Driving Action Classification via Tendency Symbolization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional techniques cannot collect information about potential dangerous driving situations unless a specific event has occurred, failing to provide warnings for areas where attention is needed due to factors like poor visibility or high parking density, even if no past events have happened.
Innovation Solution
A driving action classifying apparatus that uses sensor data from vehicles to generate symbols representing driving actions, which are then converted into 'tendency symbols' to identify patterns and alert drivers to areas requiring attention, such as changes in driving behavior or divergent actions from the norm.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If conventional techniques detect dangerous events to collect safety information, then information about actual dangerous events can be collected, but information about potential hazards without past events cannot be collected
Solution Approach 1:
The system performs preliminary classification of driving actions into symbols before dangerous events occur. By continuously collecting and typifying normal driving actions at various locations, the system builds a baseline of expected behavior patterns. This preliminary data collection enables the system to identify potential hazards by detecting deviations from normal patterns, rather than waiting for actual dangerous events to occur.
Solution Approach 2:
The patent introduces driving action symbols as an intermediary representation layer between raw sensor data and safety information. These symbols serve as a mediator that captures essential driving behavior characteristics in a standardized format, enabling comparison across multiple vehicles and locations. This intermediary representation allows the system to identify patterns and potential hazards without requiring direct detection of dangerous events.
2Loss of information
If the system monitors all driving actions to identify patterns, then potential hazards can be detected, but the complexity of data processing increases
Solution Approach 1:
The system extracts only the essential characteristics of driving actions by converting raw sensor data into simplified driving action symbols. This extraction process removes unnecessary details while preserving the core behavioral patterns needed for hazard identification. By taking out only the relevant features and representing them as discrete symbols, the system reduces data complexity while maintaining the ability to detect potential hazards.
Solution Approach 2:
The patent transforms continuous sensor data into discrete symbolic representations through parameter changes. This transformation converts complex continuous measurements into simplified categorical symbols that are easier to process and compare. By changing the parameter representation from continuous to discrete, the system reduces computational complexity while preserving the essential information needed for pattern recognition.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A driving action classifying apparatus comprises a driving-action-symbol acquiring unit configured to acquire position information on a vehicle and driving action symbols, which are data obtained by converting driving actions of the vehicle into symbols; and a tendency symbolizing unit configured to collect the driving action symbols corresponding to a same or similar place acquired from a plurality of vehicles and generate driving tendency symbols, which are data obtained by converting into a symbol a frequency distribution of the driving action symbols.