Entity Tracking System Using Historical Pattern Indexing for Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing tracking systems for entities, such as objects or individuals, face challenges with intermittent and inaccurate geospatial and temporal data, which are retrospective and do not provide meaningful information about future locations, limiting their ability to predict or identify anomalies in movement patterns.
Innovation Solution
A system and method that utilize a processor to create tracks based on training data, generate an index of these tracks, and execute queries to output feature values such as anomaly scores or predictions, incorporating attributes like city, country, and entity type, to analyze and predict the movement of test entities like people, ships, or vehicles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sensor-based tracking is used to obtain entity location data, then location information can be collected, but the data becomes intermittent and inaccurate
Solution Approach 1:
The system pre-processes historical sensor data to create cleaned, continuous tracks before they are needed for querying. This preliminary action involves filtering intermittent data points, interpolating missing locations, and establishing baseline movement patterns, so that when queries are executed, reliable continuous tracks are already available for accurate anomaly detection and prediction.
Solution Approach 2:
The system introduces an intermediary processing layer between raw sensor data and query results. This intermediary layer includes data cleaning modules, track generation algorithms, and pattern recognition systems that transform intermittent sensor readings into continuous, reliable movement tracks, mediating the gap between discontinuous input data and the need for continuous analysis.
2Loss of information
If historical tracking data is stored and analyzed, then movement patterns can be identified, but the system cannot predict future locations
Solution Approach 1:
The system implements feedback by continuously comparing actual entity movements against predicted trajectories generated from historical patterns. When deviations are detected, the system adjusts future predictions based on this feedback, creating a dynamic prediction model that learns from past errors and improves over time, enabling accurate future location forecasting.
Solution Approach 2:
The system performs preliminary analysis of historical movement patterns to establish baseline trajectories and speed profiles before future prediction is needed. By pre-processing historical data to extract meaningful movement characteristics and storing these as predictive models, the system can quickly generate accurate future location predictions when queries are executed.
3Loss of information
If comprehensive tracking analysis is performed to identify anomalies and predict movements, then meaningful insights are obtained, but the processing complexity increases
Solution Approach 1:
The system segments the complex tracking analysis into distinct modular components: data cleaning modules, track generation modules, pattern recognition modules, anomaly detection modules, and prediction modules. Each module handles a specific aspect of the analysis independently, processing data in discrete stages rather than attempting comprehensive analysis in a single complex operation, thereby reducing overall system complexity.
Solution Approach 2:
The system applies partial action by implementing selective anomaly detection that focuses on specific movement characteristics or geographic regions based on query parameters, rather than analyzing all possible aspects of every track uniformly. This allows the system to obtain meaningful insights for specific use cases without the computational overhead of exhaustive comprehensive analysis.
Data Source
AI summary
Systems and methods for tracking at least one entity. Systems according to various embodiments may include an interface for at least receiving training data representing movements of a plurality of training entities and a query regarding a test entity's movement, a memory, and at least one processor executing instructions stored on the memory to create a plurality of tracks, generate an index of the plurality of tracks, execute the received query, and output a feature value with respect to the test entity.