Vehicle Peripheral Recognition Text Generation via Object Prioritization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image recognition technologies fail to comprehensively generate text that predicts changes in a vehicle's peripheral situation, focusing primarily on segmental representations rather than overall scenarios.
Innovation Solution
A recognition processing device and method that includes a peripheral situation recognition unit, an object recognition unit, and a text generation unit to identify and prioritize objects influencing the vehicle, generating text that describes the recognized situation using prescribed grammar, with a driving control unit executing steering and acceleration/deceleration based on this text.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If conventional image recognition technology is used to generate text, then segmental word strings representing small parts of the peripheral situation can be obtained, but comprehensive text representing the overall peripheral situation cannot be generated
Solution Approach 1:
The system segments the peripheral situation recognition into distinct functional modules: a peripheral situation recognition unit that identifies objects and their positions, an object recognition unit that selects relevant words, and a text generation unit that constructs comprehensive sentences. This segmentation allows each module to specialize in specific tasks while working together to produce complete peripheral situation descriptions.
Solution Approach 2:
The system merges multiple recognition results and object information into a unified comprehensive text description. The text generation unit integrates selected words from the object recognition unit with grammatical structures to create cohesive sentences that represent the overall peripheral situation, combining fragmented recognition data into meaningful contextual information.
2Loss of information
If all recognized objects are included in the generated text, then comprehensive coverage is achieved, but text length and processing complexity increase
Solution Approach 1:
The object recognition unit extracts only the most relevant objects and their positional relationships from the complete set of recognized objects. It selects specific words that accurately represent critical elements of the peripheral situation, filtering out redundant or less important information to create concise yet comprehensive text descriptions.
Solution Approach 2:
The system applies different levels of detail to different objects based on their relevance to the vehicle. Critical objects such as pedestrians, other vehicles, and obstacles receive more detailed description with specific positional relationships, while less critical elements are described more briefly, optimizing the balance between comprehensiveness and efficiency.
3Manufacturing precision
If text generation follows strict grammar rules, then text quality and readability improve, but generation speed decreases
Solution Approach 1:
The system prepares grammatical structures and sentence templates in advance. The text generation unit has pre-defined grammatical patterns for describing different types of objects and their positional relationships, allowing it to quickly assemble grammatically correct sentences by filling in the template variables with selected object information rather than constructing grammar rules during text generation.
4Reliability
If priority is given to objects with large influence on the vehicle, then driving safety is improved, but object recognition complexity increases
Solution Approach 1:
The system changes the parameter of object selection by introducing a priority criterion based on the object's influence on vehicle operation. Objects are evaluated and ranked according to their potential impact on driving safety and vehicle control, with high-priority objects such as pedestrians, cyclists, and nearby vehicles receiving preferential treatment in text generation and driving control decisions.
Data Source
AI summary
A recognition processing device includes a peripheral situation recognition unit configured to recognize a type of object around a vehicle and a positional relationship with the vehicle, an object recognition unit configured to select a word indicating the type of object recognized by the peripheral situation recognition unit and a word indicating a positional relationship between the vehicle and the object, and a text generation unit configured to generate text for describing a peripheral situation recognized by the peripheral situation recognition unit, wherein the text includes the word indicating the type of object selected by the object recognition unit and the word indicating the positional relationship between the vehicle and the object selected by the object recognition unit.