AI diagnoses a vehicle by its sound
V2M in the US has developed AI-based acoustic sensor monitoring to detect faults in all kinds of vehicles for preventive maintenance.
The company has installed a prototype acoustic sensor system on a Tesla Model 3 and is developing a scoring model that will be able to predict potential malfunctions by the sound of the vehicle.
The next step for V2M, based in New York, is to purchase two vehicles with internal combustion engines and one hybrid vehicle for product testing to show that the product is equally suitable for every type of car/engine.
Data shows that 27% of malfunctions that led to accidents could be prevented if the noise car was making were noticed and recognized on time.
- Infineon leads EU project developing future ‘digital ear’
- Sensata partners with Nanoprecise on AI driven predictive maintenance
- Improving uptime in industrial robotic systems
“In the past I was a racing driver. In the summer of 2019 in front of my eyes the car crashed and burned (I was after two cars behind). Unfortunately, the driver died. His car had a malfunction with the rear axle. It’s a common case and for sure it had a noisy before finally jamming. Our solution could have saved his life,” said Peter Bakulov, CEO of V2M.
The ultimate aim is to work with Tier 1 automakers to create a one-time implementation into the vehicle as a part of the standard equipment.
The V2M team has developed a methodology based on a multilayer neural network that determines the presence of patterns of faulty sound and concludes which malfunction is on board. The system recognizes malfunction sounds even when the car is on the road loaded with noise.
One MEMS microphone is located in front of the car, the second in the back and both sensors are connected to the multiplexer board via an I2S link. A third sensor in a middle of the car. The software detects the sound and concludes what type of malfunction is occurring with 98% accuracy within 2s.
The system then notifies the driver (or somebody else, it depends of the client) about the malfunction and completes a report.
Initially the system is connected to the cloud via an eSIM but V2M also could be 100% on-board system without any kind of M2M communications.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
