Servicing of autonomous vehicles

1. A system, comprising:

a data storage; and

a processor, configured to:

retrieve data from the data storage;

based on the data, generate a plurality of training vectors relating to different respective historical service requests generated by respective autonomous vehicles, each of the training vectors including respective features associated with the historical request to which the training vector relates;

using the training vectors, train a model configured to generate a plurality of types of predictions relating to one or more future service requests, wherein the plurality of types of predictions includes a first type of prediction and a second type of prediction;

generate at least one updated feature vector based on newly received historical service requests;

generate a plurality of new training vectors based on the at least one updated feature vector;

return the model using the generated plurality of new training vectors in order to create a retrained model, wherein generating the predictions of the first type of prediction further comprises passing the first updated feature vectors to the retrained model; and

using the retrained model, generate predictions of the first type of prediction at a first frequency;

using the retrained model, generate the predictions of the second type of prediction at a second frequency, wherein the first frequency is higher than the second frequency; and

output the generated plurality of predictions.

2. The system according to claim 1 , wherein the historical requests include historical requests for assistance issued to a remote support staff, and wherein the future service requests include future requests for assistance.

3. The system according to claim 2 , wherein the historical requests for assistance include historical requests for navigational assistance, and wherein the future requests for assistance include future requests for navigational assistance.

4. The system according to claim 2 , wherein the prediction includes a predicted number of support-staff members required to service the future requests during a predefined interval of time.

5. The system according to claim 1 , wherein the historical requests include historical requests for maintenance, and wherein the future requests include future requests for maintenance.

6. The system according to claim 1 , wherein the prediction includes a start of an interval of time during which a particular one of the vehicles will next generate one of the future requests.

7. The system according to claim 1 , wherein the prediction is for one of the future requests being generated during a predefined interval of time.

8. The system according to claim 1 , wherein the prediction includes a predicted number of the future requests during a predefined interval of time.

9. The system according to claim 1 , wherein the features include a health score of the autonomous vehicle that generated the historical request at a time of the historical request.

10. The system of claim 1 , wherein the at least one updated feature vector includes first updated feature vectors and second updated feature vectors, wherein the processor is further configured to:

generate the first updated feature vectors each including a first set of features at the first frequency, wherein generating the predictions of the first type of prediction further comprises passing the first updated feature vectors to the retrained model; and

generate the second updated feature vectors each including a second set of features at the second frequency, wherein generating the predictions of the second type of prediction further comprises passing the second updated feature vectors to the retrained model.

11. The system of claim 10 , wherein each first feature vector and each second feature vector passed to the retrained model is passed along with a type of prediction to be generated by the model using the passed feature vector.

12. A method, comprising:

generating a plurality of training vectors relating to different respective historical service requests generated by respective autonomous vehicles, each of the training vectors including respective features associated with the historical request to which the training vector relates;

using the training vectors, training, by a processor, a model configured to generate a plurality of types of predictions relating to one or more future service requests, wherein the plurality of types of predictions includes a first type of prediction and a second type of prediction;

generating at least one updated feature vector based on newly received historical service requests;

generating a plurality of new training vectors based on the at least one updated feature vector;

retraining the model using the generated plurality of new training vectors in order to create a retrained model, wherein generating the predictions of the first type of prediction further comprises passing the first updated feature vectors to the retrained model; and

using the retrained model, generating predictions of the first type of prediction at a first frequency;

using the retrained model, generating the predictions of the second type of prediction at a second frequency, wherein the first frequency is higher than the second frequency; and

outputting the generated plurality of predictions.

13. The method according to claim 12 , wherein the historical requests include historical requests for assistance issued to a remote support staff, and wherein the future requests include future requests for assistance.

14. The method according to claim 13 , wherein the historical requests for assistance include historical requests for navigational assistance, and wherein the future requests for assistance include future requests for navigational assistance.

15. The method according to claim 13 , wherein the prediction includes a predicted number of support-staff members required to service the future requests during a predefined interval of time.

16. The method according to claim 12 , wherein the historical requests include historical requests for maintenance, and wherein the future requests include future requests for maintenance.

17. The method according to claim 12 , wherein the prediction includes a start of an interval of time during which a particular one of the vehicles will next generate one of the future requests.

18. The method according to claim 12 , wherein the prediction is for one of the future requests being generated during a predefined interval of time.

19. The method according to claim 12 , wherein the prediction includes a predicted number of the future requests during a predefined interval of time.

20. The method according to claim 12 , wherein the features include a health score of the autonomous vehicle that generated the historical request at a time of the historical request.

21. The method of claim 12 , wherein the first type of prediction relates to a time at which one of the autonomous vehicles will generate a next service request, wherein the second type of prediction relates to a predicted number of support staff members required during a given period of time.

22. A computer software product comprising a tangible non-transitory computer-readable medium in which program instructions are stored, which instructions, when read by a processor, cause the processor to:

generate a plurality of training vectors relating to different respective historical service requests generated by respective autonomous vehicles, each of the training vectors including respective features associated with the historical request to which the training vector relates;

using the training vectors, training a model configured to generate a plurality of types of predictions relating to one or more future service requests, wherein the plurality of types of predictions includes a first type of prediction and a second type of prediction;

generate at least one updated feature vector based on newly received historical service requests;

generate a plurality of new training vectors based on the at least one updated feature vector;

retrain the model using the generated plurality of new training vectors in order to create a retrained model, wherein generating the predictions of the first type of prediction further comprises passing the first updated feature vectors to the retrained model; and

using the retrained model, generating predictions of the first type of prediction at a first frequency;

using the retrained model, generating the predictions of the second type of prediction at a second frequency, wherein the first frequency is higher than the second frequency; and

outputting the generated plurality of predictions.