Vehicle mode determination based on edge-computing
Methods, computer-readable media, software, and apparatuses may determine, based upon edge-computing operations, that a vehicular trip has been initiated and cause one or more sensors to collect vehicle data. One or more trip segments for at least a portion of the vehicular trip may be determined. In some aspects, for each trip segment, a first plurality of time features and a second plurality of frequency features may be determined, and may be concatenated with a third plurality of GPS features to form a feature vector. An accuracy measure may be determined based on the feature vector, and a mode for the vehicle may be predicted.
1. A method comprising:
determining, by an on-board mobile computing device, whether a vehicular trip has been initiated;
based on a determination that the vehicular trip has been initiated, causing, by the mobile computing device, one or more sensors to collect vehicle data, wherein the vehicle data comprises at least one of: accelerometer data, gyroscope data, and GPS data;
determining, by the mobile computing device, one or more trip segments for at least a portion of the vehicular trip;
determining, by the mobile computing device and based on the accelerometer data and the gyroscope data, one or more time domain features and one or more frequency domain features for each trip segment of the one or more trip segments;
generating, by the mobile computing device and by concatenating the one or more time domain features and the one or more frequency domain features with one or more GPS features, a feature vector;
determining, by the mobile computing device and based on the feature vector, an accuracy measure;
predicting, based on the accuracy measure, a mode for the vehicle, wherein the mode for the vehicle includes a type of vehicle; and
ceasing the one or more sensors from collecting the vehicle data upon determining the mode for the vehicle can be predicted.
2. The method of claim 1 , further comprising:
determining a physical orientation of the on-board mobile device; and
adjusting the accelerometer data based on the physical orientation.
3. The method of claim 1 , further comprising:
adjusting the gyroscope data by removing drift bias.
4. The method of claim 1 , wherein generating the feature vector further comprises:
reducing, based on a principal component analysis of the feature vector, a number of components of the feature vector.
5. The method of claim 1 , further comprising:
training, based on a support vector machine (SVM) classifier with a linear kernel, a machine learning model.
6. The method of claim 5 , wherein generating the feature vector further comprises:
determining, based on the machine learning model, a respective weight for one or more components of the feature vector.
7. The method of claim 1 , further comprising:
determining a threshold for a probability measure associated with the mode; and
determining the mode by comparing the probability measure to the threshold.
8. The method of claim 1 , further comprising:
determining a stop time after the initiation of the vehicular trip, wherein the stop time is indicative of a time when the predicting for the mode may occur.
9. The method of claim 1 wherein predicting the mode further comprises:
performing a multimodal prediction.
10. The method of claim 1 , further comprising:
after prediction of the mode, determine whether the on-board mobile computing device is associated with a driver or a passenger of the vehicle.
11. The method of claim 1 , further comprising:
determining a number of features to be used for the predicting of the mode.
12. The method of claim 1 , further comprising:
determining, for each vehicle mode, a number of trips to be used for the predicting of the mode.
13. An apparatus, comprising:
a processor;
a memory unit storing computer-executable instructions, which when executed by the processor, cause the apparatus to:
determine, by an on-board mobile computing device, whether a vehicular trip has been initiated;
based on a determination that the vehicular trip has been initiated, cause, by the mobile computing device, one or more sensors to collect vehicle data, wherein the vehicle data comprises at least one of: accelerometer data, gyroscope data, and GPS data;
determine, by the mobile computing device, one or more trip segments for at least a portion of the vehicular trip;
determine, by the mobile computing device and based on the accelerometer data and the gyroscope data, one or more time domain features and one or more frequency domain features for each trip segment of the one or more trip segments;
generate, by the mobile computing device and by concatenating the one or more time domain features and the one or more frequency domain features with one or more GPS features, a concatenated vector;
reduce, based on a principal component analysis of the concatenated vector, a number of components of the concatenated vector to determine a feature vector;
determine, by the mobile computing device and based on the feature vector, an accuracy measure; and
predict, based on the accuracy measure, a mode for the vehicle.
14. The apparatus of claim 13 , wherein the computer-executable instructions, when executed by the processor, further cause the apparatus to:
after prediction of the vehicle mode, determine whether the on-board mobile computing device is associated with a driver or a passenger of the vehicle.
15. The apparatus of claim 13 , wherein the computer-executable instructions, when executed by the processor, further cause the apparatus to:
determine a number of features to be used for the predicting of the vehicle mode.
16. The apparatus of claim 13 , wherein the computer-executable instructions, when executed by the processor, cause the apparatus to:
train, based on a support vector machine (SVM) classifier with a linear kernel, a machine learning model.
17. The apparatus of claim 13 , wherein the computer-executable instructions, when executed by the processor, cause the apparatus to:
determine a respective weight for one or more components of the feature vector.
18. The apparatus of claim 13 , wherein the computer-executable instructions, when executed by the processor, cause the apparatus to:
determine a stop time after the initiation of the vehicular trip, wherein the stop time is indicative of a time when the predicting for the mode may occur.
19. One or more non-transitory computer-readable media storing instructions that, when executed by a computing device, cause the computing device to:
determine, by an on-board mobile computing device, whether a vehicular trip has been initiated;
based on a determination that the vehicular trip has been initiated, cause, by the mobile computing device, one or more sensors to collect vehicle data, wherein the vehicle data comprises at least one of: accelerometer data, gyroscope data, and GPS data;
determine, by the mobile computing device, one or more trip segments for at least a portion of the vehicular trip;
determine, by the mobile computing device and based on the accelerometer data and the gyroscope data, one or more time domain features and one or more frequency domain features for each trip segment of the one or more trip segments;
generate, by the mobile computing device and by concatenating the time domain features and the frequency domain features with one or more GPS features, a feature vector;
determine, based on a machine learning model, a respective weight for one or more components of the feature vector;
determine, by the mobile computing device and based on the feature vector, an accuracy measure; and
predict, based on the accuracy measure, a mode for the vehicle.