Asynchronous Camera Pair Detection for Low-Lying Road Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing perception systems in autonomous or semi-autonomous vehicles struggle to reliably detect small or low-lying objects at long ranges, such as debris or potholes, which can lead to insufficient time for collision avoidance maneuvers.
Innovation Solution
Utilizing asynchronous image pairs captured by cameras positioned at different locations to form approximate stereo pairs, enhancing stereo baselines orthogonal to the driving direction, and applying optical flow algorithms to compute displacement fields for object detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional synchronized stereo camera systems are used, then object detection can be performed, but small or low-lying objects at long ranges cannot be reliably detected
Solution Approach 1:
The system transitions from static synchronized image capture to dynamic asynchronous image capture with different exposure times. The first image is captured with a first exposure time and the second image with a second exposure time, allowing the system to adaptively capture objects at different ranges and orientations, thereby improving detection reliability for small and low-lying objects.
Solution Approach 2:
The patent introduces temporal dimension by using asynchronous image capture with different exposure times, transforming the traditional spatial stereo vision into a spatio-temporal detection system. This adds the time dimension to the stereo baseline, creating an enhanced detection capability that goes beyond conventional spatial arrangements.
2Reliability
If asynchronous image pairs with different exposure times are used, then detection range and reliability are improved, but system complexity increases
Solution Approach 1:
The system changes the exposure time parameter for different images in the pair. The first image is captured with a first exposure time and the second image with a second exposure time, allowing optimization of detection for different object ranges and types without adding physical sensors or complex hardware.
Solution Approach 2:
The patent replaces the need for complex mechanical stereo baseline adjustments with computational methods. By using asynchronous temporal stereo with different exposure times and applying optical flow algorithms, the system achieves enhanced detection capability through software processing rather than mechanical reconfiguration.
3Measurement precision
If synchronized stereo image pairs are required, then accurate depth perception is achieved, but system bandwidth and processing requirements increase
Solution Approach 1:
The system uses periodic alternating capture of image pairs from two sensors with different exposure times. This periodic asynchronous capture pattern allows the system to process images in manageable pairs rather than continuous streams, reducing bandwidth requirements while maintaining depth perception accuracy through the temporal stereo paradigm.
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
Enhances the ability to detect low-lying objects at greater distances, providing sufficient time for collision avoidance operations and reducing system bandwidth by staggering sensor triggers.
Implementation Method 1
systems and methods described herein may detect objects in an environment using optical flow-based algorithms (which may execute using one or more optical flow hardware accelerators) to compute displacement fields for images captured using asynchronous cameras
Implementation Method 2
by using a series of alternating, asynchronous images to form approximate stereo pairs, the systems of the present disclosure are able to enhance stereo baselines orthogonal to a driving direction
Data Source
AI summary
In various examples, optical flow-based algorithms may be used to detect objects in an environment by computing displacement fields for images captured using asynchronous cameras. As an example, an asynchronous set of cameras (e.g., two or more cameras) may capture a series of asynchronous images of an environment. Additionally, in some examples, the cameras may be positioned at different locations and capture different fields of view of the environment. Based at least on the differing image capture times and/or the differing fields of view of the images, image pixels corresponding to the same, physical locations in the environment may move locations between images of the series of images. The disclosed systems and methods may use optical flow algorithms to compute scores associated with the displacement/movement of the pixels throughout the series of images, as well as use these scores to detect objects in the environment.