Automated Driving Object Classification Without Environment Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automated driving systems require extensive computing resources to interpret object relations in complex environments, necessitating the use of environment maps that are costly and inefficient.
Innovation Solution
A method that utilizes a relational classifier to recognize object relations directly from individual images using geometrical and other relation features, forming aggregation data to determine pairwise object relations without the need for environment maps, and uses an aggregation module to confirm these relations over multiple images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If environment maps are used to interpret object relations, then object relations can be determined, but computing resources and data volume increase significantly
Solution Approach 1:
The patent extracts only the necessary relational information (geometrical relations, lateral relations, longitudinal relations) directly from sensor data without constructing complete environment maps. The relational classifier identifies object relations by analyzing spatial positions and geometrical features of individual objects, taking out only the essential relational data needed for trajectory planning while discarding unnecessary environmental details.
Solution Approach 2:
The patent segments the environment interpretation task into independent object-level analyses. Instead of processing the entire environment map to determine object relations, the system processes each object individually using the relational classifier, which evaluates geometrical relations, lateral relations, and longitudinal relations of each object separately based on sensor data, thereby reducing overall computational complexity.
2Loss of information
If environment maps include all individual objects, then complete environmental information is available, but computing resources are wasted on irrelevant objects
Solution Approach 1:
The patent applies local quality by focusing computational resources only on objects that are locally relevant to the motor vehicle's trajectory planning. The relational classifier evaluates each object's geometrical relation, lateral relation, and longitudinal relation to the vehicle, processing only those objects that have meaningful spatial relationships with the vehicle's path, rather than uniformly processing all detected objects.
Solution Approach 2:
The patent performs partial action by determining only the specific object relations that are necessary for trajectory planning (geometrical relations, lateral relations, longitudinal relations) rather than computing all possible environmental attributes. The system processes sensor data to extract only the relational features needed for safe and efficient driving decisions.
3Measurement precision
If tracking is performed over multiple images, then object properties are deepened, but processing time increases
Solution Approach 1:
The patent applies preliminary action by pre-defining the types of relations to be evaluated (geometrical relations, lateral relations, longitudinal relations) and preparing the relational classifier with these relation types before processing sensor data. This allows the system to efficiently process multiple images by consistently applying the same relational evaluation framework without requiring complex adaptive processing for each image.
Data Source
AI summary
An automated driving function in a motor vehicle comprises: a processor circuit of the motor vehicle recognizes respective individual images of an environment of the motor vehicle from sensor data of a least one sensor of the motor vehicle by means of at least one object classifier. At least one relational classifier using the object data for at least some of the individual objects additionally recognizes a respective pairwise object relation with the aid of predetermined relation features of the individual objects in the respective individual image determined from the sensor data, which relation is described by relational data, and an aggregation module is used to aggregate the relational data throughout multiple consecutive individual images to produce aggregation data, which describe aggregated object relations.