Event-Based Vehicle Object Detection for Imminent Motion Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vehicle safety systems lack the necessary speed and accuracy in detecting impending collisions with objects in their vicinity, often due to limitations in sensor technology, leading to potential accidents that could be avoided with additional time to react.
Innovation Solution
The method employs event-based sensors with light-sensitive pixels to continuously monitor the environment, analyzing subtle internal movements and changes in objects, allowing for faster and more precise detection of potential collisions, and combines this data with camera information for enhanced object classification and tracking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If camera-based sensors are used to detect objects in the vehicle's surroundings, then object classification and tracking can be achieved, but the response time is insufficient due to the need to wait for complete image capture
Solution Approach 1:
The patent replaces traditional camera-based sensing systems with event-based sensors that detect changes in the optical environment. Instead of capturing complete images at fixed intervals, the event-based sensors only register changes exceeding a threshold, eliminating the need to wait for full frame capture and enabling immediate response to detected events.
Solution Approach 2:
The patent implements periodic threshold-based sampling where sensors continuously monitor the optical environment and trigger event detection only when changes exceed predetermined thresholds. This periodic evaluation with event-driven output reduces unnecessary data processing while maintaining detection accuracy for significant changes.
2Loss of time
If event-based sensors are used to detect light intensity changes, then response time is reduced, but the system must process and filter event data to distinguish relevant events from noise
Solution Approach 1:
The patent applies preliminary filtering and classification to event data streams, preprocessing events to identify patterns and characteristics before full analysis. By evaluating event properties such as temporal patterns, spatial distribution, and intensity changes in advance, the system reduces the complexity of subsequent processing while maintaining rapid response capabilities.
3Measurement precision
If the detection threshold for light intensity changes is set low to detect subtle movements, then detection sensitivity is improved, but the bandwidth requirement increases due to more frequent events
Solution Approach 1:
The patent dynamically adjusts detection parameters including threshold levels and sensitivity settings based on environmental conditions and detection needs. By adapting parameters in real-time, the system maintains high detection sensitivity for subtle movements while controlling event generation rates to manage bandwidth requirements effectively.
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 enables faster reaction times, reduces bandwidth requirements, and improves collision avoidance by accurately predicting object movements and changes in state, thereby enhancing vehicle safety and reducing the risk of accidents.
Implementation Method 1
an area of the surroundings is detected with at least one event-based sensor, wherein the event-based sensor has light-sensitive pixels
Data Source
Figure 1
Figure 2
AI summary
A method (100) for identifying potentially dangerous or endangered objects (41-43) in the surroundings (2) of a vehicle (1) includes the steps of: • detecting (110) a region (2a) of the surroundings (2) using at least one event-based sensor (3), wherein the event-based sensor (3) comprises light-sensor pixels (31) and wherein a relative change of the light intensity incident on a pixel (31) by at least a predetermined percentage causes the sensor (3) to output an event (31a) assigned to this pixel (31); • assigning (120) events (31a) output by the sensor (3) to objects (41-43) in the region (2a); • analysing (130) events (31a) assigned to an object (41-43) in respect of current movements (41a-43a) of the object (41-43) for at least one object (41-43) which has events (31a) assigned thereto; • ascertaining (140) an imminent movement (41b-43b) of the object (41-43) and/or an imminent change in state (41c-43c) of the object (41-43) from the current movements (41a-43a). The invention also relates to an associated computer program.