Indoor Vehicle Positioning via Semantic Feature Points
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing indoor positioning technologies for autonomous driving, such as vision acquisition mapping and lidar mapping, require high computational power, result in complex and costly hardware, and face challenges in data integration and feature point accumulation, especially in satellite signal-absent environments like underground parking lots.
Innovation Solution
A vehicle positioning method and device that utilizes semantic feature points, identified by physical quantities like attitude, speed, and steering wheel data, to create and correct indoor semantic maps, simplifying the processing algorithm and reducing data volume, enabling low-cost indoor positioning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If vision acquisition mapping or lidar mapping is used for indoor positioning, then positioning accuracy is improved, but hardware complexity and cost increase significantly
Solution Approach 1:
The patent extracts and uses only the necessary semantic feature points (entry/exit slope points, speed bumps, road connection points) from the environment rather than processing all visual or lidar data. This selective extraction reduces hardware requirements while maintaining positioning accuracy through meaningful feature identification.
Solution Approach 2:
The patent replaces expensive vision acquisition and lidar systems with simpler, cheaper sensors that can identify semantic feature points. The solution uses basic vehicle sensors (speed, steering angle, acceleration) to detect and map semantic features, eliminating the need for costly dedicated positioning hardware.
2Measurement precision
If vision acquisition mapping or lidar mapping is used for indoor positioning, then positioning accuracy is improved, but computational power requirements increase
Solution Approach 1:
The patent extracts only essential semantic feature points (entry/exit slope points, speed bumps, road connection points) from sensor data rather than processing complete visual or point cloud maps. This extraction approach reduces computational load significantly while maintaining positioning accuracy through selective feature processing.
Solution Approach 2:
The patent performs partial mapping by only recording and processing semantic feature points rather than complete environmental maps. This partial action approach reduces computational requirements for map generation and processing while providing sufficient information for accurate vehicle positioning.
3Measurement precision
If vision acquisition mapping or lidar mapping is used for indoor positioning, then positioning accuracy is improved, but data integration difficulty increases
Solution Approach 1:
The patent merges multiple vehicle sensor data streams (speed, steering angle, acceleration, attitude) with semantic feature point identification into a unified positioning system. This integration approach simplifies data processing by combining available vehicle data with semantic map matching, avoiding the complexity of integrating separate vision and lidar systems.
Solution Approach 2:
The patent creates a semantic map that serves multiple functions: positioning, navigation, and path planning. The semantic feature points (entry/exit slope points, speed bumps, road connection points) provide universal reference markers that can be used for various autonomous driving functions, reducing the need for separate specialized systems.
4Measurement precision
If feature points are constantly accumulated in indoor positioning, then positioning accuracy is improved, but data volume increases significantly
Solution Approach 1:
The patent extracts only essential semantic feature points (entry/exit slope points, speed bumps, road connection points) from the environment rather than accumulating all possible feature data. This selective extraction maintains positioning accuracy by focusing on meaningful features while significantly reducing data volume through purposeful data selection.
Data Source
Figure 1
Figure 2
AI summary
The embodiments of the disclosure relate to a vehicle positioning method and device, a vehicle and a storage medium. The vehicle positioning method comprises: identifying (102) a site entrance; acquiring (103) an indoor semantic map of the site according to the site entrance, the indoor semantic map comprising semantic feature points of the site; correcting (104) positioning information of the vehicle in the indoor semantic map according to the semantic feature points acquired during the movement of the vehicle; and acquiring (105) the positioning information of the vehicle according to movement information of the vehicle relative to a previous semantic feature point. Methods for identifying these points used in the method are mainly physical quantities when the vehicle runs. These physical quantities are calculated by other parts of the vehicle, and the identifying process is greatly simplified accordingly.