Workshop on

Pedestrian Behaviour Prediction

at ICRA 2026, 1 June in Vienna, Austria

Venue: VIECON / Messe Wien   |   Room: Schubert 1   |   Time: 14:00-17:30

Introduction

Interactive scenarios such as factories, hospitals, shopping malls, and traffic scenes are highly socialized, which necessitates an effective and precise understanding of the behaviors of surrounding pedestrians. Nevertheless, it faces critical challenges: highly complex interactions, multi-modal dependencies, real-time requirements, uncertain pedestrian intentions, and dynamically changing contexts. This ICRA 2026 workshop will explore prediction problems in these real-world settings, with the aim of improving system safety, interaction fluency, and social acceptance. By integrating multidisciplinary approaches from machine learning, artificial intelligence, kinematics, and cognitive psychology, it will transcend conventional technical frameworks to collectively explore next-generation prediction paradigms suited for dynamic social environments.

Important Dates

  • Submission deadline: Mar 20, 2026 Mar 31, 2026 (subject to possible further extension)
  • Notification: Apr 10, 2026
  • Camera-ready deadline: Apr 20, 2026
  • Workshop day: Jun 1, 2026, 14:00 - 17:30 CEST

Deadlines are 23:59 PDT (Pacific Daylight Time, UTC-7). Please refer to this page for the latest updates in case of any changes.

Call for Papers

We welcome either original research contributions or previously published works (e.g., at ICRA 2026 or other venues) that are highly relevant and of strong technical quality, submitted via the following three tracks.

  • Extended Abstract: late-breaking results or recently published results for collaborative sharing.
  • Short Papers: original technical contributions with new ideas/results.
  • Brave New Ideas: bold, unconventional contributions, including novel problem formulations, early-stage concepts, surprising findings or negative results, and ideas that challenge existing paradigms and stimulate discussion.

To align with the ICRA workshop guidelines, previously published work is accepted only via the Extended Abstract track; the Short Papers and Brave New Ideas tracks require substantial original contributions that have not been published elsewhere. This workshop is a non-archival venue and will not have official proceedings; submissions can be subsequently or concurrently submitted to other venues.

Accepted submissions will be presented as oral talks or posters, and we will present a Best Paper Award and a Best Academic Poster Award to recognise outstanding contributions. All the accepted submissions will be available on our workshop website, though authors could indicate explicitly if they want to opt out.

Topics of Interest

welcome submissions on (but not limited to) the following topics:(click to view)

  • Dynamic Interaction Modeling
    • Pedestrian-vehicle interactions (PVI)
    • Pedestrian-robot interactions (PRI)
    • Pedestrian-pedestrian social interactions, group behavior, and crowd dynamics
  • Context, Perception & Representations
    • Scene context and traffic semantics (traffic lights, crosswalks, lane markings, road signs, HD maps, lane graphs)
    • Human-centric cues (gaze, head pose, body keypoints / pose)
    • Multi-modal spatial representations (2D images / egocentric views, BEV, 3D point clouds / meshes)
  • Multimodality, Uncertainty & Stochastic Prediction
    • Multimodal forecasting (multiple plausible futures)
    • Uncertainty-aware / probabilistic prediction (aleatoric & epistemic)
    • Stochastic behavior modeling and calibrated confidence
  • Methods & Interdisciplinary Approaches
    • Deep learning, probabilistic, and hybrid frameworks
    • Causal modeling and counterfactual reasoning
    • Interdisciplinary methods (psychology, cognitive science, neuroscience)
  • Systems, Datasets & Safety-Critical Applications
    • Real-time / efficient prediction systems and deployment
    • Datasets, benchmarks, evaluation protocols, and metrics
    • Safety-critical applications in autonomous driving and robotics

Author Guidelines

  • Template: Manuscripts should follow the ICRA 2026 paper template. Direct links to ieeeconf.zip and IEEEtranBST.zip.
  • Length: Extended Abstract submissions should be concise, with a maximum length of 2 pages, including text, figures, and references. Short Papers and Brave New Ideas papers are limited to 4 pages (including references), with up to 2 extra pages allowed (i.e., up to 6 pages total, including references).
  • Review: All submissions will be reviewed by the workshop program committee. The review process will be single-blind. As an author, you do not need to anonymize your submission. At least one co-author of each submission must volunteer to review for the workshop. Submissions with no co-author reviewer participation may be desk-rejected.
  • Presentation: If accepted, at least one author must register for in-person participation and present the paper on-site at the workshop. The program committee may retract accepted papers that are not presented in person.
  • Submission: Please submit your manuscript via the CMT system; see the CMT author submission form guide for details.
Submit Now!

