Robert Bosch GmbH (20240247976). METHOD AND SYSTEM FOR VEHICLE DAMAGE DETECTION simplified abstract

METHOD AND SYSTEM FOR VEHICLE DAMAGE DETECTION

Organization Name

Robert Bosch GmbH

Inventor(s)

Tome Silva of Braga (PT)

Afonso Costa of Braga (PT)

Ana Carolina Silva of Braga (PT)

Andre Ferreira of Braga (PT)

Bruno Faria of Braga (PT)

Carolina Goncalves of Braga (PT)

Joao Sa of Braga (PT)

Joao Fernandes of Braga (PT)

Pedro Cunha of Braga (PT)

Pedro Silva of Braga (PT)

Rita Pereira of Braga (PT)

METHOD AND SYSTEM FOR VEHICLE DAMAGE DETECTION - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240247976 titled 'METHOD AND SYSTEM FOR VEHICLE DAMAGE DETECTION

The abstract describes a computer-implemented method for detecting transient events, such as damage or contact events, to a vehicle using sound and vibration signals.

• Acquiring a sound signal over time with an audio sensor and a vibration signal with a motion sensor mounted on the vehicle.
• Detecting if the acquired signals are above predetermined thresholds.
• Converting the signals into an input data record and obtaining an input feature record.
• Using a pretrained machine-learning model to predict transient events based on the input feature record.

Potential Applications: - Vehicle safety and maintenance systems - Insurance claim processing - Fleet management and monitoring

Problems Solved: - Early detection of damage or contact events to vehicles - Improved accuracy in identifying potential issues - Streamlining maintenance processes

Benefits: - Increased safety for vehicle occupants - Cost savings through proactive maintenance - Enhanced efficiency in handling insurance claims

Commercial Applications: "Vehicle Damage Detection System: Enhancing Safety and Efficiency in Fleet Management"

Questions about the technology: 1. How does the machine-learning model differentiate between different types of transient events? 2. What are the potential limitations of using sound and vibration signals for detecting vehicle damage?

Original Abstract Submitted

a computer-implemented method for detecting at least a transient event, which is a damage and/or a contact event to a vehicle. the method includes: acquiring a sound signal over time by an audio sensor mounted on the vehicle to capture air-borne sound waves; acquiring at least a vibration signal over time by a motion sensor mounted on the vehicle to capture vehicle vibration; detecting if the acquired sound signal is above a predetermined sound threshold and/or acquired vibration signal is above a predetermined vibration threshold; converting the acquired sound signal and the vibration signal into an input data record; obtaining an input feature record from the input data record; feeding a pretrained machine-learning model with the input feature record to provide a transient event prediction output, wherein the pretrained model has been pretrained with a training dataset comprising input feature training records and event output training records.