Sim-to-Real Causal Transfer: A Metric Learning Approach to Causally-Aware Interaction Representations

TL;DR

This paper introduces a metric learning approach to enhance causally-aware interaction representations in multi-agent systems, improving robustness and generalization from simulation to real-world scenarios.

Contribution

It proposes a novel causal regularization method and a sim-to-real transfer technique that together improve causal understanding and robustness in multi-agent interaction models.

Findings

Enhanced causal awareness in latent representations.

Improved out-of-distribution robustness.

Significant generalization gains in pedestrian datasets.

Abstract

Modeling spatial-temporal interactions among neighboring agents is at the heart of multi-agent problems such as motion forecasting and crowd navigation. Despite notable progress, it remains unclear to which extent modern representations can capture the causal relationships behind agent interactions. In this work, we take an in-depth look at the causal awareness of these representations, from computational formalism to real-world practice. First, we cast doubt on the notion of non-causal robustness studied in the recent CausalAgents benchmark. We show that recent representations are already partially resilient to perturbations of non-causal agents, and yet modeling indirect causal effects involving mediator agents remains challenging. To address this challenge, we introduce a metric learning approach that regularizes latent representations with causal annotations. Our controlled experiments show that this approach not only leads to higher degrees of causal awareness but also yields stronger out-of-distribution robustness. To further operationalize it in practice, we propose a sim-to-real causal transfer method via cross-domain multi-task learning. Experiments on pedestrian datasets show that our method can substantially boost generalization, even in the absence of real-world causal annotations. We hope our work provides a new perspective on the challenges and pathways towards causally-aware representations of multi-agent interactions. Our code is available at https://github.com/vita-epfl/CausalSim2Real.