Avoidance Map Segmentation for Autonomous Mobile Constraint Regions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous mobile devices (AMDs) face challenges in navigating physical spaces efficiently while avoiding obstacles and ensuring safe movement, as existing methods are prone to noise in sensor data and lack effective algorithms for determining constraint regions to avoid collisions and optimize route planning.
Innovation Solution
The system generates an avoidance map by processing occupancy maps using despeckling, Gaussian blur, and medial axis transform algorithms to identify constraint regions, allowing the AMD to determine safe movement paths and avoid blocking users by classifying areas as open, narrow, or closed based on user and device diameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sensor data is used directly for navigation, then the AMD can detect obstacles, but the noise in sensor data reduces measurement precision
Solution Approach 1:
The system applies preprocessing operations (despeckling filter, Gaussian blur) to the occupancy map before constraint region determination. This preliminary action cleans the sensor data by removing noise and filling gaps, improving measurement precision before the data is used for navigation decisions, thereby resolving the contradiction between detection accuracy and navigation safety
Solution Approach 2:
The patent introduces an intermediary processing pipeline between raw sensor data and navigation decisions. The occupancy map is transformed through multiple processing stages (despeckling → Gaussian blur → skeletonization → constraint region determination) to produce a refined representation that eliminates noise while preserving essential obstacle information, thus improving both measurement precision and navigation reliability
2Productivity
If the AMD moves freely without constraint regions, then the AMD can reach destinations quickly, but the risk of blocking users and collisions increases
Solution Approach 1:
The system determines constraint regions with different characteristics (open, narrow, closed) based on local spatial conditions. By analyzing the skeleton of the occupancy map and calculating distance transforms, the system identifies regions where the AMD should be constrained and applies appropriate movement restrictions only in those specific areas, allowing free movement in safe zones while preventing collisions and user obstruction in hazardous zones, thus resolving the contradiction between navigation efficiency and safety
Solution Approach 2:
The constraint regions are dynamically determined based on the current occupancy map and spatial configuration. The system continuously updates the avoidance map by processing the latest sensor data, allowing the AMD to adapt its movement constraints in real-time according to changing environmental conditions, maintaining navigation efficiency while responding to dynamic safety requirements
3Measurement precision
If complex processing algorithms are applied to occupancy maps, then constraint regions are determined more accurately, but the computational complexity increases
Solution Approach 1:
The patent segments the constraint determination process into distinct modular stages: despeckling filter application, Gaussian blur processing, skeletonization, and distance transform calculation. Each stage performs a specific function and can be independently optimized or implemented, reducing overall computational complexity while maintaining determination accuracy through systematic progressive processing
Solution Approach 2:
The system applies a sequence of processing operations that progressively refine the occupancy map. Rather than using a single complex algorithm, the patent employs multiple simpler operations (despeckling, then Gaussian blur, then skeletonization) where each operation addresses specific aspects of noise and spatial representation, achieving high accuracy through cumulative partial actions that are computationally more manageable than a single comprehensive complex algorithm
Data Source
AI summary
A system uses an obstacle map to determine an avoidance map that indicates one or more constraint regions that are present in a physical space. The obstacle map is processed using a smoothing function and the smoothed map processed with a medial axis transform function to generate a MAT skeleton. The skeleton is pruned to remove spurs and then segmented based on MAT values that are representative of distances in the physical space between the skeleton and an obstacle. The avoidance map is determined by associating constraint regions with the segments. Constraint regions may indicate areas in the physical space that are open for the autonomous mobile device (AMD) to stop, narrow areas for which stopping should be avoided, and closed areas within which the AMD is not to stop during ordinary operation.