Aerial Target Tracking Control for Occlusion and Re-Targeting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current object tracking systems for aerial vehicles face challenges such as target deformation, occlusion, and long-term tracking due to limited computing resources and power, especially in environments where targets may become out of view.
Innovation Solution
A system and method for target tracking using a controller that extracts features from images captured by an aerial vehicle, applies a feature model to determine similarity, and generates control signals to maintain target tracking, including re-targeting capabilities once the target is lost.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If object tracking is implemented on aerial vehicles with limited computing resources, then automatic tracking function is added, but tracking reliability deteriorates due to target deformation, occlusion, and long-term tracking challenges
Solution Approach 1:
The tracking system is segmented into multiple independent modules: detection module, feature extraction module, tracking module, and re-targeting module. Each module handles specific aspects of the tracking task, allowing the system to maintain reliability through modular error handling and continued operation even when individual modules face challenges with difficult targets
Solution Approach 2:
The system performs preliminary actions by pre-processing images to extract robust features before tracking begins, and by implementing prediction mechanisms that anticipate target position and appearance changes. This preparatory work enables the tracker to handle deformation and occlusion more effectively during the tracking process
2Measurement precision
If feature extraction and similarity determination are performed continuously, then tracking accuracy is maintained, but computing resource consumption increases
Solution Approach 1:
The system maintains continuous tracking by updating feature models and similarity measurements at each frame, ensuring uninterrupted tracking accuracy. The continuous action is optimized by using efficient feature extraction methods that maintain precision while reducing computational burden compared to full-image processing
Solution Approach 2:
Instead of processing entire images for feature extraction, the system applies partial action by focusing only on relevant regions (such as bounding boxes or regions of interest) for feature extraction and similarity comparison. This selective processing maintains tracking accuracy while significantly reducing the computational resources required
3Use of energy by moving object
If the tracking system operates with limited power resources, then energy efficiency is improved, but tracking duration is limited and the system cannot maintain tracking over long periods
Solution Approach 1:
The system dynamically adjusts tracking parameters such as feature extraction frequency, similarity threshold, and model update rate based on available power resources and tracking conditions. This parameter adaptation allows the system to extend tracking duration by optimizing the balance between energy consumption and tracking performance throughout the operation
Data Source
AI summary
A method for controlling target tracking includes determining whether a tracking mode of a movable object is a manual tracking mode or an automatic tracking mode; when it is determined that the tracking mode is the manual tracking mode, determining a target for the movable object from one or more images based on a user input; when it is determined that the tracking mode is the automatic tracking mode, having the movable object determine the target automatically; and directing the movable object to track a target object using the target.