[35]  P. Krause,et al.  COMPARISON OF DIFFERENT EFFICIENCY CRITERIA FOR HYDROLOGICAL MODEL ASSESSMENT , 2005 .

[36]  John P. Wann,et al.  Perceiving Time to Collision Activates the Sensorimotor Cortex , 2005, Current Biology.

[37]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[38]  Jose Manuel Barrios,et al.  Assessing the benefit of the brake assist system for pedestrian injury mitigation through real-world accident investigations , 2013 .

[39]  Rico Auerswald,et al.  The Principles of Operation Framework: A Comprehensive Classification Concept for Automated Driving Functions , 2020 .

[40]  Nitesh V. Chawla,et al.  SMOTE: Synthetic Minority Over-sampling Technique , 2002, J. Artif. Intell. Res..

[41]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.