The Microsoft CMT service was used for managing the peer-reviewing process for this conference. This service was provided for free by Microsoft and they bore all expenses, including costs for Azure cloud services as well as for software development and support.


Presentation Instructions

    There will be two types of presentations at the workshop: oral presentations and poster presentations. You can find your assigned presentation type in your notification email.
  • Oral presentations: Each oral presentation will be 5 minutes in total, including Q&A. Please prepare your slides in 16:9 format and send them to this email address by May 31, 2026.
  • Please note that authors selected for oral presentation are also required to prepare a poster. Please follow the poster instructions below.
  • Poster presentations: Posters should be prepared in A0 format, portrait orientation. Posters will be displayed on the walls of the workshop room, and the venue will provide stickers for attaching them.

Invited Speakers





Workshop Schedule

14:00--14:10 Welcome and Opening Remarks
14:10--14:45 Invited Talk by Prof. Hubert P. H. Shum
Understanding Pedestrian Behaviour: Trajectory Prediction from Human-Human to Human-Vehicle Interactions
Talk Cancelled Invited Talk by Prof. Philippe Martinet
Proactive Autonomous Navigation in Human Populated Environment 📄 Slides
14:45--15:20 Invited Talk by Dr. Amir Rasouli
Trustworthy Pedestrian Prediction
15:20--15:40 Coffee Break & Poster Session
15:40--16:10 Invited Talk by Prof. Alexandre Alahi
Pedestrian Behavior Prediction
16:10--16:20 Sponsor Remarks
16:20--17:10 Paper Presentation
17:10--17:30 Concluding Remarks

Accepted List

Short Papers

HuNavSim 2.5: Human Navigation Simulator for evaluation of Human-Aware Robot Navigation
Noé Pérez-Higueras, Miguel Escudero, Fernando Caballero, and Luis Merino

Simulation and Prediction of Human Flows in Hospital Environments from Assistive Robot Observations
Maxime Bernard, Damien Rambeau, Jacques Saraydaryan, and Olivier Simonin

Vehicle Collision Detection Model Based on Multiple Traffic Factors
Xin Cheng, Jingmei Zhou, Lizhuoyi Mi, Chongfeng Wei, and Can Cheng

Extended Abstracts

Dynamic Clustering for Efficient Trajectory Prediction in Dense Crowd Environments
Antonius Bima Murti Wijaya, Paul Henderson, and Marwa Mahmoud

ESIA: Energy-Based Spatiotemporal Interaction-Aware Framework for Pedestrian Intention Prediction
Yanping Wu, Meiting Dang, Lin Wu, Edmond S. L. Ho, Zhenghua Chen, and Chongfeng Wei

Depth-Lift: Edge-Deployable Pedestrian Trajectory Prediction via Lightweight Monocular Depth Estimation
Li Zhang, Yi Zhang, Xinyu Wang, Zhihao Lin, Xiang Luo, and Fengmin Cheng

DADA: Dual-Alignment Domain Adaptation for Cross-Scene Pedestrian Trajectory Prediction
Wenzhan Li, Fuhao Li, Xinghui Jing, Pingfa Feng, and Long Zeng

AFMCT: Adaptive Cross-modal Transformer Fusion for Stable 3D Pedestrian Perception Toward Robust Behavior Prediction
Yixin Wang, Bingli Zhang, Chengbiao Zhang, Xinyu Wang, Junzhao Jiang, Yifan Wang, and Shuqing Zhao

FogFusion: robust camera-LiDAR fusion for fog-resilient 3D perception in pedestrian-vehicle interaction scenarios
Yifan Wang, Bingli Zhang, Yixin Wang, Chengbiao Zhang, Xingyu Wang, Junzhao Jiang, Xiang Luo, and Shuqing Zhao

Organizers

Contact