Learning Probabilistic Responsibility Allocations for Multi-Agent Interactions

arXiv cs.RO / 4/16/2026

💬 OpinionIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper introduces a probabilistic model that learns how responsibility is allocated among agents in multi-agent interactions, reflecting how actors deviate from their desired behavior to satisfy shared constraints like safety.
It uses a conditional variational autoencoder latent space combined with multi-agent trajectory forecasting to represent multimodal uncertainty in responsibility allocations conditioned on scene and agent context.
Because direct responsibility labels are not available, the method stays trainable by using a differentiable optimization layer that converts responsibility allocations into induced control actions.
Experiments on the INTERACTION driving dataset show strong predictive performance and provide interpretable responsibility-based insights into interaction patterns.

Abstract

Human behavior in interactive settings is shaped not only by individual objectives but also by shared constraints with others, such as safety. Understanding how people allocate responsibility, i.e., how much one deviates from their desired policy to accommodate others, can inform the design of socially compliant and trustworthy autonomous systems. In this work, we introduce a method for learning a probabilistic responsibility allocation model that captures the multimodal uncertainty inherent in multi-agent interactions. Specifically, our approach leverages the latent space of a conditional variational autoencoder, combined with techniques from multi-agent trajectory forecasting, to learn a distribution over responsibility allocations conditioned on scene and agent context. Although ground-truth responsibility labels are unavailable, the model remains tractable by incorporating a differentiable optimization layer that maps responsibility allocations to induced controls, which are available. We evaluate our method on the INTERACTION driving dataset and demonstrate that it not only achieves strong predictive performance but also provides interpretable insights, through the lens of responsibility, into patterns of multi-agent interaction.