Vehicle Object Detection Using Radar and Monocular Camera Domain Overlap
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Vehicle collision avoidance systems using a monocular camera instead of stereoscopic cameras face challenges in accurately detecting object distance and azimuth angle, leading to lower accuracy compared to systems using stereoscopic cameras.
Innovation Solution
An object detection apparatus that defines object domains using both radar and monocular camera data, determining overlap on an XY-plane to identify objects and adjust error domains based on object types to prevent false determinations, with the monocular camera's azimuthal range defined to match its detection characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a monocular camera is used instead of a stereoscopic camera, then device complexity is reduced, but measurement precision of object distance deteriorates
Solution Approach 1:
The patent combines radar and monocular camera detection systems to compensate for the monocular camera's inferior depth perception. By merging the distance measurement capability of radar with the visual recognition capability of the monocular camera, the system achieves accurate object detection without requiring complex stereoscopic camera hardware.
Solution Approach 2:
The patent introduces an intermediary coordinate transformation process that converts monocular camera image coordinates to radar coordinate system. This intermediary transformation, combined with error domain definitions, enables the monocular camera's limited depth information to be effectively integrated with radar data, resolving the measurement precision issue.
2Measurement precision
If object domains are defined with strict coincidence criteria, then measurement precision is improved, but reliability of object identification deteriorates due to detection errors
Solution Approach 1:
The patent changes the parameter for object domain comparison from strict coordinate coincidence to error domain overlap. By defining error domains that account for detection uncertainties and checking for overlaps rather than exact matches, the system maintains measurement precision while significantly improving object identification reliability.
Solution Approach 2:
The patent applies beforehand cushioning by pre-defining error domains that anticipate detection uncertainties. These error domains act as a buffer zone that accommodates expected detection variations, preventing false negative identifications while maintaining accurate object detection.
3Ease of operation
If the second object domain is defined by X-coordinate range, then ease of operation is improved, but measurement precision of azimuth angle deteriorates
Solution Approach 1:
The patent changes the parameter for defining the second object domain from X-coordinate range to azimuthal range. This parameter change aligns the domain definition with the monocular camera's actual detection characteristics, improving azimuth angle measurement precision while maintaining ease of operation through straightforward angular range specifications.
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 accuracy of object detection by utilizing the strengths of both radar and monocular camera data, reducing false determinations and improving the system's ability to identify objects accurately, especially in scenarios where the cameras' detection points do not coincide perfectly.
Implementation Method 1
a first detection point indicative of a position of a first object detected by using a vehicle-mounted radar
Implementation Method 2
a second detection point indicative of a position of a second object detected on the basis of an image captured by a vehicle-mounted monocular camera
Data Source
AI summary
An object detection apparatus mounted in a vehicle, includes a first domain definition unit, a second domain definition unit, and a determination unit. The first domain definition unit defines a first object domain including a first detection point which is indicative of a position of a first object detected by using a vehicle-mounted radar. The second domain definition unit defines a second object domain including a second detection point which is indicative of a position of a second object detected on the basis of an image captured by a vehicle-mounted monocular camera. The determination unit determines whether or not an overlapping domain of the first and second object domains is present, and when it is determined that an overlapping domain of the first and second object domains is present, then determines that the first and second objects are the same.