Methods and systems for monitoring driving automation
The present disclosure provides methods and systems for monitoring a driving automation system of a vehicle. An example system comprises a vehicle interface for connecting to a forward-facing camera mounted on the vehicle to receive video data, a microphone for recording audio signals from inside the vehicle, an internal measurement unit (IMU) for generating vehicle motion signals comprising at least a lateral acceleration signal, a longitudinal acceleration signal, and a yaw angular acceleration signal, and a processor connected to process the video data, audio signals and vehicle motion signals to determine whether the driving automation signal is engaged or disengaged.
1 . A system for monitoring a driving automation system of a vehicle, the system comprising:
a microphone for recording audio signals from inside the vehicle;
a vehicle interface for connecting to a vehicle camera controller to receive video data from a forward-facing camera mounted on the vehicle;
at least one inertial measurement unit (IMU) for generating vehicle motion signals comprising at least a lateral acceleration signal, a longitudinal acceleration signal, and a yaw angular acceleration signal;
a processor connected to receive the audio signals, video data and vehicle motion signals; and
a memory accessible to the processor,
wherein the processor is configured to:
generate driving mode determination signals indicative of whether the driving automation system is engaged or disengaged based on the audio signals, the video data, and the vehicle motion signals by:
continuously sampling the audio signals to obtain a plurality of audio samples;
filtering and transforming the audio samples to obtain an audio spectrogram for each audio sample;
comparing the audio spectrogram of each audio sample to a plurality of chime spectrograms;
when the audio spectrogram matches an engagement chime spectrogram, recording a start time of the audio sample for that spectrogram as a driving automation engagement time; and
when the audio spectrogram matches a disengagement chime spectrogram, recording a start time of the audio sample for that spectrogram as a driving automation disengagement time;
store the driving mode determination signals, the video data, and the vehicle motion signals in the memory; and
export the driving mode determination signals, the video data, and the vehicle motion signals to one or more remote systems for further processing.
2 . The system of claim 1 wherein the memory comprises a first partition for storing video data received from the forward-facing camera and a second partition for storing the vehicle motion signals and the driving mode determination signals, and the first partition is accessible by the processor and the vehicle camera controller, and the second partition is only accessible by the processor.
3 . The system of claim 1 wherein the processor is configured to discard audio signals after generating a corresponding driving mode determination signal.
4 . The system of claim 1 wherein the vehicle interface comprises a USB connector for insertion into a USB port on the vehicle.
5 . The system of claim 1 wherein the processor is configured to:
process the video data to generate a lane-center signal;
continuously sampling each of the lateral acceleration, longitudinal acceleration, yaw angular acceleration and lane-center signals to obtain a plurality of sample sets;
feed each sample set into a recurrent neural network (RNN) to determine whether the sample set matches a driving automation control signature;
when the sample set matches the driving automation control signature, determine that the driving automation system is engaged; and
when the sample set does not match the driving automation control signature, determine that the driving automation system is disengaged.
6 . A method for monitoring a driving automation system of a vehicle, the method comprising:
receiving audio signals from one or more microphones in the vehicle;
continuously sampling the audio signals to obtain a plurality of audio samples;
filtering and transforming the audio samples to obtain an audio spectrogram for each audio sample;
comparing the audio spectrogram of each audio sample to a plurality of chime spectrograms;
when the audio spectrogram matches an engagement chime spectrogram, recording a start time of the audio sample for that spectrogram as a driving automation engagement time;
when the audio spectrogram matches a disengagement chime spectrogram, recording a start time of the audio sample for that spectrogram as a driving automation disengagement time;
recording associated vehicle data corresponding to each driving automation engagement time and each driving automation disengagement time to generate a plurality of driving automation scenarios; and
exporting the plurality of driving automation scenarios to one or more remote systems for further processing.
7 . The method of claim 6 comprising discarding each audio sample after comparing its audio spectrogram to the plurality of chime spectrograms.
8 . The method of claim 6 wherein recording associated vehicle data comprises recording image data corresponding to the driving automation engagement time or driving automation disengagement time.
9 . The method of claim 6 wherein recording associated vehicle data comprises recording vehicle motion signals corresponding to the driving automation engagement time or driving automation disengagement time.
10 . The method of claim 6 wherein recording associated vehicle data comprises recording image data and vehicle motion signals corresponding to the driving automation engagement time or driving automation disengagement time.
11 . A method for monitoring a driving automation system of a vehicle, the method comprising:
receiving vehicle motion signals comprising a lateral acceleration signal, a longitudinal acceleration signal, and a yaw angular acceleration signal from one or more inertial measurement units (IMUs) mounted on the vehicle;
receiving image data from a forward facing camera mounted on the vehicle and processing the image data to generate a lane-center signal;
continuously sampling each of the lateral acceleration, longitudinal acceleration, yaw angular acceleration and lane-center signals to obtain a plurality of sample sets;
feeding each sample set into a recurrent neural network (RNN) to determine whether the sample set matches a driving automation control signature;
when the sample set matches the driving automation control signature, determining that the driving automation system is engaged;
when the sample set does not match the driving automation control signature, determining that the driving automation system is disengaged;
storing each sample set that matches the driving automation control signature with corresponding image data to generate a plurality of driving automation scenarios; and
exporting the plurality of driving automation scenarios to one or more remote systems for further processing.