Convolution Kernel Parallel Scanning for ADAS Image Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Advanced driver assistance systems (ADAS) and autonomous vehicle (AV) systems face challenges in effectively processing and protecting data generated by object classifiers, which are crucial for detecting objects and controlling vehicles or alerting drivers, especially as these systems progress towards fully autonomous operation.
Innovation Solution
The system employs a combination of cameras and processors, including neural networks or deep learning algorithms, to process images and estimate a vehicle's future path, integrating multiple image capture devices and sensors like cameras, GPS, and radar systems for enhanced object detection and vehicle control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If neural networks or deep learning algorithms are used to process images and estimate vehicle paths, then object detection accuracy and vehicle control are improved, but data processing complexity and computational requirements increase
Solution Approach 1:
The patent segments the convolution operation into multiple processing stages: separating the convolution kernel application from the pooling operation, and dividing the neural network processing into distinct layers (convolutional layers, pooling layers). This segmentation allows each component to be optimized independently, reducing overall processing complexity while maintaining detection accuracy.
Solution Approach 2:
The patent applies preliminary actions by performing convolution operations with multiple kernels in advance before the pooling operation. The convolution kernels are pre-configured with specific parameters (depth, channels, padding) to extract relevant features from input data before downsampling, which simplifies the subsequent pooling stage and improves overall efficiency.
2Reliability
If multiple image capture devices and sensors are integrated for enhanced object detection, then detection reliability is improved, but system complexity and data protection requirements increase
Solution Approach 1:
The patent merges data from multiple input sources (different input channels representing different sensors or image capture devices) into a unified neural network processing framework. The convolution operation processes data across multiple channels simultaneously, combining information from various sensors to improve detection reliability while managing system complexity through integrated processing.
Solution Approach 2:
The neural network architecture designed in the patent is universal and can process data from multiple types of sensors and input devices through the same convolution and pooling operations. The system handles diverse input data types (images, sensor readings) using a common processing framework, reducing the need for separate processing pipelines for each sensor type.
Data Source
AI summary
An improved convolution kernel system and method may be used to improve performance of analysis of input image data for autonomous or semi-autonomous vehicle navigation. A processing circuit may be used to apply a convolution kernel on the input data to provide output data that comprises output data segments. The application may include repeating scanning and summing, including parallel scanning input data segments of the different input channels and of the input data depth value to provide first intermediate results associated with the input data depth value, and summing first intermediate results associated with a same output data channel and with different input depth values to provide, per each output data channel, a second result. The output analyzed image data may be used to generate a vehicle control signal, such as automatic control of braking, acceleration, or steering of a vehicle.