Situation-Aware Pedestrian Trajectory Prediction with Spatio-Temporal Attention Model

Sirin Haddad,Meiqing Wu,He Wei,Siew Kei Lam

DOI: https://doi.org/10.3217/978-3-85125-652-9

2019-02-14

Abstract:Pedestrian trajectory prediction is essential for collision avoidance in autonomous driving and robot navigation. However, predicting a pedestrian's trajectory in crowded environments is non-trivial as it is influenced by other pedestrians' motion and static structures that are present in the scene. Such human-human and human-space interactions lead to non-linearities in the trajectories. In this paper, we present a new spatio-temporal graph based Long Short-Term Memory (LSTM) network for predicting pedestrian trajectory in crowded environments, which takes into account the interaction with static (physical objects) and dynamic (other pedestrians) elements in the scene. Our results are based on two widely-used datasets to demonstrate that the proposed method outperforms the state-of-the-art approaches in human trajectory prediction. In particular, our method leads to a reduction in Average Displacement Error (ADE) and Final Displacement Error (FDE) of up to 55% and 61% respectively over state-of-the-art approaches.

Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to predict the trajectories of pedestrians in crowded environments. Specifically, pedestrian trajectory prediction is crucial for collision avoidance in autonomous driving and robot navigation. However, predicting the trajectories of pedestrians in crowded environments is not easy because it is affected by the movements of other pedestrians and the static structures in the scene. These human - human and human - space interactions lead to the non - linearity of trajectories. Therefore, the paper proposes a new long - short - term memory (LSTM) network based on spatio - temporal graphs for predicting the trajectories of pedestrians in crowded environments. This model takes into account the interactions between static (physical objects) and dynamic (other pedestrians) elements in the scene. The main contributions of the paper include: 1. Proposing a spatio - temporal graph that explicitly captures the global interactions of all pedestrians in the scene as well as the local interactions with static objects. 2. Proposing a new spatio - temporal attention mechanism for the trajectory prediction of each pedestrian. This mechanism takes into account the local interactions between pedestrians and the interactions with objects. The experimental results on two widely - used datasets show that the proposed method outperforms the existing state - of - the - art methods in human trajectory prediction, especially achieving a significant reduction of 55% and 61% in the average displacement error (ADE) and the final displacement error (FDE) respectively.

Situation-Aware Pedestrian Trajectory Prediction with Spatio-Temporal Attention Model

Spatiotemporal Attention-Based Pedestrian Trajectory Prediction Considering Traffic-Actor Interaction

An Attention-Based Interaction-Aware Spatio-Temporal Graph Neural Network for Trajectory Prediction

CF-LSTM: Cascaded Feature-Based Long Short-Term Networks for Predicting Pedestrian Trajectory

Spatial-Temporal Graph Attention Network for Pedestrian Trajectory Prediction

Spatio-Temporal Interaction Aware and Trajectory Distribution Aware Graph Convolution Network for Pedestrian Multimodal Trajectory Prediction

PTPGC: Pedestrian trajectory prediction by graph attention network with ConvLSTM

Multi-Relational Pedestrian Trajectory Prediction in Complex Scenes.

Pedestrian Trajectory Prediction Based on Deep Convolutional LSTM Network

Holistic Spatio-Temporal Graph Attention for Trajectory Prediction in Vehicle-Pedestrian Interactions

Human Trajectory Prediction for Automatic Guided Vehicle with Recurrent Neural Network

EvoSTGAT: Evolving Spatiotemporal Graph Attention Networks for Pedestrian Trajectory Prediction

PTP-STGCN: Pedestrian Trajectory Prediction Based on a Spatio-temporal Graph Convolutional Neural Network

Multimodal Pedestrian Trajectory Prediction Based on Relative Interactive Spatial-Temporal Graph

Causal Temporal–Spatial Pedestrian Trajectory Prediction with Goal Point Estimation and Contextual Interaction

Human trajectory prediction using LSTM with Attention mechanism

Step Attention: Sequential Pedestrian Trajectory Prediction

Dynamic Pedestrian Trajectory Forecasting with LSTM-based Delaunay Triangulation

Using Appearance to Predict Pedestrian Trajectories Through Disparity-Guided Attention and Convolutional LSTM

Static-dynamic Global Graph Representation for Pedestrian Trajectory Prediction

Self-Growing Spatial Graph Network for Context-Aware Pedestrian Trajectory Prediction