SunnyParking: Multi-Shot Trajectory Generation and Motion State Awareness for Human-like Parking

TL;DR

SunnyParking introduces a dual-branch end-to-end framework for human-like autonomous parking, explicitly modeling motion states and employing Fourier features for high-precision target interaction, improving trajectory realism and gear-shift accuracy.

Contribution

The paper presents a novel dual-branch architecture with motion state awareness and Fourier feature-based target representation, addressing limitations of existing methods in complex parking maneuvers.

Findings

More robust and human-like parking trajectories

Significant improvement in gear-shift point localization

Open-sourced parking dataset for complex maneuvers

Abstract

Autonomous parking fundamentally differs from on-road driving due to its frequent direction changes and complex maneuvering requirements. However, existing End-to-End (E2E) planning methods often simplify the parking task into a geometric path regression problem, neglecting explicit modeling of the vehicle's kinematic state. This "dimensionality deficiency" easily leads to physically infeasible trajectories and deviates from real human driving behavior, particularly at critical gear-shift points in multi-shot parking scenarios. In this paper, we propose SunnyParking, a novel dual-branch E2E architecture that achieves motion state awareness by jointly predicting spatial trajectories and discrete motion state sequences (e.g., forward/reverse). Additionally, we introduce a Fourier feature-based representation of target parking slots to overcome the resolution limitations of traditional bird's-eye view (BEV) approaches, enabling high-precision target interactions. Experimental results demonstrate that our framework generates more robust and human-like trajectories in complex multi-shot parking scenarios, while significantly improving gear-shift point localization accuracy compared to state-of-the-art methods. We open-source a new parking dataset of the CARLA simulator, specifically designed to evaluate full prediction capabilities under complex maneuvers.