<ul data-eligibleForWebStory="true"><li>Modeling spatial-temporal interactions among neighboring agents is crucial for multi-agent problems like motion forecasting and crowd navigation.</li><li>Recent representations may not fully capture the causal relationships in agent interactions.</li><li>A metric learning approach is introduced to enhance causal awareness in representations by regularizing latent features with causal annotations.</li><li>Experiments demonstrate that the proposed approach improves causal awareness and enhances out-of-distribution robustness.</li><li>A sim-to-real causal transfer method through cross-domain multi-task learning is proposed to apply these concepts in real-world scenarios.</li><li>Experiments on pedestrian datasets show significant performance improvements even without real-world causal annotations.</li><li>The research offers insights into challenges and solutions for developing causally-aware representations of multi-agent interactions.</li><li>The code for the approach is available at https://github.com/vita-epfl/CausalSim2Real.</li></ul>

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

Discover more