Occupancy Grid Segmentation Using Speed Feature Graphs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Occupancy grids in driver assistance systems struggle to effectively represent and analyze dynamic obstacles, as they primarily rely on position data without efficiently incorporating speed and other object-related information, leading to artifacts in speed estimation and limited object classification.
Innovation Solution
The method involves transforming grid cells into a feature space that includes speed information, constructing a graph with proximity relationships, and using efficient graph segmentation algorithms to assign objects to grid cells, thereby enhancing object representation and reducing over- or under-segmentation issues.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If occupancy grids rely primarily on position data without incorporating speed information, then the grid structure remains simple and easy to process, but the accuracy of dynamic obstacle representation and classification deteriorates
Solution Approach 1:
The patent extends the traditional 2D occupancy grid by incorporating speed information as an additional dimension, creating a 4D occupancy grid (x, y, vx, vy). This dimensional expansion allows the system to represent dynamic obstacles more accurately by encoding both position and velocity information, directly resolving the contradiction between maintaining simple grid structure and improving object classification accuracy
Solution Approach 2:
The patent changes the parameter representation within grid cells from simple occupancy indicators to multi-parameter descriptors including speed components (vx, vy) and object classification labels. This parameter enrichment enables the grid to capture dynamic characteristics of obstacles, improving measurement precision while the systematic organization of these parameters prevents excessive complexity
2Measurement precision
If occupancy grids use basic segmentation without graph-based methods, then the processing speed remains high, but the segmentation accuracy and object representation quality deteriorates
Solution Approach 1:
The patent applies graph-based segmentation algorithms to divide the occupancy grid into distinct object regions based on proximity relationships and feature similarities. By constructing graphs where nodes represent grid cells and edges represent spatial relationships, the system achieves accurate object segmentation while using efficient algorithms to maintain processing speed
Solution Approach 2:
The patent transforms grid cell data into feature space using speed and position information, then applies graph segmentation based on these transformed features. This parameter transformation enables more accurate segmentation by capturing dynamic characteristics, while the graph-based approach provides computationally efficient processing through optimized algorithms
3Loss of information
If occupancy grids incorporate multiple object information items, then the object representation quality improves, but the computational complexity and processing time deteriorates
Solution Approach 1:
The patent extracts and separates different object information items (position, speed, classification) into distinct components within the occupancy grid framework. By organizing these information items systematically and using graph-based methods to process only relevant relationships, the system maintains complete object information while reducing unnecessary computational overhead
Solution Approach 2:
The patent transforms multiple object parameters into a unified feature representation that captures essential characteristics for segmentation. By changing the parameter space and using graph-based algorithms that operate on these transformed features, the system processes comprehensive object information efficiently, minimizing processing time while preserving information completeness
Data Source
AI summary
A method is described for segmenting an occupancy grid for a surroundings model of a driver assistance system for a vehicle. The method encompasses a step of reading in an occupancy grid having a plurality of grid cells, each grid cell of the plurality of grid cells having assigned to it a grid cell information item that encompasses an information item about a degree of occupancy and at least one additional information item about the grid cell; and a step of assigning at least one object and/or a plurality of objects to the plurality of grid cells using the grid cell information item, in order to segment the occupancy grid for a surroundings model.