Pedestrian Trajectory Prediction

Follow this topic Ask about this topic

The field predicts pedestrians’ future paths to help autonomous vehicles and robots navigate safely around people. Methods range from recurrent networks and Transformers to state-space models, often using social interaction graphs, scene maps, gaze, or intent cues; probabilistic approaches such as CVAEs and flow matching represent multiple possible futures. Evaluation commonly uses displacement errors on ETH/UCY and SDD, with newer work also testing driving datasets, dense crowds, and robot-collected data. A central challenge is balancing accurate forecasts with calibrated uncertainty, social and physical safety, robustness to occlusion and domain shift, and manageable computation.

Latest papers

36 papers6 papers in the last four weeks, level with the four weeks before. 0.0% of all new papers.

TIGFlow-GRPO: Trajectory Forecasting via Interaction-Aware Flow Matching and Reward-Guided Optimization
Mar 26, 2026Xuepeng Jing, Wenhuan Lu, Hao Meng +2Trajectory Prediction Pedestrian Trajectory Prediction
Tianjin University, Tianjin, China
Eye Gaze-Informed and Context-Aware Pedestrian Trajectory Prediction in Shared Spaces with Automated Shuttles: A Virtual Reality Study
Mar 20, 2026Danya Li, Yan Feng, Rico KruegerPedestrian Trajectory Prediction Pedestrian Behavior Modeling
Department of Technology, Management and Economics at the Technical University of Denmark. · Department of Transport & Planning, Civil Engineering Geosciences at Delft University of Technology.
Efficient Dense Crowd Trajectory Prediction Via Dynamic Clustering
Mar 18, 2026Antonius Bima Murti Wijaya, Paul Henderson, Marwa MahmoudPedestrian Trajectory Prediction Clustering
University of Glasgow, Glasgow G12 8QQ, United Kingdom
TrajMamba: An Ego-Motion-Guided Mamba Model for Pedestrian Trajectory Prediction from an Egocentric Perspective
Mar 16, 2026Yusheng Peng, Gaofeng Zhang, Liping ZhengPedestrian Trajectory Prediction Trajectory Prediction
School of Computer Science and Information Engineering, Hefei University of Technology, Hefei 230601, China · School of Software, Hefei University of Technology, Hefei 230601, China
Rethinking Gaussian Trajectory Predictors: Calibrated Uncertainty for Safe Planning
Mar 11, 2026Fatemeh Cheraghi Pouria, Mahsa Golchoubian, Katherine Driggs-CampbellPedestrian Trajectory Prediction Safe Motion Planning
Department of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Champaign, IL 61820 USA · Department of Mathematical and Computational Sciences, University of Toronto, Toronto, Canada
From GEV to ResLogit: Spatially Correlated Discrete Choice Models for Pedestrian Movement Prediction
Mar 1, 2026Rulla Al-Haideri, Bilal FarooqDiscrete Choice Modeling Pedestrian Trajectory Prediction
Laboratory of Innovations in Transportation (LiTrans), Toronto Metropolitan University, 350 Victoria Street, Toronto, Ontario, Canada

Mar 26, 2026J/K move · Enter open · S save

TIGFlow-GRPO: Trajectory Forecasting via Interaction-Aware Flow Matching and Reward-Guided Optimization

Xuepeng Jing, Wenhuan Lu, Hao Meng, Zhizhi Yu+1

Tianjin University, Tianjin, China

Trajectory Prediction Pedestrian Trajectory Prediction Flow-Based Generative Modeling Conditional Flow Matching

Open paper AskPDF arXiv

Human trajectory forecasting is important for intelligent multimedia systems operating in visually complex environments, such as autonomous driving and crowd surveillance. Although Conditional Flow Matching (CFM) has shown strong ability in modeling trajectory distributions from spatio-temporal observations, existing approaches still focus primarily on supervised fitting, which may leave social norms and scene constraints insufficiently reflected in generated trajectories. To address this issue, we propose TIGFlow-GRPO, a two-stage generative approach that aligns flow-based trajectory generation with behavioral rules. In the first stage, we build a CFM-based predictor with a Trajectory-Interaction-Graph (TIG) module to model fine-grained visual-spatial interactions and strengthen context encoding. This stage captures both agent-agent and agent-scene relations more effectively, providing more informative conditional features for subsequent alignment. In the second stage, we perform Flow-GRPO post-training, where deterministic flow rollout is reformulated as stochastic ODE-to-SDE sampling to enable trajectory exploration, and a composite reward combines view-aware social compliance with map-aware physical feasibility. By evaluating trajectories explored through SDE rollout, GRPO progressively steers multimodal predictions toward behaviorally plausible futures. Experiments on the ETH/UCY and SDD datasets show that TIGFlow-GRPOimproves forecasting accuracy and long-horizon stability while generatingtrajectories that are more socially compliant and physically feasible.These results suggest that the proposed approach provides an effective way to connectflow-based trajectory modeling with behavior-aware alignment in dynamic multimedia environments.

Latest papers

TIGFlow-GRPO: Trajectory Forecasting via Interaction-Aware Flow Matching and Reward-Guided Optimization

Eye Gaze-Informed and Context-Aware Pedestrian Trajectory Prediction in Shared Spaces with Automated Shuttles: A Virtual Reality Study

Efficient Dense Crowd Trajectory Prediction Via Dynamic Clustering

TrajMamba: An Ego-Motion-Guided Mamba Model for Pedestrian Trajectory Prediction from an Egocentric Perspective

Rethinking Gaussian Trajectory Predictors: Calibrated Uncertainty for Safe Planning

From GEV to ResLogit: Spatially Correlated Discrete Choice Models for Pedestrian Movement Prediction