SR-LSTM: State Refinement for LSTM Towards Pedestrian Trajectory Prediction

Abstract

In crowd scenarios, reliable trajectory prediction of pedestrians requires insightful understanding of their social behaviors. These behaviors have been well investigated by plenty of studies, while it is hard to be fully expressed by hand-craft rules. Recent studies based on LSTM networks have shown great ability to learn social behaviors. However, many of these methods rely on previous neighboring hidden states but ignore the important current intention of the neighbors. In order to address this issue, we propose a data-driven state refinement module for LSTM network (SR-LSTM), which activates the utilization of the current intention of neighbors, and jointly and iteratively refines the current states of all participants in the crowd through a message passing mechanism. To effectively extract the social effect of neighbors, we further introduce a social-aware information selection mechanism consisting of an element-wise motion gate and a pedestrian-wise attention to select useful message from neighboring pedestrians. Experimental results on two public datasets, i.e. ETH and UCY, demonstrate the effectiveness of our proposed SR-LSTM and we achieves state-of-the-art results.

What the Constant Velocity Model Can Teach Us About Pedestrian Motion Prediction

“…In Sec. V we provided our models with a history of eight timesteps, which is common practice in the domain of pedestrian motion prediction [1], [2], [3], [4], [5], [14], [21]. However, the performance of the CVM that only uses the last two timesteps to make predictions suggests that for pedestrian motion prediction long histories are not as relevant as believed.…”

“…T HE accurate prediction of pedestrians' future motion is an essential capability for autonomous robots and vehicles to operate safely and to not endanger humans. Recently, many models based on neural networks have been proposed to address this problem [1], [2], [3], [4], [5]. Neural networks are powerful function approximators and believed to be able to take into account the pedestrians' motion histories and to learn how they interact.…”

A Location-Velocity-Temporal Attention LSTM Model for Pedestrian Trajectory Prediction

“…In this section, we give a brief review of the literature on pedestrian trajectory prediction, focusing especially on techniques that we compare with our proposed method. A large number of existing methods employ context information such as human-human interaction (e.g., [11], [16], [20], [22], [28]) and/or human-space interaction (e.g., [12], [18], [21], [24], [29], [30]). Similar to our work, there are also methods that incorporate attention mechanisms.…”

Trajectory Prediction in Heterogeneous Environment via Attended Ecology Embedding

“…Conventional methods focus on using egocentric features, e.g., previous trajectory, for trajectory prediction, which lacks considering some surrounding information, i.e., multi-agents, in traffic environments. For example, some works in trajectory prediction mainly focus on road-agents in homogeneous environments [2,4,10,11,21,30,31,34,39], which only consists of a single type of road-agent in a scene. However, in real driving environment, it is necessary to differentiate the interactions among different types of road-agents such as the difference between pedestrians and bikes or bikes and trucks.…”