Occupancy Grid Mapping for Accurate Self-Position Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating the self-position of a moving body, such as vehicles and robots, can be inaccurate when similar environmental structures are present, leading to incorrect positioning.
Innovation Solution
An information processing apparatus and method that generates an attribute-attached occupancy grid map to represent the existence probability and attributes of obstacles around a moving body, allowing for accurate position estimation by matching the map with a pre-generated map, using sensors like cameras and LiDAR to differentiate between moving and non-moving objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If self-position estimation uses shape matching of environmental structures, then positioning can be performed using simple geometric features, but positioning accuracy deteriorates when similarly shaped structures are present
Solution Approach 1:
The patent segments the environmental structure into multiple feature points and divides the matching process into hierarchical stages: first matching coarse geometric shapes, then progressively refining with detailed attribute information (colors, textures, patterns) at each feature point. This segmentation allows the system to maintain simplicity while achieving high accuracy through multi-level feature comparison.
Solution Approach 2:
The patent transitions from two-dimensional shape matching to multi-dimensional feature space matching by incorporating attribute information (colors, textures, patterns, reflectivity) as additional dimensions. This dimensional expansion enables differentiation between similarly shaped structures by comparing their unique attribute signatures across multiple features simultaneously.
2Measurement precision
If attribute information is added to occupancy grid map, then positioning accuracy is improved through better feature discrimination, but data processing complexity increases
Solution Approach 1:
The patent segments attribute information extraction into modular components: color extraction module, texture extraction module, pattern recognition module, and reflectivity analysis module. Each module processes specific attribute types independently and outputs structured data that can be efficiently integrated with occupancy grid information, reducing overall processing complexity.
Solution Approach 2:
The patent performs preliminary attribute extraction and organization during the map generation phase, storing pre-processed attribute data in structured formats. This preliminary action eliminates the need for real-time attribute computation during positioning, significantly reducing processing complexity while maintaining high positioning accuracy through efficient data retrieval and comparison.
3Measurement precision
If multiple feature types are used for matching, then discrimination between similar structures is improved, but computational load increases
Solution Approach 1:
The patent implements periodic feature matching with varying intensity: using all feature types (color, texture, pattern, reflectivity) at critical matching stages, and switching to simplified feature subsets during routine updates or when confidence is high. This periodic full-featured matching reduces computational load while maintaining discrimination capability when needed.
Solution Approach 2:
The patent applies partial feature matching by selecting and using only the most discriminative feature types for each specific matching scenario. For example, using only color and pattern features when texture information is unavailable, or prioritizing reflectivity features for metallic objects. This selective partial action reduces computational load while maintaining sufficient discrimination capability.
Data Source
AI summary
Provided is a map generation unit that generates an attribute-attached occupancy grid map including an existence probability of an obstacle in a space around a moving body for each grid, and an attribute of the obstacle labelled in the occupancy grid map. A position estimation unit that estimates a position of the moving body by matching in a shape of a non-moving body and the attribute between the attribute-attached occupancy grid map and a pre-map that is the attribute-attached occupancy grid map prepared beforehand.