Collaborative Tracking for Embedded Devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object tracking systems in video streams face challenges with input quality, content complexity, and object interaction, requiring significant computational resources, which are impractical for embedded devices like cameras, leading to inefficient processing and bandwidth constraints.
Innovation Solution
The Collaborative Tracking (CTR) approach transfers processing between a camera and a remote computing device over a network, using a simple tracker on the camera and switching to advanced tracking on a remote PC when necessary, determining when and what data to transfer based on track quality measures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If advanced tracking methods are used to improve tracking robustness against input quality and content complexity, then tracking reliability is improved, but computational resource requirements increase making them impractical for embedded devices
Solution Approach 1:
The tracking system is segmented into two distinct components: a simple tracker embedded in the camera device and an advanced tracker on a remote computer. The simple tracker handles routine tracking tasks locally, while the advanced tracker processes complex cases remotely, dividing the computational workload to achieve both reliability and resource efficiency.
Solution Approach 2:
A collaborative tracking mechanism acts as an intermediary between the simple embedded tracker and the advanced remote tracker. This mediator monitors tracking quality and automatically transfers control between the two trackers, enabling the embedded device to achieve advanced tracking reliability without permanently requiring advanced computational resources.
2Measurement precision
If all tracking processing is performed on a separate computer system to handle complex situations, then tracking accuracy is improved, but bandwidth requirements increase significantly
Solution Approach 1:
Only the necessary tracking data (object positions, track history, and quality metrics) is extracted and transmitted to the remote computer, rather than transmitting entire video frames or excessive data. This selective data extraction maintains tracking accuracy while significantly reducing bandwidth consumption.
Solution Approach 2:
The system performs partial processing locally with the simple tracker and only engages the advanced remote tracker when needed for complex situations. This partial action approach achieves sufficient accuracy for most cases while avoiding the excessive bandwidth consumption that would result from always using the advanced tracker.
3Device complexity
If simple tracking methods are used on embedded devices to reduce computational requirements, then device complexity is reduced, but tracking reliability deteriorates in complex situations
Solution Approach 1:
The collaborative tracking mechanism serves as an intermediary that monitors the performance of the simple embedded tracker and automatically activates the advanced remote tracker when tracking reliability is at risk. This allows the embedded device to maintain simplicity while achieving reliable tracking through the mediator's coordination.
Solution Approach 2:
The system replaces the need for permanently complex embedded hardware with a software-based collaborative approach. The simple embedded device is enhanced through software coordination with a remote advanced tracker, substituting mechanical/hardware complexity with a flexible software architecture that achieves reliability on demand.
4Quantity of substance
If the simple tracker operates autonomously without remote assistance to reduce bandwidth usage, then bandwidth consumption is reduced, but tracking accuracy deteriorates when encountering complex situations
Solution Approach 1:
The tracking system dynamically adjusts its operation mode based on the complexity of the current situation. The simple tracker operates autonomously in normal conditions to minimize bandwidth usage, but automatically transitions to remote advanced tracking when complex situations are detected, achieving adaptive accuracy without excessive bandwidth consumption.
Solution Approach 2:
The collaborative tracking mechanism incorporates feedback by monitoring tracking quality metrics and automatically initiating remote assistance when accuracy deteriorates. This feedback loop ensures that the system maintains tracking accuracy by engaging the advanced tracker only when necessary, avoiding unnecessary bandwidth consumption while preserving precision when needed.
Data Source
AI summary
Disclosed is a system (200) and method (101) for collaborative tracking of an object, the method comprising updating (105) the track with an object measurement using a camera tracking module (230), determining (110) a track quality measure for the updated track, based on the track quality measure, determining (120) whether a second tracking module (260), remotely located from the camera, should be applied, if the second tracking module (260) is to be applied, selecting (130) data describing the track and the object, transmitting (140) the selected data to the second tracking module over a network (240) that imposes constraints of bandwidth and/or latency, and applying (150) the second tracking module (260) to determine the next position of the object in the track.