Recurrent Neural Network Vehicle Occupancy Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing human detection systems in vehicles, such as those used at border crossings and secure sites, face challenges with false positives and require extensive training due to the use of feedforward neural networks, which are not effective in windy conditions and require numerous experiments to differentiate between vehicle occupant conditions.
Innovation Solution
A system utilizing a neural network with recurrent nodes that can feed its output back into itself or other nodes, allowing for a time delay and increased accuracy in distinguishing between vehicle conditions, using a single vibration sensor to detect vibrations and process signals to indicate the presence or absence of a human.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If feedforward neural networks are used for human detection in vehicles, then the system can process sensor inputs, but the training requires numerous experiments (1,000,000 experiments) and produces false positives in windy conditions
Solution Approach 1:
The patent applies feedback by implementing recurrent nodes that feed their outputs back into the network. This creates a feedback mechanism where previous states influence current processing, allowing the network to learn temporal patterns and reduce false positives without requiring extensive training experiments.
Solution Approach 2:
The patent introduces dynamic elements by using recurrent nodes that maintain state information over time. This transforms the static feedforward network into a dynamic system that can adapt to changing conditions (such as wind) while maintaining detection accuracy, thereby reducing both training time and false positives.
2Adaptability or versatility
If feedforward neural networks are used to differentiate occupant conditions, then the system can classify different scenarios, but it requires extensive training under various experimental conditions
Solution Approach 1:
Recurrent nodes provide feedback loops that allow the network to learn from past classifications and improve future differentiation. This feedback mechanism enables the system to adapt to various occupant conditions more efficiently, reducing the number of training experiments needed while maintaining high differentiation capability.
Solution Approach 2:
The recurrent nodes perform preliminary processing by maintaining state information about previous inputs. This preliminary action allows the network to pre-process temporal patterns before final classification, thereby reducing the overall training burden while preserving the ability to differentiate complex occupant conditions.
3Reliability
If multiple geophones and long testing periods (10-20 seconds) are used to detect heartbeat vibrations, then false negatives are eliminated, but false positives increase particularly in windy conditions
Solution Approach 1:
The recurrent neural network introduces dynamic temporal processing that can distinguish between different vibration patterns over time. This dynamic approach allows the system to maintain high reliability in detecting true occupancy while dynamically adapting to filter out false positives caused by environmental factors like wind, without requiring extended testing periods.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach significantly reduces the number of experiments needed for network training, increases accuracy, and effectively differentiates between occupied and unoccupied vehicle states, even in windy conditions, by mathematically combining inputs and utilizing time delays in the neural network processing.
Implementation Method 1
a vibration sensor configured to detect vibration of the vehicle and to output signals related to the sensed vibrations
Data Source
AI summary
A system for detecting the presence of a human in a vehicle is provided. The system includes a vibration sensor that is configured to detect vibrations of the vehicle, and to output signals related to the sensed vibrations. A processor is configured to receive the signals output from the vibration sensor. The processor also operates a neural network that has a plurality of nodes, at least some of which are recurrent. The use of the recurrent nodes allows the output of a recurrent node to be fed back into itself, or another node. In addition, the output that is fed back can be combined with other inputs entering the node. In this way, the neural network can quickly learn to distinguish between various conditions, including an occupied state and an unoccupied state of the vehicle. The neural network provides an output indicating whether the vehicle is occupied.