WS13: The 5th Workshop on Naturalistic Driving Data Analytics

Workshop Code: p67bw

Time: 13:30-17:30


  • Huijing Zhao
    Affiliation: the Key Lab of Machine Perception (MOE), Peking University, Beijing, China.


  • Donghao Xu
    Affiliation: the Key Lab of Machine Perception (MOE), Peking University, Beijing, China.

Scope and Goals

This workshop is organized by Naturalistic Driving Data Analytics Technical Committee of IEEE ITS Society. It will be co-located at the 2018 IEEE Intelligent Vehicles Symposium (IV) which will be held in ChangShu, China, June 26 - July 1, 2018. Four previous editions of this workshop have also been co-located at the IEEE IV series (2014-2017).

Large datasets of naturalistic driving data have been collected, and diverse type of analysis could be done using such data, and an immense potential is still to be explored. The purposes of this workshop are two folds: 1) to foster discussions on challenges related to developing and using the right methods to make meaningful inferences and interpretations from large scale unlabeled naturalistic driving data to among others further develop ITS technologies (e.g., intelligent vehicles, advanced driver assistance systems) and extend knowledge on crash causation and normal driving behaviors; and 2) to exchange ideas on how to optimally use naturalistic driving data especially in the era of automation.

The workshop will be for 8 hours and consist of 9-12 speakers, coffee and lunch breaks, and a poster session for more deep discussion. The oral session consists of three parts, i.e., Data Collection, Data Analytics, and NDD Applications, and each session includes one or two invited speakers.

Data collection part is focusing on how to collect, maintain and share the huge amount of ND data. Because recent intelligent vehicles including self-driving vehicles have a lot of sensors to investigate drivers behavior and surrounding road environment, the collected data amount exceeds more than 500GB for one-hour driving even in the case with one camera and several LIDARs. Thus considering how to maintain and share such a huge amount of data is one of the essential problems in ND data analytics area as well as how to collect data.

Data analytics part is focusing on the more theoretical area than data collection part. In this part, we will discuss how can we establish more efficient methods to deal the huge amount of ND data. In addition to the traditional eventbased analytics, the several machine learning approaches, e.g., non-parametric Bayesian and deep neural networks, have been well studied especially in recent days. By using these methods, we can introduce data-driven knowledge statistically extracting not only from some parts of the ND data around the events but also from the whole parts of the data with integrating multimodal information such as driving behavior signals, images and point clouds of surrounding road environments.

In the NDD application parts, we will discuss potential applications of ND data and try to clarify how ND data contributes to improving ITS systems as well as self-driving vehicles. We will also invite some speakers from industries and startups to discuss based on actual examples.

Topics of Interest

  • Data Collection
    – Naturalistic driving data sharing
    – Robust data compression and annotation
    – Feature reduction
  • Data Analytics
    – Driver behavior analysis
    – Image processing for ND videos
    – Machine learning for ND data
  • NDD Applications
    – Infrastructure/roadway features extraction
    – ND data for automation


  • 13:30-14:10
    On Annotation, Augmentation & Analytics of Naturalistic Data for Autonomous Driving
    Mohan Trivedi, University of California, San Diego

  • 14:10-14:50
    Vehicle-Pedestrian Encountering Scenarios and Pedestrian Behavior Recognition in Naturalistic Driving Environment
    Renran Tian, Indiana University-Purdue University Indianapolis
    Yaobin Chen, Indiana University-Purdue University Indianapolis
    David Good, School of Public and Environmental Affairs, Indiana University - Bloomington

  • 14:50-15:10
    Coffee break with demo & posters

  • 15:10-15:50
    Automan: Making research corpus from real-world autonomous driving data
    Shunya Seiya, Nagoya University
    Yuto Jumonji, The University of Tokyo
    Hidenaga Ushijima, The University of Tokyo
    Abraham Monrroy, Nagoya University
    Yuki Tsuji, The University of Tokyo
    Kazumasa Sakiyama, The University of Tokyo
    Eiji Sekiya, Tier4
    Yuki Iida, Tier4
    Shinpei Kato, The University of Tokyo; Tier4
    Kazuya Takeda, Nagoya University; Tier4

  • 15:50-16:30
    Scene-Aware Driving Behavior Modeling
    Donghao Xu, Peking University
    Huijing Zhao, Peking University

  • 16:30-17:30
    Round Table Discussion: How to effectively collect/share/utilize the Naturalistic Driving Data – Challenges & future of NDS activities
    Pujitha Gunaratne, Toyota Motor North America