Transport Robot LiDAR Mapping for Dynamic Obstacle Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing SLAM algorithms for transport robots often incorrectly identify dynamic obstacles as fixed environments, leading to unnecessary route changes, failure to find routes, or inaccurate location estimation, and they also face challenges with data processing and storage as data accumulation increases with workspace size.
Innovation Solution
A transport robot equipped with a PC that includes modules for node generation, motion constraint calculation, loop constraint calculation, node pose optimization, LiDAR data transformation, depth image generation, matching, projection, weight adjustment, and map generation, which allows for efficient map creation and dynamic obstacle removal by processing LiDAR scan data and optimizing node poses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all tracking information for cells through which the laser passes and the cell on which the laser focuses is stored, then dynamic obstacles can be detected, but data processing and storage increase in proportion to the square of the workspace length
Solution Approach 1:
The patent segments the workspace into a grid of cells and processes tracking information cell by cell rather than storing all data globally. Each cell maintains only its own tracking information (first tracking information for focused cells and second tracking information for passed cells), dividing the large data processing task into many small independent cell-level operations.
Solution Approach 2:
The patent extracts only the essential tracking information needed for dynamic obstacle detection at the cell level, rather than storing complete laser scan data for the entire workspace. By extracting and storing only the count of focused hits and passed laser beams per cell, the data volume is dramatically reduced while preserving detection capability.
2Loss of information
If tracking information is accumulated for all cells in the workspace, then complete map coverage is achieved, but processing time and computational resources increase proportionally to the square of the workspace length
Solution Approach 1:
The patent divides the map into discrete cells and processes each cell independently and in parallel. This segmentation allows the system to achieve complete map coverage by iterating through all cells, but the computational burden is distributed across many small cell-level operations rather than one large workspace-level operation, reducing overall processing time.
Solution Approach 2:
The patent applies a simplified detection criterion using predetermined thresholds for first and second tracking information values. Rather than performing complex analysis on all cell data, the system uses partial action by comparing accumulated counts against fixed thresholds, achieving sufficient detection accuracy with minimal processing effort.
3Measurement precision
If general SLAM algorithms are applied in workspaces with frequent movement and temporary objects, then location estimation and mapping are performed, but dynamic obstacles are mistakenly recognized as fixed environments
Solution Approach 1:
The patent introduces dynamic tracking information that is continuously updated as the robot moves and the laser scans the environment. By maintaining first tracking information (number of focused hits) and second tracking information (number of passed beams) that evolve over time, the system can distinguish between static environmental features and dynamic obstacles based on changes in tracking patterns.
Solution Approach 2:
The patent uses feedback from accumulated tracking information to dynamically adjust obstacle classification. By continuously monitoring the first and second tracking information values for each cell and comparing them against predetermined thresholds, the system receives feedback that enables accurate distinction between fixed environments and dynamic obstacles, correcting misclassifications that occur in static SLAM approaches.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This solution reduces data generation and processing time by focusing on workspace boundaries, significantly lowering the risk of misidentifying obstacles and improving the robot's ability to navigate effectively in dynamic environments.
Implementation Method 1
a light detection and ranging (LiDAR) part configured to emit light, detect light in which the emitted light is reflected from an object in a global space
Data Source
AI summary
The present invention relates to a robot capable of autonomous traveling without mistaking a temporary obstacle as part of a fixed environment, and a dynamic obstacle removal method for the robot. A transport robot capable of removing a dynamic obstacle, according to the present invention, comprises: a body part (100); a transport unit (200) for moving the body part under the control of the body part; and a lidar unit (300) that emits light, detects the light reflected by an object in a global space to generate lidar scan data, and transmits the lidar scan data to the body part (100).