Abstract
Pedestrian trajectory prediction under crowded circumstances is a challenging problem owing to human interaction and the complexity of the trajectory pattern. Various methods have been proposed for solving this problem, ranging from traditional Bayesian analysis to Social Force model and deep learning methods. However, most existing models heavily depend on specific scenarios because the trajectory model is constructed in absolute coordinates even though the motion trajectory as well as human interaction are in relative motion. In this study, a novel trajectory prediction model is proposed to capture the relative motion of pedestrians in extremely crowded scenarios. Trajectory sequences and human interaction are first represented with relative motion and then integrated to our model to predict pedestrians’ trajectories. The proposed model is based on Long Short Term Memory (LSTM) structure and consists of an encoder and a decoder which are trained by truncated back propagation. In addition, an anisotropic neighborhood setting is proposed instead of traditional neighborhood analysis. The proposed approach is validated using trajectory data acquired at an extremely crowded train station in Tokyo, Japan. The trajectory prediction experiments demonstrated that the proposed method outperforms existing methods and is stable for predictions of varying length even when the model is trained with a controlled short trajectory sequence.
Highlights
Pedestrian trajectory prediction is a challenging, open task attracting increasing attentions owing to its potential applications in multi-object tracking, human surveillance, socio-robot navigation, and autonomous driving [1,2,3,4,5]
Trajectory prediction under crowded scenarios is highly complex because it can be affected by various factors, such as trajectory pattern, human interaction, and obstacles
Our main task of interest is in pedestrian trajectory prediction using recurrent neural networks (RNNs), in particular, Long Short Term Memory (LSTM) based architectures
Summary
Pedestrian trajectory prediction is a challenging, open task attracting increasing attentions owing to its potential applications in multi-object tracking, human surveillance, socio-robot navigation, and autonomous driving [1,2,3,4,5]. LSTM can model long sequence data, unlike methods based on hand-crafted features, so that it can learn more trajectory cues, including the trajectory pattern, from past observation sequences. Social LSTM, a prediction model proposed for crowded scenarios, has attracted attention It models human interaction by pooling the latent states of all people inside a controlled neighborhood [24,27]. LSTM-based methods model human interaction using absolute coordinates by pooling current-state features, as in the case of Social LSTM. The focus is on the characteristic “relativity” of pedestrian motion, and an LSTM based data-driven architecture is proposed for trajectory prediction in extremely crowded scenarios. An LSTM based prediction model for extremely crowded scenarios is proposed that can model both motion trajectory and human interaction with relative motion. The paper is concluded with contributions and suggestions for future research
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
