Autonomous Parking via Visual Sensors and T-Box
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automatic parking systems rely on fixed monitoring systems and high-precision maps, limiting their applicability and increasing computing costs, as they require a parking lot server for dynamic path planning, making them unsuitable without such infrastructure.
Innovation Solution
An automatic parking method and device that utilize a T-box for information interaction with a parking lot transmission device to obtain a rough waypoint map, then use vehicle visual information for global and local path planning, obstacle avoidance, and autonomous parking, eliminating the need for high-precision maps and fixed monitoring systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a fixed monitoring system and high-precision maps are used for automatic parking, then positioning accuracy is improved, but device complexity and renovation cost increase
Solution Approach 1:
The patent replaces the mechanical/fixed monitoring system with an onboard visual recognition system. Instead of using external fixed cameras and high-precision maps, the vehicle uses its own visual sensors (cameras) to detect lane lines, parking space markings, and obstacles, substituting complex external infrastructure with simpler onboard optical systems.
Solution Approach 2:
The vehicle performs self-positioning and self-navigation using its onboard visual sensors and processing units. Rather than relying on external monitoring systems to track and guide the vehicle, the system uses the vehicle's own camera to capture images, process visual information, and generate navigation commands independently.
2Productivity
If a parking lot server performs dynamic path planning, then path optimization is improved, but computing cost and system complexity increase
Solution Approach 1:
The patent extracts the path planning function from the external parking lot server and relocates it to the vehicle's onboard processing unit. The visual recognition system captures images, and the processor directly generates navigation commands locally, eliminating the need for continuous server communication and reducing computing energy consumption at the server end.
Solution Approach 2:
The system segments the automatic parking function into independent onboard modules: visual information acquisition, image processing, path planning, and control execution. This segmentation allows the vehicle to perform complete path planning independently without requiring a centralized server, distributing the computational workload to the vehicle itself.
3Difficulty of detecting and measuring
If visual information processing is used for obstacle detection, then detection capability is improved, but processing time and computational load increase
Solution Approach 1:
The system focuses visual processing on critical areas only - detecting lane lines, parking space markings, and potential obstacles in the vehicle's path. Rather than processing every pixel in the entire image, the processor targets specific regions of interest, reducing computational load while maintaining effective detection capability for safety-critical elements.
Data Source
AI summary
Provided are an automatic parking method and device. The automatic parking method includes: acquiring a global waypoint map of a parking lot where a vehicle is to be parked and position information of available parking spaces, and performing global path planning according to the global waypoint map and the available parking space position information; starting automatic driving according to the global path planning, acquiring vehicle visual information, and performing local path planning and obstacle avoidance processing according to the vehicle visual information; and searching for a parking space when automatically driving to the vicinity of the available parking spaces, and performing parking after finding an available parking space that satisfies a parking condition.