Radar Object Classification Fusion for Drone–Bird Differentiation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object detection systems, particularly radar systems, struggle with inefficient and slow classification of objects, leading to numerous false alarms and inability to reliably differentiate between objects like drones and birds, especially in swarm formations, and lack of control over system updates.
Innovation Solution
A classification method involving multiple iterative steps using a combination of statistical algorithms and machine learning-based recognition to quickly and accurately classify objects by filtering data, calculating kinematic characteristics, and applying weighted probabilities to determine a final class.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional classification methods are used, then system simplicity is maintained, but classification speed is slow (several seconds to tens of seconds) and reliability is poor (high false alarm rate)
Solution Approach 1:
The classification method is divided into multiple independent modules: a first classification module using statistical algorithms, a second classification module using machine learning recognition algorithms, and a fusion module that combines their results. Each module operates independently and can be processed in parallel, significantly improving classification speed while maintaining manageable complexity through modular design.
Solution Approach 2:
The patent combines multiple classification approaches (statistical algorithms and machine learning recognition algorithms) into a unified system. The results from both classification modules are fused to produce the final classification, leveraging the strengths of each method to achieve both high speed and high reliability simultaneously.
2Measurement precision
If conventional classification methods are used, then system complexity is low, but measurement precision is poor (inability to reliably differentiate between drones and birds)
Solution Approach 1:
The patent introduces an intermediary fusion module that combines the results from the first classification module (statistical algorithms) and the second classification module (machine learning recognition algorithms). This fusion module acts as a mediator that integrates multiple classification perspectives, significantly improving measurement precision by cross-validating results and reducing false alarms while maintaining systematic complexity through a clear intermediary structure.
3Adaptability or versatility
If radar manufacturers lock down access to their systems, then manufacturer control is maintained, but system adaptability is reduced (no control over operation or evolution)
Solution Approach 1:
The patent creates a copy of the classification functionality by implementing an independent classification device that interfaces with the radar system. This copy operates autonomously with its own algorithms and database, allowing the system to be adapted and updated without requiring access to or approval from the original radar manufacturer, thereby improving adaptability while maintaining ease of operation through independent control.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
- The method comprises at least one reception step (E1) for receiving tracks and a first classification determined by the radar for each of the detected objects, a data processing step (E2) for filtering the data received in the reception step (E1), a first calculation step (E3) for calculating kinematic characteristics relating to each of the tracks, a second calculation step (E4) for calculating statistical elements representing statistical distributions of the kinematic characteristics, a classification step (E5) for determining a second classification using a statistical algorithm and a third classification using a recognition algorithm, a final classification step (E6) for determining a final classification from the first, second and third classifications,and to assign to each of the tracks a final class and a transmission step (E7) for transmitting to a user system at least each of the tracks and the final class assigned to it, said method making it possible to quickly and reliably determine a class for each of the detected objects.,