Asynchronous Event-Based Shape Tracking for Fast-Moving Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional ICP algorithms for object tracking in vision systems are limited by high computational redundancy and latency, making them unsuitable for real-time tracking of fast-moving objects, especially with traditional cameras, and lack efficiency in handling asynchronous event-based vision sensors.
Innovation Solution
A method for shape tracking using asynchronous event-based vision sensors, where events from a light sensor are processed to update a model of an object by minimizing a distance criterion, allowing for fast and precise tracking without waiting for complete image acquisition, and incorporating spatial and temporal constraints to resolve ambiguities between multiple objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If conventional ICP algorithms are used with traditional cameras, then object tracking can be performed, but the frame rate is insufficient and latency is high, making real-time tracking of fast-moving objects impossible
Solution Approach 1:
The patent transitions from static frame-based processing to dynamic event-based processing, where the system adapts its processing rate to the motion dynamics of objects in the scene. Events are processed as they occur rather than waiting for complete frames, enabling the system to dynamically respond to moving objects at their actual speed rather than camera frame rate limits.
Solution Approach 2:
The patent performs preliminary actions by maintaining and updating object models continuously in the background using accumulated events. When tracking is needed, the model is already prepared and can be immediately applied, eliminating the need to wait for complete image acquisition and enabling real-time tracking performance.
2Productivity
If conventional ICP algorithms process successive images, then object tracking is achieved, but computational redundancy occurs because the same information is repeatedly calculated in each frame
Solution Approach 1:
The patent extracts only the relevant changes from the scene by processing individual events rather than complete images. By taking out only the motion information carried by events and applying it to pre-maintained object models, the system eliminates the computational redundancy of processing entire frames while preserving all necessary tracking information.
Solution Approach 2:
The patent changes the fundamental parameter of data representation from complete frame images to sparse event representations. This parameter change transforms the processing from redundant full-frame analysis to efficient event-based model updates, dramatically reducing computational load while maintaining tracking accuracy.
3Measurement precision
If asynchronous event-based vision sensors are used, then temporal precision is increased and redundancy is reduced, but existing tracking methods lack efficiency in handling these asynchronous events
Solution Approach 1:
The patent implements self-service by having the asynchronous event stream directly update object models in real-time without requiring complex synchronization or frame assembly. The event-based system naturally serves the tracking function by providing continuous updates as events occur, eliminating the need for complex algorithms to reconcile asynchronous data from multiple sources.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables fast and accurate shape tracking with high temporal precision, reducing computational load and latency, and effectively handling moving objects in dynamic scenes, even when the sensor itself is in motion.
Implementation Method 1
receiving asynchronous information from a light sensor having a matrix of pixels arranged facing the scene, the asynchronous information comprising, for each pixel of the matrix, successive events originating from said pixel
Data Source
Figure 1~2C
Figure 3A~3B
Figure 4~5
AI summary
Asynchronous information is provided by a sensor (10) having a matrix of pixels disposed opposite the scene. The asynchronous information comprises, for each pixel of the matrix, successive events (ev(p, t)) originating from this pixel, that may depend on variations of light in the scene. A model representing the tracked shape of an object is updated after detecting events attributed to this object in the asynchronous information. Following the detection of an event, the updating of the model comprises an association of a point of the model with the event detected by minimizing a criterion of distance with respect to the pixel of the matrix from which the detected event originates. The updated model is then determined as a function of the pixel of the matrix from which the detected event originates and attributed to the object and of the associated point in the model, independently of the associations performed before the detection of this event.