Abstract:The autonomous valet parking (AVP) functionality in self-driving vehicles is currently capable of handling most simple parking tasks. However, further training is necessary to enable the AVP algorithm to adapt to complex scenarios and complete parking tasks in any given situation. Training algorithms with real-world data is time-consuming and labour-intensive, and the current state of constructing simulation environments is predominantly manual. This paper introduces an approach to automatically generate 3D underground garage simulation scenarios of varying difficulty levels based on pre-input 2D underground parking structure plans.

What problem does this paper attempt to address?

The main objective of this paper is to address the adaptability and parking task completion capability of Automated Valet Parking (AVP) algorithms in complex scenarios. Specifically, the paper attempts to improve existing AVP systems through the following points: 1. **Automatic Generation of Underground Parking Garage Simulation Scenarios**: The current method of constructing simulation environments is mainly manual, time-consuming, and labor-intensive. The paper proposes a method to automatically generate 3D underground parking garage simulation scenarios of varying difficulty levels based on pre-input 2D underground parking garage layout diagrams. 2. **Improving the Accuracy of Perception Algorithms**: To ensure that AVP vehicles can safely complete parking tasks in complex environments, it is essential to ensure that their perception algorithms can timely and accurately detect surrounding traffic participants and identify lane markings. However, in underground parking garage environments, due to factors such as building structure occlusion, vehicle occlusion, and lighting arrangements, perception algorithms may fail to accurately identify surrounding vehicles or pedestrians in certain situations. 3. **Evaluating the Impact of Different Factors on Perception**: The paper focuses on studying the impact of occlusion and lighting intensity on perception algorithms and conducts multiple experimental verifications using the YOLO algorithm. The experimental results show that the specific position of occlusion and the type of occlusion have a significant impact on vehicle recognition, while lighting intensity has almost no impact on the performance of the YOLO algorithm. Through these efforts, the paper aims to create more challenging underground parking garage simulation test scenarios to enhance the performance of AVP algorithms in complex environments.

Exploration of the Assessment for AVP Algorithm Training in Underground Parking Garages Simulation Scenario

Optimization-based Motion Planning for Autonomous Parking Considering Dynamic Obstacle: A Hierarchical Framework

PCG-based Static Underground Garage Scenario Generation

Development of Occupancy Prediction Algorithm for Underground Parking Lots

Acceleration method for generating perception failure scenarios based on editing Markov process

Automated Parking Planning with Vision-Based BEV Approach

Research on Reliable and Safe Occupancy Grid Prediction in Underground Parking Lots

Autonomous Exploration for Automated Valet Parking Based on Road Structure

Human-Like Reverse Parking Using Deep Reinforcement Learning with Attention Mechanism

Hierarchical Parking Path Planning Based on Optimal Parking Positions

A Trajectory Planning Method for Autonomous Valet Parking via Solving an Optimal Control Problem

A Unified Motion Planning Method for Parking an Autonomous Vehicle in the Presence of Irregularly Placed Obstacles

Reservation Enhanced Autonomous Valet Parking Concerning Practicality Issues

Control algorithm of path planning and path tracking for automated valet parking

Modelling autonomous vehicle parking: An agent‐based simulation approach

Toward Efficient Physical and Algorithmic Design of Automated Garages

Assessment of future parking systems with autonomous vehicles through agent-based simulation: A case study of Hangzhou, China

Dynamic indoor mapping for AVP: Crowdsourcing mapping without prior maps

A Priori Map-Based Automated Valet Parking with Accurate Adjustment Ability for Automatic Charging

An Intelligent Parking Assist System

Optimal Planning of Parking Facilities in the Era of Automated Vehicles