Event-Based Vision Filtering for Early Secondary Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automatic emergency brake assistants face challenges in detecting secondary objects in time to prevent collisions due to processing delays and false positives, which can lead to rear-end collisions.
Innovation Solution
A system utilizing event-based vision sensors with a discriminator module to differentiate between events caused by the primary object's motion and those from secondary objects, focusing processing resources on relevant events and incorporating a classifier module for accurate classification, along with a mitigation module for collision avoidance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional vision sensors with frame-based processing are used, then complete image information is available for analysis, but processing time increases and detection speed decreases
Solution Approach 1:
The patent changes the fundamental parameter of data representation from continuous frame-based images to discrete event-based signals. Event-based pixels output individual events whenever light intensity changes exceed a threshold, transforming the temporal and spatial structure of visual data. This parameter change enables asynchronous processing of only relevant visual changes, eliminating the need to process complete frames and significantly reducing processing time while maintaining detection accuracy for moving objects.
Solution Approach 2:
The system dynamically adapts its processing based on scene activity. Instead of processing all pixels in every frame at a fixed rate, the event-based sensor dynamically generates events only when and where visual changes occur. The discriminator module dynamically filters events based on motion patterns, adapting processing resources to the actual scene dynamics rather than operating at a static frame rate, thus reducing processing time while maintaining detection precision.
2Loss of time
If event-based vision sensors are used to increase detection speed, then processing time is reduced, but the volume of event data increases significantly
Solution Approach 1:
The discriminator module extracts and separates relevant events from the large stream of raw event data. It identifies and extracts events that match the motion pattern of the primary object while filtering out events caused by static objects or irrelevant motion. This extraction process reduces the effective data volume that requires further processing, transforming the overwhelming event stream into a manageable set of relevant detection candidates.
Solution Approach 2:
The system initially processes all events excessively to ensure no potential threats are missed, then applies the discriminator module to filter out irrelevant events. This partial action approach processes the full event stream upfront (excessive action) but then reduces the workload by eliminating static object events (partial action), achieving a balance between comprehensive detection and manageable processing load.
3Reliability
If sophisticated processing is applied to reduce false positives, then detection reliability improves, but processing time increases
Solution Approach 1:
The discriminator module performs preliminary filtering of events based on motion pattern matching before full object detection and classification. By pre-processing events to identify those matching the primary object's motion characteristics, the system establishes a filtered event stream early in the processing pipeline. This preliminary action reduces the number of events requiring sophisticated analysis, thereby maintaining detection reliability while reducing overall processing time.
Solution Approach 2:
The processing pipeline is segmented into distinct stages: event generation by vision sensor, motion-based filtering by discriminator module, and sophisticated analysis by object detection algorithms. This segmentation allows each stage to operate independently and optimally, with the discriminator module handling the time-critical filtering of relevant events while more computationally intensive reliability checks are applied only to the reduced set of candidate events, balancing both reliability and speed.
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
This approach significantly reduces processing time and false detections, enabling earlier detection of secondary objects and reducing the risk of collisions by prioritizing relevant events and providing timely collision avoidance actions.
Implementation Method 1
a relative change in the light intensity impinging onto an event-based pixel of the vision sensor by at least a predetermined percentage causes the vision sensor to emit an event associated with this event-based pixel
Data Source
AI summary
A system (1) for detecting dynamic secondary objects (55) that have a potential to intersect the trajectory (51) of a moving primary object (50), comprising a vision sensor (2) with a light-sensitive area (20) that comprises event-based pixels (21), so that a relative change in the light intensity impinging onto an event-based pixel (21) of the vision sensor (2) by at least a predetermined percentage causes the vision sensor (2) to emit an event (21a) associated with this event-based pixel (21), wherein the system (1) further comprises a discriminator module (3) that gets both the stream of events (21a) from the vision sensor (2) and information (52) about the heading and/or speed of the motion of the primary object (50) as inputs, and is configured to identify, from said stream of events (21a), based at least in part on said information (52), events (21b) that are likely to be caused by the motion of a secondary object (55), rather than by the motion of the primary object (50).Vision sensors (2) for use in the system (1).A corresponding computer program.