On-Vehicle Position Estimation Using Parking Lot Landmark Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automated driving systems face challenges in accurately estimating the position of a vehicle in outdoor environments due to disturbances and low accuracy in position estimation.
Innovation Solution
An on-vehicle processing device that utilizes a combination of sensors, including cameras, GPS receivers, and vehicle speed sensors, to detect landmarks, calculate movement quantities, and estimate the vehicle's position by matching local peripheral information with pre-recorded parking lot data, using coordinate transformation and outlier exclusion techniques to enhance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If position estimation is performed using external sensors and map matching in automated driving systems, then the vehicle position can be calculated, but the accuracy becomes low and the system is affected by environmental disturbances
Solution Approach 1:
The patent segments the position estimation process into multiple independent components: external sensor-based map matching, internal sensor-based dead reckoning, and landmark recognition. Each component operates independently and contributes to the overall position estimation, allowing the system to maintain accuracy even when one component is affected by disturbances.
Solution Approach 2:
The patent creates a composite positioning system that combines multiple sensing modalities (external sensors, internal sensors, and landmark data) to form a robust position estimation mechanism. This composite approach leverages the strengths of each sensing method while compensating for their individual weaknesses, particularly regarding disturbance resistance.
2Measurement precision
If multiple sensors and processing methods are used to improve position estimation accuracy, then measurement precision increases, but device complexity increases
Solution Approach 1:
The patent merges the results from multiple independent position estimation methods (map matching, dead reckoning, and landmark recognition) into a unified position estimate. By combining these methods rather than using them in sequence, the system achieves higher accuracy without proportionally increasing complexity, as the methods work concurrently and independently.
Solution Approach 2:
The patent creates a multi-functional positioning system where the same processing unit handles multiple sensing modalities and estimation methods. This universal approach allows a single system to perform map matching, dead reckoning, and landmark recognition, reducing overall system complexity while maintaining high measurement precision.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided is an on-vehicle processing device that can estimate the position of a vehicle with higher accuracy. A storage unit 124 stores a parking lot point group 124A including a plurality of coordinates of points of a part of an object in a parking lot coordinate system. A sensor input unit (I/F 125) acquires peripheral information from a camera 102. A movement information acquisition unit (I/F 125) acquires movement information. A local peripheral information creation unit 121B generates local peripheral information 122B expressing second point group data including a position of the vehicle in a local coordinate system and a plurality of coordinates of points of a part of the object in the local coordinate system on the basis of the peripheral information and the movement information. A position estimation unit 121C estimates a correlation between the parking lot coordinate system and the local coordinate system on the basis of the parking lot point group 124A and the local peripheral information 122B, and estimates the position of the vehicle 1 in the parking lot coordinate system from the position of the vehicle 1 in the local coordinate system and the correlation.