SceneMotion: From Agent-Centric Embeddings to Scene-Wide Forecasts

TL;DR

SceneMotion is an attention-based model that transforms local agent-centric embeddings into scene-wide motion forecasts, enabling joint interaction modeling and improving prediction accuracy for autonomous driving scenarios.

Contribution

The paper introduces a novel latent context module that learns a scene-wide latent space from agent embeddings, facilitating scene-wide forecasting and interaction analysis.

Findings

Achieved competitive performance in the Waymo Open Interaction Prediction Challenge.

Clustered future waypoints to quantify and analyze agent interactions.

Merged and analyzed modes to identify interaction-induced conflicts.

Abstract

Self-driving vehicles rely on multimodal motion forecasts to effectively interact with their environment and plan safe maneuvers. We introduce SceneMotion, an attention-based model for forecasting scene-wide motion modes of multiple traffic agents. Our model transforms local agent-centric embeddings into scene-wide forecasts using a novel latent context module. This module learns a scene-wide latent space from multiple agent-centric embeddings, enabling joint forecasting and interaction modeling. The competitive performance in the Waymo Open Interaction Prediction Challenge demonstrates the effectiveness of our approach. Moreover, we cluster future waypoints in time and space to quantify the interaction between agents. We merge all modes and analyze each mode independently to determine which clusters are resolved through interaction or result in conflict. Our implementation is available at: https://github.com/kit-mrt/future-motion