Automated Driving Recognition System Using Sensor Fusion for Latency Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image recognition processing systems for automated driving face challenges with high computational load and power consumption, leading to increased frame delay and latency, making them unsuitable for small-scale systems like vehicles.
Innovation Solution
A recognition processing system that includes a camera for imaging, a radar for object detection, and a selection unit that chooses the appropriate recognition process based on detected object type, allowing for efficient preprocessing and recognition using specialized recognizers or programs, thereby reducing computational load and accelerating processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If comprehensive image recognition processing is performed on all captured images, then recognition accuracy is improved, but computational load and power consumption increase significantly
Solution Approach 1:
The patent extracts and processes only the regions containing detected objects rather than performing comprehensive recognition on the entire image. The object detection unit identifies candidate regions, and the recognition processing unit focuses computational resources only on these extracted regions, thereby maintaining recognition accuracy while significantly reducing power consumption.
Solution Approach 2:
The patent segments the image processing task into two stages: first, object detection to identify candidate regions; second, detailed recognition only within those segmented regions. This segmentation allows the system to achieve comprehensive recognition accuracy while concentrating computational power only where needed, reducing overall energy consumption.
2Measurement precision
If comprehensive image recognition processing is performed on all captured images, then recognition accuracy is improved, but processing time and latency increase
Solution Approach 1:
The patent extracts only the necessary regions containing objects for detailed recognition, avoiding processing of empty or irrelevant areas. This extraction approach maintains comprehensive recognition accuracy while significantly reducing the time required for processing, as the recognition unit focuses only on relevant image portions.
Solution Approach 2:
The patent segments the processing workflow into rapid object detection followed by focused recognition on detected regions. This segmentation enables the system to achieve thorough recognition accuracy while minimizing processing latency by avoiding unnecessary computation on entire images.
3Measurement precision
If high-performance recognition processing is implemented, then recognition capability is improved, but system size and complexity increase
Solution Approach 1:
The patent extracts only the essential processing components needed for effective recognition by focusing computation on detected object regions rather than entire images. This approach maintains high recognition capability while simplifying the system architecture by eliminating redundant processing operations.
Solution Approach 2:
The patent segments the recognition system into a lightweight detection component and a focused recognition component that operates only on detected regions. This segmentation reduces overall system complexity while preserving recognition capability, as each component performs its specialized function efficiently.
Data Source
AI summary
To perform high-speed recognition processing of an image signal acquired by imaging.A recognition processing system of the present disclosure includes: a first sensor device that acquires an image signal by imaging; a second sensor device that performs an object detection process, a selection unit that selects one of a plurality of recognition processes based on information on the object detected by the detection process; and a recognition processing unit that executes the recognition process selected by the selection unit based on the image signal.