Ultrasonic Sensor Array Obstacle Detection Neural Network
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Unmanned vehicles equipped with ultrasonic sensors face low detection precision and are prone to detection errors or omissions due to the working principles of ultrasonic radars, which affect their obstacle-avoiding capabilities.
Innovation Solution
A method and system that utilize a pre-trained neural network model, specifically a convolutional neural network or Long Short-Term Memory (LSTM) neural network, to process obstacle coordinates from ultrasonic sensor arrays by comparing them with LiDAR data, generating true or false labels, and improving the accuracy of obstacle detection by selecting feature values from adjacent sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If ultrasonic sensors are used for obstacle detection in unmanned vehicles, then the obstacle-avoiding function can be achieved, but the detection precision is low and detection errors or omissions occur
Solution Approach 1:
The patent combines multiple ultrasonic sensors into an array configuration, where multiple sensors work together to detect obstacles. By merging the detection capabilities of multiple sensors and processing their combined data, the system achieves higher detection precision and reliability compared to single sensor operation, while maintaining cost-effectiveness.
Solution Approach 2:
The patent introduces a neural network model as an intermediary between raw sensor data and obstacle detection results. This intermediary processes the coordinate data from ultrasonic sensors, identifies false detections through pattern recognition, and outputs corrected obstacle information, thereby improving detection precision without requiring expensive alternative sensors.
2Measurement precision
If a neural network model is used to process obstacle coordinates, then detection accuracy is improved, but system complexity increases
Solution Approach 1:
The neural network model is pre-trained offline using labeled training samples before deployment. During actual obstacle detection, the pre-trained model directly processes sensor coordinates without requiring real-time complex computations. This preliminary preparation transfers computational burden from runtime to training time, reducing real-time system complexity while maintaining high detection accuracy.
Solution Approach 2:
The patent uses a simplified neural network architecture that copies successful patterns from training data rather than implementing complex physical models. The network learns to recognize detection patterns and false alarms by copying statistical relationships from training samples, enabling accurate detection with a relatively simple network structure that reduces computational requirements.
Data Source
AI summary
The present disclosure provides a method and system for processing an obstacle detection result of an ultrasonic sensor array. The method comprises: obtaining obstacle coordinates collected by ultrasonic sensors in an ultrasonic sensor array; inputting the obstacle coordinates collected by ultrasonic sensors in the ultrasonic sensor array into a pre-trained neural network model, to obtain true or false labels for the obstacle coordinates collected by the ultrasonic sensors output by the pre-trained neural network model; process the obstacle coordinates collected by the ultrasonic sensors in the ultrasonic sensor array according to the true or false labels. The present disclosure improves the accuracy in the obstacle detection of the ultrasonic sensor array, avoids detection error and omission and improves the driving safety.