Active Vision Neural Network for Vehicle Feature Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in adapting to varying operational environments due to the limitations of Deep Neural Networks (DNNs), which rely on passive vision and require significant computational resources and large training datasets, leading to performance deterioration when environmental conditions differ from the training dataset.
Innovation Solution
A low-resolution recurrent neural network with active vision capabilities, allowing for adaptive input selection and processing, is employed to identify features in the vehicle's environment, using a continuous time recurrent neural network architecture with fewer input nodes, reducing computational power and enabling efficient processing on standard hardware.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Deep Neural Networks are used for feature identification, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The image processing task is segmented into multiple passes where the neural network processes only selected regions of interest rather than the entire image. The system divides the visual field into manageable segments, processing only those areas containing potential features, thereby reducing computational complexity while maintaining identification accuracy.
Solution Approach 2:
The system dynamically adjusts the processing strategy based on the input image characteristics. The neural network adaptively determines which regions require detailed analysis and modifies processing parameters in real-time, allowing optimal balance between accuracy and computational efficiency for varying scene complexities.
2Measurement precision
If Deep Neural Networks with large training datasets are used, then measurement precision is improved, but loss of time increases due to retraining requirements
Solution Approach 1:
The neural network is pre-trained on a comprehensive dataset covering diverse environmental conditions and feature types. This preliminary training equips the network with generalizable knowledge that enables accurate feature identification across varying operational environments without requiring retraining, thereby eliminating time loss while maintaining precision.
3Device complexity
If passive vision systems are used, then device complexity is reduced, but adaptability worsens when environmental conditions differ from training data
Solution Approach 1:
The system incorporates feedback mechanisms where the neural network's output and performance metrics are continuously monitored. Based on this feedback, the system adjusts processing parameters, selects appropriate analysis strategies, and adapts to varying environmental conditions in real-time, enabling passive vision hardware to achieve high adaptability without increasing physical system complexity.
Data Source
AI summary
A computer-implemented method and software stack for identifying a feature using active vision is provided. The method is for use in a vehicle for identifying a feature of the environment of the vehicle, and includes: receiving an original image from a sensor or camera; pre-processing the original image to produce an input image; presenting the input image to a neural network; wherein the neural network is trained to classify a feature in an image presented to it, the neural network having an input layer, a hidden layer and an output layer, the output layer including three outputs: a first feedback output for selecting pixels from the input image to input at the input layer at each iteration of the neural network; a second feedback output for selecting a colour channel of the selected pixels to input at the input layer at each iteration; and a third output for outputting an output value indicative of a classification result from the neural network; obtaining the output value from the neural network; and post-processing the output value from the neural network to identify a feature of the environment of a vehicle. The software stack comprises four layers configured to perform this method.