Probabilistic Occupancy Mapping for Multi-Sensor Target Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As search missions progress, the log of potential target locations becomes unmanageable due to multiple sensor readings, leading to uncertainty and inefficiency in locating targets within a search area.
Innovation Solution
A dynamic probabilistic tracking method and system that generate an occupancy map, dividing the search area into cells, updating confidence levels based on sensor data, and removing detected targets from the list, allowing for efficient tracking and location of targets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple sensor readings are taken during search missions, then measurement precision improves, but the quantity of data and complexity of target location tracking increases uncontrollably
Solution Approach 1:
The search area is divided into discrete grid cells, and the target location problem is segmented into tracking probabilities for each cell independently. This segmentation transforms the unmanageable continuous search space into discrete, manageable units that can be processed systematically
Solution Approach 2:
A probabilistic occupancy map is introduced as an intermediary data structure between sensor readings and target location determination. This map accumulates and processes multiple sensor readings through probability updates, mediating between raw data and final target location decisions
2Reliability
If multiple sensor readings are accumulated, then target detection reliability improves, but the log size becomes unmanageable
Solution Approach 1:
The system changes the parameter representation from storing raw sensor readings to storing probability values for each grid cell. This parameter transformation compresses the data volume while maintaining detection reliability through cumulative probability updates based on sensor readings
3Ease of operation
If the search area is divided into discrete cells, then ease of operation improves, but the area coverage requires more data processing
Solution Approach 1:
The search area is divided into discrete grid cells, making the search space manageable and easier to operate with. Each cell can be independently processed and updated, simplifying the overall search management
Solution Approach 2:
Multiple sensor readings and observations are merged into a single probabilistic occupancy map that represents the entire search area. This merging consolidates data processing efforts and improves overall productivity by avoiding redundant processing
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A system and method for dynamic probabilistic tracking of search targets within a search area (100) generates a displayable occupancy map (212) of the search area. The occupancy map is divided into individual map cells (300) and seeded with a list (214) of unlocated targets (e.g., targets of one or more types known or thought to be within the search area but which have not been precisely located therewithin). Each map cell of the occupancy map reflects the confidence or likelihood that an unlocated target will be detected within that cell. As a search mission proceeds, sensor datasets are received from the searching assets and correlated to identify unlocated targets. As unlocated targets are detected, located with sufficient accuracy, and removed from the list, the confidence levels (304, 306, 308, 310) of other map cells are reflected to indicate changes in the probability that any remaining unlocated targets will be detected therein.