Autonomous Object Identification Using Pairing Data and Ambient Sound

Solution Overview

Problem

Autonomous driving systems face challenges in achieving high accuracy for object recognition while maintaining complexity at manageable levels, particularly in real-time processing, and often fail to utilize ambient sound information effectively.

Innovation Solution

A data processing method that incorporates sensor data from vehicles, including sound sensors, and utilizes pairing data from external sources to enhance object identification accuracy, allowing for the recognition of objects not detected by primary sensors and adjusting vehicle controls based on ambient sounds.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If more sensor data and deeper learning models are used to improve object recognition accuracy, then recognition accuracy increases, but computational complexity and processing time increase

Engineering Contradiction:

Improveobject recognition accuracyVSAvoidcomputational complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments object recognition into two stages: initial detection using lightweight models on vehicle sensors, and subsequent verification/enhancement using pairing data from external devices. This division allows the system to maintain low computational complexity for real-time processing while achieving high accuracy through collaborative external resources.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces pairing data from external devices (smartphones, other vehicles, infrastructure) as an intermediary resource to enhance object recognition. This external data acts as a mediator that provides additional verification and information without requiring the vehicle's onboard system to perform all computational tasks, thus reducing overall system complexity.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If more sensor data and deeper learning models are used to improve object recognition accuracy, then recognition accuracy increases, but real-time processing becomes difficult

Engineering Contradiction:

Improveobject recognition accuracyVSAvoidreal-time processing capability

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The patent segments object recognition into two stages: initial detection using lightweight models on vehicle sensors, and subsequent verification/enhancement using pairing data from external devices. This division allows the system to maintain low computational complexity for real-time processing while achieving high accuracy through collaborative external resources.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary object detection using lightweight onboard models before seeking pairing data. This preliminary action ensures that critical real-time detection occurs quickly, while more computationally intensive verification can occur subsequently or in parallel, maintaining real-time responsiveness for safety-critical functions.

Inventive Principle:

Principle #10Preliminary action

3Device complexity

If traditional sensor-only approaches are used, then system complexity is low, but objects not detected by primary sensors cannot be identified

Engineering Contradiction:

Improvesystem complexityVSAvoidundetected objects

Core Design Contradiction:

Device complexityVSLoss of information

Solution Approach 1:

The patent introduces pairing data from external devices (smartphones, other vehicles, infrastructure) as an intermediary resource to enhance object recognition. This external data acts as a mediator that provides additional verification and information without requiring the vehicle's onboard system to perform all computational tasks, thus reducing overall system complexity.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent enables multiple device types (vehicle sensors, smartphones, other vehicles, infrastructure) to contribute to object detection through a universal pairing data framework. This multi-functional approach allows any device in the ecosystem to contribute detection capabilities, reducing information loss without requiring every device to have complex sensor suites.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS20230333557A1Data processing method for object detection and identification and autonomous deriving device therefor

Publication Date: 2023.10.19 HONG JOONYOUNG

AI summary

A method for controlling a vehicle based on object identification for autonomous driving according to the disclosure of this document includes obtaining sensor data based on sensors positioned on a vehicle, performing object identification based on a result of applying the sensor data to a machine learning model, adjusting a control parameter of the vehicle based on a result of the object identification, wherein performing the object identification comprises receiving pairing data through a network, wherein the object identification is performed further based on the pairing data. Based on this, it is possible to increase the accuracy of object identification in the blind area.