Big Cell Grid Mapping for Real-Time Indoor Robot Pose Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for estimating the posture of a moving object, such as GPS and grid maps, face challenges in indoor use due to high error ranges and memory limitations, while feature maps struggle with real-time calculations and accessibility issues.
Innovation Solution
A big cell grid map method that divides the map into finite cells, allowing for efficient storage and access of map feature point information, enabling accurate and fast posture estimation by composing local and global map feature points and determining the object's position and posture using sensing information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a grid map is used to represent the environment, then real-time calculation is enabled, but the memory required increases significantly
Solution Approach 1:
The patent divides the map into a hierarchical structure of big cells (macro-level) and small cells (micro-level). Each big cell contains multiple small cells, allowing the system to work with coarse-grained representations for global navigation (reducing memory) while maintaining fine-grained details when needed (enabling real-time calculation for local operations). This segmentation resolves the contradiction by enabling real-time processing at appropriate scales without requiring the entire fine-grained map in memory simultaneously.
Solution Approach 2:
The patent applies different levels of detail to different regions of the map. Big cells provide coarse environmental information for areas far from the robot, while small cells provide detailed information for areas near the robot where precise navigation is needed. This local quality approach allows real-time calculation for local navigation while minimizing overall memory requirements by not maintaining fine detail everywhere.
2Quantity of substance
If a feature map is used to compress and store only necessary information, then memory efficiency is improved, but accessibility by position information is lost
Solution Approach 1:
The patent implements a nested structure where big cells (coarse representation) contain multiple small cells (fine representation). The big cell level provides position-based accessibility for general navigation, while small cells within each big cell provide detailed feature information when needed. This nesting allows the system to access position information efficiently at the big cell level while maintaining memory efficiency, and drill down to small cells only when detailed information is required.
Solution Approach 2:
The patent adds a hierarchical dimension to the traditional 2D map representation by introducing multiple levels of cell granularity. This dimensional change allows the system to access map information at different resolutions depending on the task, providing position-based accessibility for global operations while maintaining compressed storage through the hierarchical structure.
3Measurement precision
If small cells are used to express the environment in sufficient detail, then environmental representation accuracy is improved, but the amount of data for calculation increases
Solution Approach 1:
The patent segments the environmental representation into hierarchical levels: big cells for global environmental structure and small cells for local detailed representation. This segmentation allows the system to maintain high measurement precision where needed (in small cells near the robot) while reducing the total data quantity by using coarse representations (big cells) for the rest of the environment.
Solution Approach 2:
The patent applies high-detail small cells only to regions where precise environmental representation is critical (near the robot's current position and planned path), while using lower-detail big cells for distant regions. This local quality approach maintains measurement precision for navigation-critical areas while minimizing the overall data quantity.
Data Source
AI summary
Embodiments are proposed, including: a method for estimating the posture of a moving object by using a big cell grid map; a recording medium in which a program for implementing the method is stored; and the computer program stored in the medium to implement the method. More particularly, there are provided the embodiments including: a method for estimating the posture of a moving object by using a big cell grid map, the method dividing a map into a plurality of finite cells, and estimating and correcting the posture of a moving robot in the map (the big cell grid map) where map feature point information about a map feature point corresponding to a position of a corresponding cell is matched for each cell; a recording medium where a program for implementing the method is stored, and the computer program stored in the medium to implement the method.