Interpretable Social Anchors for Human Trajectory Forecasting in Crowds

TL;DR

This paper introduces an interpretable neural network architecture for human trajectory forecasting in crowds, combining discrete choice models for rule-based intent learning with neural networks for scene-specific residuals, achieving accurate and explainable predictions.

Contribution

The paper presents a novel hybrid model that integrates discrete choice models with neural networks to produce interpretable and accurate human trajectory forecasts in crowded scenes.

Findings

Effective explanation of predictions without accuracy loss

Outperforms existing methods on TrajNet++ benchmark

Demonstrates the value of interpretability in trajectory forecasting

Abstract

Human trajectory forecasting in crowds, at its core, is a sequence prediction problem with specific challenges of capturing inter-sequence dependencies (social interactions) and consequently predicting socially-compliant multimodal distributions. In recent years, neural network-based methods have been shown to outperform hand-crafted methods on distance-based metrics. However, these data-driven methods still suffer from one crucial limitation: lack of interpretability. To overcome this limitation, we leverage the power of discrete choice models to learn interpretable rule-based intents, and subsequently utilise the expressibility of neural networks to model scene-specific residual. Extensive experimentation on the interaction-centric benchmark TrajNet++ demonstrates the effectiveness of our proposed architecture to explain its predictions without compromising the accuracy.