Instance-Aware Predictive Navigation in Multi-Agent Environments

TL;DR

This paper introduces an Instance-Aware Predictive Control approach for multi-agent driving, forecasting interactions and scene structures to enable safe, efficient end-to-end autonomous navigation in dynamic environments.

Contribution

It proposes a novel multi-instance event prediction module and sequential action sampling strategy, advancing multi-agent driving policy learning without expert demonstrations.

Findings

Achieves state-of-the-art performance in CARLA multi-agent driving simulations.

Improves explainability and sample efficiency over existing methods.

Demonstrates effective anticipation of agent interactions and scene changes.

Abstract

In this work, we aim to achieve efficient end-to-end learning of driving policies in dynamic multi-agent environments. Predicting and anticipating future events at the object level are critical for making informed driving decisions. We propose an Instance-Aware Predictive Control (IPC) approach, which forecasts interactions between agents as well as future scene structures. We adopt a novel multi-instance event prediction module to estimate the possible interaction among agents in the ego-centric view, conditioned on the selected action sequence of the ego-vehicle. To decide the action at each step, we seek the action sequence that can lead to safe future states based on the prediction module outputs by repeatedly sampling likely action sequences. We design a sequential action sampling strategy to better leverage predicted states on both scene-level and instance-level. Our method establishes a new state of the art in the challenging CARLA multi-agent driving simulation environments without expert demonstration, giving better explainability and sample efficiency.