Crane Collision Avoidance With Neural Network Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for ensuring collision-free movement of cranes in automated loading processes lack reliability in detecting obstacles such as persons and objects, particularly in varying conditions like day and night times and weather, leading to potential collisions.
Innovation Solution
A method utilizing a first neural network trained with raw data from crane operations outside normal conditions, combined with a second neural network for object detection, to identify anomalies and obstacles during crane operation, using optical sensors and a central IT infrastructure for data processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional obstacle detection methods are used during crane operations, then the system can detect some obstacles, but the detection reliability is insufficient particularly in varying conditions like day and night times and weather
Solution Approach 1:
The system performs preliminary training of neural networks using data collected during crane operations under various conditions (day/night, different weather). This preliminary training enables the system to adapt to varying conditions before actual obstacle detection is needed, improving both reliability and adaptability simultaneously
Solution Approach 2:
The system changes parameters by using multiple neural networks with different training data sets corresponding to different environmental conditions. Each neural network is optimized for specific parameters (time of day, weather conditions), allowing the system to maintain high detection reliability across varying conditions
2Reliability
If neural networks are trained with comprehensive training data sets to improve detection accuracy, then obstacle detection reliability improves, but data processing time and system complexity increase
Solution Approach 1:
The system segments the training data into multiple data sets, each corresponding to different environmental conditions (daytime, nighttime, various weather). Multiple specialized neural networks are trained on these segmented data sets, allowing for faster and more accurate detection in specific conditions compared to a single large network processing all data
Solution Approach 2:
The neural networks are trained in advance with comprehensive training data sets covering various conditions. This preliminary training action is performed during system setup or maintenance periods, so that during actual crane operations, the pre-trained networks can perform rapid anomaly detection without processing delays
Data Source
AI summary
In a method for a collision-free movement of a crane in a crane lane, a sensor captures a first training data set of raw data during a movement of the crane outside a crane operation in the crane lane. The first training data set is evaluated while teaching a first neural network based on the captured raw data and first training data are determined from the evaluated first training data set. The sensor captures current sensor data during a movement of the crane during the crane operation in the crane lane. The current sensor data is compared with the first training data, and an anomaly is detected between the current sensor data and the first training data.