Vehicle Semantic Grid Mapping With Height-Based Free Space Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 2D DST grid maps struggle to accurately represent complex environments, particularly with large objects and obstacles, as they lack semantic categorization and height information, leading to poor localization features and inaccurate free space representation.
Innovation Solution
The method extends DST grid maps by adding layers with semantic meaning, using sensor data fusion from multiple sensors like cameras, radar, and lidar to assign height information and allocate semantic categories such as 'can be driven over' or 'can be driven under' to grid cells, enabling more precise environment representation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If 2D DST grid maps are used for environment representation, then the system is simple and computationally efficient, but the representation accuracy and semantic information are insufficient
Solution Approach 1:
The patent extends traditional 2D DST grid maps by adding a height dimension, transforming them into 3D semantic grid maps. Each grid cell now contains not only occupancy probability but also height information and semantic categories, enabling accurate representation of objects with varying heights such as curbstones, barriers, and vegetation while maintaining the computational efficiency of the grid map structure
Solution Approach 2:
The patent applies different semantic categories to different grid cells based on their local characteristics. Each grid cell is assigned a semantic label (e.g., curbstone, barrier, vegetation, large object) according to its height and properties, allowing the system to treat different regions of the environment differently with appropriate driving behaviors for each semantic type
2Loss of information
If height information is added to grid cells, then the semantic representation improves, but the difficulty of detecting and measuring increases
Solution Approach 1:
The patent employs multiple environment detection sensors (cameras, radar, lidar, ultrasonic sensors) that serve dual purposes: detecting object presence/occupancy and measuring object height. This multi-functional approach allows the system to extract both occupancy probability and height information from the same sensor data without requiring separate measurement systems
Solution Approach 2:
The patent introduces an intermediary processing step where sensor measurements are transformed into height information through data fusion and interpretation. The system uses the measured data from multiple sensors as an intermediary to infer height characteristics, bridging the gap between raw sensor data and semantic height classification
3Reliability
If semantic categories are allocated to grid cells, then the localization features improve, but the device complexity increases
Solution Approach 1:
The patent segments the environment into discrete grid cells, each independently classified with a semantic category. This segmentation approach allows the system to handle complex semantic information in a modular fashion, where each grid cell is processed and categorized independently, simplifying the overall localization and decision-making process
Solution Approach 2:
The patent changes the parameter representation from simple occupancy probability to a composite parameter including occupancy probability, height information, and semantic category. This parameter transformation enables richer localization features while maintaining the grid map structure, as the additional information is integrated into the existing grid cell data model
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 provides a more accurate and detailed semantic representation of the vehicle's environment, enhancing localization and enabling adaptive driving functions and improved obstacle avoidance.
Implementation Method 1
height information can be detected with a stereo camera
Implementation Method 2
a radar sensor
Implementation Method 3
a lidar sensor
Data Source
AI summary
The invention relates to a method for creating a semantic representation of the environment of a vehicle, having the following steps:acquiring (S1) sensor measuring data by means of at least one first (2a) and/or second environment detection sensor (2b) of the vehicle,detecting (S2) objects and free spaces in the environment of the vehicle,creating (S3) a grid map having occupied and free grid cells based on the detected objects and free spaces;determining (S4) height information by means of the at least one first (2a) and/or second environment detection sensor (2b);adding (S5) the height information to each grid cell;allocating (S6) semantic information to the grid cells based on the height information.