Transportation Research Board – AED50 Data Challenge


Transportation Forecasting Competition (TRANSFOR 22) – Call for Participation – 2022 TRB Annual Meeting Workshop

Call for Participation – 2022 TRB Annual Meeting Workshop

Transportation Forecasting Competition (TRANSFOR 22)

Organized by: AED50 Artificial Intelligence and Advanced Computing Applications

Supported by: IEEE ITSS Technical Activities Sub-Committee “Smart Mobility and Transportation 5.0”

Sponsored by: The Center for Urban Informatics and Progress (CUIP) at The University of Tennessee at Chattanooga (UTC), NSF, City of Chattanooga, Ouster LiDAR, and Seoul Robotics.

Pedestrian wheelchair users are 36% more likely to be killed in a road accident than the general public. While this is devastating, it is more devastating knowing that there is no data on the pedestrians’ categories in both fatal and nonfatal accidents, since police reports often do not keep a record of whether a victim was using a wheelchair or has a disability. This leads to lack of a clear understanding of the risk factors that make the traffic environment more dangerous for this pedestrian category compared to others.

Pedestrians, cyclists, wheelchair users and other vulnerable road users (VRUs) currently rely on pre-programmed signal timing and pedestrian phases to safely cross a signal-controlled intersection. Nonetheless, VRUs would benefit from more adequate and proactive protection that recognizes, differentiates and distinguishes a walking pedestrian from a handicapped one. In essence, real-time detection of these VRUs using advanced traffic sensors installed at the infrastructure side would have great potential to significantly improve traffic safety at intersections.

In support of the aforementioned technology, the Transportation Research Board is happy to announce the 2022 Transportation Forecasting (TRANSFOR) Competition, which will be held at the Transportation Research Board (TRB) 101st Annual Meeting, for the third time since 2013. This year’s competition is also supported by the IEEE ITSS Technical Activities Sub-Committee “Smart Mobility and Transportation 5.0” and sponsored by The Center for Urban Informatics and Progress (CUIP) at The University of Tennessee at Chattanooga (UTC), NSF, City of Chattanooga, Ouster LiDAR, and Seoul Robotics.

The scope of this TRANSFOR 22 competition is to 1) evaluate the accuracy of sub-classifying pedestrians at higher-risk; 2) predict the time needed to cross the street and whether the pedestrian can safely cross within the allocated pedestrian signal time. Sub-classifications are up to the entrants, but some examples may include wheelchair users, elderly citizens, individuals pushing strollers, etc. As an example, a scenario such as a pedestrian crossing the road but cannot reach their destination based on the position and speed during the current timing cycle should be flagged. The competition is open to teams (single participant or group of participants) that are willing to present their work at the 2022 TRB Annual Meeting.

Teams will have access to robust LiDAR data labeled with pedestrian, vehicle, and bike, and subclassification features of object size and velocity, as well as other related features including Signal Phasing and Timing (SPaT). These data are graciously provided by the CUIP at UTC, Ouster, and Seoul Robotics who are searching for a deeper understanding of these microscopic traffic moments, in hopes that this understanding leads to solutions that make all cities more accessible.

The competition will run in three stages: in the first stage the interested teams can register by sending their information (contestant's name and email address; number of member(s) in the team, their names, and email addresses; affiliation; student or professional; highest degree obtained; and name of the team) to the Organizing Committee ( This will be followed by the second stage (pre-selection), in which data will be made available through a link on our website (links and access information will be provided to registrants) and the submissions will be comprehensively evaluated and ranked according to the quality, feasibility, implementation, and potential value. The top three submissions will proceed to the third stage, which will take place on during the TRB Annual Meeting as part of the AED50 Committee activities, where the selected teams will briefly present their work (10 minutes for each presentation). The Organizing Committee will select the winner based on four criteria: i. Classification accuracy (weight = 35%), ii. Novelty of solutions (weight = 35%) and iii. Quality of the code (weight = 15%), and iv. Quality of the presentation (weight = 15%). All finalists will receive monetary prizes as follows: the first prize is $3,000, the second prize $2,000, and the third prize $1,000, which will be announced during the annual AED50 Committee Meeting in January 2022.

The Dataset

TRNSFOR 22 contestants will be using data collected by Ouster digital LiDAR sensors located on the MLK Smart Corridor in downtown Chattanooga, Tennessee. This competition focuses on one intersection (MLK and Georgia Ave) that is equipped with three Ouster OS1-128 LiDAR sensors, as shown in the figure below. Ouster’s LiDAR data will then be processed by Seoul Robotics software. This output data will include LiDAR point clouds with labeled objects, including pedestrians, vehicles, and bicyclists, as well as subclassification features such as object size and velocity.

Challenge Image

Problem Formulation

Contestants are asked to explore advanced approaches for detecting high-risk pedestrians attempting to cross the street at signalized intersections, and predict the time needed for every subclass of VRUs to cross the street, which can be used to adjust the signal pedestrian phases. Often, these populations are severely disadvantaged at crosswalks and intersections because they may have difficulty completing the crossing within the pre-programmed pedestrian phases. The proposed research could aid cities and municipalities in developing data-driven solutions that bring us closer to Vision Zero goals and create a safer traffic environment for all road users.

Panel Members

Sherif Ishak, Old Dominion University (, Eleni I. Vlahogianni, National Technical University of Athens (, Osama A. Osman, University of Tennessee (, Mohamed Zaki, University of Central Florida (, Mecit Cetin, Old Dominion University (, Yaobin Chen, IEEE ITS Society’s VP for Technical Activities (, David Reinke (, William Muller, Seoul Robotics (, Mina Sartipi, Center for Urban Informatics and Progress (

Competition Rules:

Teams should build solutions from scratch using the provided dataset. Third party OPEN-SOURCE tools and frameworks are allowed as well as your normal tooling. You may also incorporate pre-existing material that is freely available to the public into your project, such as public domain images, Creative Commons music, open-source libraries, existing APIs and platforms, and the like. However, the core development of the code MUST be the original development of the team.

  • At least one member of the team must register for the 2022 TRB meeting. Proof of TRB registration must be submitted via email along with the data competition registration.
  • Following the registration deadline of November 14th, the data will be provided to all registered teams.
  • The final submission must include source codes, paper, and slides.
  • Teams must agree to share their code on the committee’s GitHub page. The code will be archived and open to the public through an open project that will be created in GitHub by the Committee.


If you have questions, please email

Important Dates:

  • November 14, 2021: Deadline for team registration.
  • November 15, 2021: Data released to the registered teams.
  • December 15, 2021: Deadline for submission of results.
  • December 20, 2021: Notification of acceptance of top 3 submissions (Pre-selection stage).
  • January 9-13 2022 TRB Annual Meeting (Final stage for nominating top three winners).

Intellectual Property

An Entrant submission provided the Submission components are solely the Entrant’s work product and the result of the Entrant’s ideas and creativity, and the submission must be open-sourced. An Entrant may submit a Submission that includes the use of open source software or hardware, provided the Entrant complies with applicable open source licenses and, as part of the Submission, creates software that enhances and builds upon the features and functionality included in the underlying open source product. By entering the Competition, you represent, warrant, and agree that your Submission meets these requirements.


Plagiarism is the use of information or concepts from another article, website, or report without clearly attributing the source. Plagiarism is not acceptable. Phrases, sentences, or sections taken from another document, even if written by the same author(s), must appear within quotation marks and the source must be credited.


Please upload your solutions as a single zip folder. Be sure to include your organization and team id in the folder name.