Occupancy Grid Fusion for Mobile Robot Navigation Around Difficult Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing perception mechanisms for mobile industrial robots face difficulties in navigating constrained and populated environments, particularly with objects of varying heights, dimensions, transparency, reflectivity, darkness, or blackness, leading to inefficient and imprecise navigation.
Innovation Solution
A method utilizing a robot-mounted color-depth sensor to generate an occupancy grid by merging elevation and chromaticity maps, assigning occupancy probabilities based on dominant chromaticity values and low elevations to improve navigation accuracy and safety.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional perception mechanisms are used to detect objects in the environment, then the robot can identify physical objects, but it fails to accurately detect mixed, small, transparent, reflective, dark, or black objects
Solution Approach 1:
The patent transitions from traditional 2D image-based perception to 3D spatial perception by introducing elevation maps. By adding the vertical dimension (elevation) to the detection space, the system can distinguish objects based on their height relative to the substrate, effectively detecting transparent, reflective, dark, or small objects that are invisible or indistinguishable in 2D chromaticity space alone.
Solution Approach 2:
The patent changes the detection parameters from relying solely on chromaticity (color) to combining chromaticity with elevation (height). This parameter change allows the system to detect objects that have similar chromaticity values to the substrate but differ in elevation, such as transparent or dark objects, by computing occupancy probability based on both chromaticity distance and elevation threshold.
2Measurement precision
If the occupancy grid is updated frequently to improve navigation accuracy, then the navigation precision increases, but the time spent generating or updating the occupancy grid increases
Solution Approach 1:
The patent applies partial action by selectively updating only those cells in the occupancy grid where objects are detected or where the robot has moved. Instead of recalculating the entire occupancy grid from scratch, the system performs localized updates based on sensor data changes, reducing computational load and update time while maintaining navigation precision.
Solution Approach 2:
The patent implements continuous refinement of the occupancy grid by incrementally updating cell occupancy probabilities as new sensor data arrives. The system continuously integrates elevation and chromaticity information to refine the occupancy map in real-time, maintaining up-to-date navigation information without requiring complete reprocessing of the entire environment.
3Measurement precision
If the robot uses a color-depth sensor to compute occupancy probabilities based on elevation and chromaticity, then the navigation accuracy improves, but the data processing complexity increases
Solution Approach 1:
The patent segments the complex task of occupancy grid computation into distinct processing stages: (1) computing elevation maps from depth data, (2) computing chromaticity maps from color data, (3) calculating chromaticity distances, (4) determining occupancy probabilities based on elevation and chromaticity thresholds, and (5) merging partial probabilities. This segmentation simplifies the overall processing complexity by breaking down the problem into manageable, independent steps.
Solution Approach 2:
The patent introduces intermediary data structures (elevation maps and chromaticity maps) that mediate between the raw sensor data and the final occupancy probability computation. These intermediary representations organize the complex sensor data into structured formats that facilitate efficient processing and reduce the complexity of the final occupancy grid update operation.
Data Source
AI summary
A robot movable on a substrate is controlled on the basis of an occupancy grid of cells, where each cell is associated with an occupancy probability that some physical object is present in the cell. From an elevation map and a chromaticity map, an occupancy probability may be assigned as follows: a first partial occupancy probability p1 is derived from a mean elevation of each cell; a distribution of the chromaticity is estimated and dominant chromaticity values are identified; a rule is applied that assigns a lower second partial occupancy probability p2 to a cell when the chromaticity of the cell if its chromaticity is one of dominant chromaticity values in the chromaticity map; and the occupancy probability is obtained by merging the first and second partial occupancy probabilities.