PanguMotion: Continuous Driving Motion Forecasting with Pangu Transformers

TL;DR

PanguMotion introduces a novel Transformer-based framework for continuous driving motion forecasting, leveraging Pangu-1B language model features to improve trajectory predictions in real-world scenarios.

Contribution

The paper presents PanguMotion, integrating Pangu-1B Transformers into motion forecasting to model temporal continuity and contextual correlations in autonomous driving.

Findings

Enhanced trajectory prediction accuracy.

Effective modeling of temporal continuity.

Improved performance on Argoverse 2 dataset.

Abstract

Motion forecasting is a core task in autonomous driving systems, aiming to accurately predict the future trajectories of surrounding agents to ensure driving safety. Existing methods typically process discrete driving scenes independently, neglecting the temporal continuity and historical context correlations inherent in real-world driving environments. This paper proposes PanguMotion, a motion forecasting framework for continuous driving scenarios that integrates Transformer blocks from the Pangu-1B large language model as feature enhancement modules into autonomous driving motion prediction architectures. We conduct experiments on the Argoverse 2 datasets processed by the RealMotion data reorganization strategy, transforming each independent scene into a continuous sequence to mimic real-world driving scenarios.