Micron have a newly listed patent titled:
Predictive Maintenance of Automotive Transmission
Patent: 20210192860
I find it particularly interesting because not only do they describe using multiple sensor types such as vibration and temperature sensors, but they also talk about using spiking neural networks to detect patterns from the sensor outputs. The spiking neural networks seem to be specifically used and not mentioned like an afterthought. Predictive maintenance is a key application that Brainchip has regularly been talking about and Micron is also one of the companies Brainchip was allegedly talking to. Plus a year and a half ago would be a reasonable time for EAP customers to have begun engaging with Brainchip.
@BarrelSitter, can you let the ogre loose and see what it thinks? I've had a look and haven't seen anything that would rule out Akida being used as the SNN.
There was another patent filed the same day from Micron (Patent 20190190913) which spoke about using spiking neural networks for intelligent radar in autonomous vehicles, but it talked about using memristors in the crossbar array. So I'm not as convinced with that one.
AbstractSystems, methods and apparatuses of predictive maintenance of automotive transmission of vehicles. For example, the transmission has at least one sensor to measure a temperature in transmission fluid, the torque applied on a shaft of the transmission, a vibration sensor, and/or a microphone. During a period in which the vehicle is assumed to be operating normally, the sensor data generated by the transmission sensor(s) is used to train an artificialneuralnetwork to recognize the normal patterns in the sensor data. Subsequently, the trained artificialneuralnetwork is used to determine whether the current sensor data from the transmission sensor(s) are abnormal. A maintenance alert can be generated for the vehicle in response to a determination that the operations of the transmission are abnormal according to the artificialneuralnetwork and the current sensor data.
1. A vehicle, comprising: an automotive transmission; at least one sensor configured on the automotive transmission to generate sensor data; a computer system configured to determine operating parameters of the vehicle, the computer system having: one or more memory components configured to store the sensor data; and an inference engine configured to determine whether operations of the automotive transmission are abnormal, using an artificialneuralnetwork, the sensor data, and the operating parameters; wherein the computer system is configured to generate a maintenance alert in response to a determination that the operations of the automotive transmission are abnormal according to the artificialneuralnetwork.2. The vehicle of claim 1, wherein the artificialneuralnetwork includes aspiking neuralnetwork.3. The vehicle of claim 2, wherein the computer system includes a data storage device having the inference engine and a host interface configured to receive the sensor data; and wherein the data storage device is further configured to generate training data and train thespiking neuralnetwork to determine whether the operations of the automotive transmission are abnormal.4. The vehicle of claim 1, wherein the at least one sensor includes a temperature sensor configured in transmission fluid of the automotive transmission.5. The vehicle of claim 1, wherein the at least one sensor includes a torque sensor configured on a shaft of the automotive transmission.6. The vehicle of claim 1, wherein the at least one sensor includes a first torque sensor configured on an input shaft of the automotive transmission and a second torque sensor configured on an output shaft of the automotive transmission.7. The vehicle of claim 1, wherein the at least one sensor includes a vibration sensor configured in transmission fluid of the automotive transmission.https://appft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&u=%2Fnetahtml%2FPTO%2Fsearch-adv.html&r=1&f=G&l=50&d=PG01&p=1&S1=(Spiking+AND+neural)&OS=Spiking+AND+neural&RS=(Spiking+AND+neural)
(20min delay)