Vehicle Control Unit Environment Data Segmentation for Parallel Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current vehicle systems face challenges in performing a reliable and precise analysis of driving situations due to the need for real-time processing, which often results in a reduction of environmental data volume, leading to qualitatively poorer results.

Innovation Solution

The method involves breaking down environmental data into temporally and logically independent segment records, allowing for parallel processing across multiple calculations, utilizing a graphics processor to maintain high processing detail and data volume, thereby enabling detailed analysis without data abstraction.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If environmental data is reduced to a subset through preselection to enable fast calculation, then real-time processing capability is improved, but measurement precision and analysis quality deteriorate

Engineering Contradiction:
Improvecalculation speedVSAvoiddriving situation analysis quality
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The evaluation process is divided into multiple independent evaluation steps that can be executed in parallel. Each step processes a specific aspect of the environmental data (e.g., obstacle detection, trajectory prediction, risk assessment) simultaneously, allowing full data to be analyzed without reduction while maintaining real-time processing capability through parallel computation architecture

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transitions from sequential single-threaded processing to parallel multi-threaded processing architecture. By utilizing graphics processing units (GPUs) or multi-core processors, the system evaluates environmental data across multiple computational dimensions simultaneously, achieving both high processing detail and real-time performance without data reduction

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Measurement precision

If processing detail is increased to improve analysis quality, then measurement precision is improved, but calculation time increases and real-time capability is lost

Engineering Contradiction:
Improvedriving situation analysis qualityVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The complex evaluation process is segmented into multiple independent evaluation steps (e.g., environmental perception, obstacle identification, trajectory prediction, risk assessment). Each segment processes specific aspects of environmental data in parallel, maintaining high processing detail while reducing overall processing time through simultaneous execution of multiple evaluation threads

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system creates multiple parallel copies of the evaluation process, each handling different aspects or portions of the environmental data simultaneously. This parallel copying approach allows comprehensive analysis of all environmental data without sequential processing delays, achieving both high detail and real-time performance

Inventive Principle:
Principle #26Copying

Data Source

PatentEP3433152B1Method for evaluating environment data in a control unit of a motor vehicle, control unit, and motor vehicle
Publication Date: 2022.09.28 AUDI AG
  • EP3433152B1 patent drawingFigure 1
  • EP3433152B1 patent drawingFigure 2~3

AI summary

The invention relates to a method for evaluating environment data, which are at least partially derived from sensor data of at least one environment sensor (8), in a control unit of a motor vehicle (7), wherein at least one evaluation step uses an environment data set (13) as input data, wherein the environment data set (13) is broken down into temporally and logically independent segment data sets (5), which are evaluated in the evaluation step at least partially in temporally parallelized partial steps (6, S3) by using a parallel processing architecture (11) which, in particular, permits more than 100 temporally parallel computations.