Vehicle Environment Sensor Data Strip Quantization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current driver assistance systems face inefficiencies in representing environmental features, as model-based detection only accounts for objects, while grid-based approaches require high computing power and lack specificity, failing to effectively utilize information about free spaces and unobserved areas.
Innovation Solution
A method using strips in the longitudinal direction of a vehicle for quantization, with continuous transverse position representation, allowing for an efficient and low-computational description of environmental features, retaining neighborhood relationships and reducing memory and computational load.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If grid-based representation with fixed cell size is used, then neighborhood relationship and free space information are preserved, but computing power requirements increase significantly
Solution Approach 1:
The environment is segmented into strips along the longitudinal direction of the motor vehicle, with each strip containing multiple cells. This segmentation allows the system to focus computational resources on relevant areas while maintaining comprehensive environmental coverage, resolving the contradiction between model completeness and computing power requirements
Solution Approach 2:
The patent introduces a longitudinal dimension to the traditional grid representation by organizing cells into strips along the vehicle's direction of travel. This dimensional transformation enables more efficient memory access patterns and reduces the computational complexity of environment modeling while preserving all necessary spatial information
2Loss of information
If model-based detection and tracking is used, then object information is obtained, but information about free spaces and unobserved areas is lost
Solution Approach 1:
The cell-based representation within strips serves multiple functions simultaneously: it tracks objects, identifies free spaces, and represents unobserved areas. Each cell can store occupancy probability and object class information, making the system universal in handling different types of environmental information without requiring separate detection mechanisms
Solution Approach 2:
The patent changes the fundamental parameter representation from object-centric models to cell-centric probability values. Each cell contains an occupancy probability parameter that can represent objects, free spaces, or unobserved areas uniformly, allowing the system to capture all environmental information types through a single parameter change approach
3Reliability
If grid-based representation with fixed cell size is used, then environment coverage is comprehensive, but memory occupancy and data extraction complexity increase
Solution Approach 1:
The environment is segmented into strips along the longitudinal direction of the motor vehicle, with each strip containing multiple cells. This segmentation allows the system to focus computational resources on relevant areas while maintaining comprehensive environmental coverage, resolving the contradiction between model completeness and computing power requirements
Solution Approach 2:
Different strips can have different numbers of cells and different resolution levels based on local requirements. Areas closer to the vehicle or with higher relevance can use finer granularity, while distant or less critical areas can use coarser representation, optimizing memory usage while maintaining comprehensive coverage where needed
Data Source
Figure 1~2
Figure 3A~3C
AI summary
Method for evaluating sensor data relating to the environment of a motor vehicle from at least one environmental sensor, wherein, to describe the position of an environmental feature determined within the evaluation, this feature is assigned to one of several successive strips in the longitudinal direction of the motor vehicle together with at least one indication of its transverse position within the strip perpendicular to the longitudinal direction of the motor vehicle in the road plane.