Qualitative Prediction of Multi-Agent Spatial Interactions

TL;DR

This paper introduces and benchmarks three novel approaches for modeling and predicting multi-agent spatial interactions in dense scenes, utilizing qualitative representations and attention mechanisms, with a focus on robot and human scenarios.

Contribution

It presents three new methods combining qualitative trajectory calculus with neural networks for improved multi-agent interaction prediction in complex environments.

Findings

The data-driven approach outperforms the relation-based models in experiments.

All methods demonstrate generalization to different scenarios.

Qualitative and attention-based models effectively capture spatial interactions.

Abstract

Deploying service robots in our daily life, whether in restaurants, warehouses or hospitals, calls for the need to reason on the interactions happening in dense and dynamic scenes. In this paper, we present and benchmark three new approaches to model and predict multi-agent interactions in dense scenes, including the use of an intuitive qualitative representation. The proposed solutions take into account static and dynamic context to predict individual interactions. They exploit an input- and a temporal-attention mechanism, and are tested on medium and long-term time horizons. The first two approaches integrate different relations from the so-called Qualitative Trajectory Calculus (QTC) within a state-of-the-art deep neural network to create a symbol-driven neural architecture for predicting spatial interactions. The third approach implements a purely data-driven network for motion prediction, the output of which is post-processed to predict QTC spatial interactions. Experimental results on a popular robot dataset of challenging crowded scenarios show that the purely data-driven prediction approach generally outperforms the other two. The three approaches were further evaluated on a different but related human scenarios to assess their generalisation capability.