Vehicle dispatch using machine learning
A dispatch database maintains, for a plurality of vehicles available for dispatch, vehicle data and constraints data. A processor is programmed to execute a dispatch server to perform operations including to receive a dispatch request requesting a vehicle to arrive at a request location, utilize a machine-learning model to identify one or more of the plurality of vehicles to respond to the dispatch request, the machine-learning model utilizing the vehicle data and the constraints data as inputs to determine the one or more of the plurality of vehicles, and inform the one or more of the plurality of vehicles of the dispatch request.
1. A system for use of machine learning for dispatch of mobile aid and service, comprising:
a dispatch database maintaining, for a plurality of vehicles available for dispatch, vehicle data and constraints data; and
a processor programmed to execute a dispatch server to perform operations including to:
receive a dispatch request requesting a vehicle to arrive at a request location,
utilize a machine-learning model to identify a subset of the plurality of vehicles to respond to the dispatch request, the subset including vehicles indicated, by the machine-learning model, as being most probable choices to handle the dispatch request, the machine-learning model utilizing the vehicle data and the constraints data as inputs to determine the subset of the plurality of vehicles,
inform the subset of the plurality of vehicles of the dispatch request,
receive a result indicative of which one of the subset of the plurality of vehicles actually performed the dispatch request, and
update training of the machine-learning model using the vehicle data, the constraints data, and the result to improve the machine-learning model in learning to identify the subset one or more of the plurality of vehicles that are the most probable choices to handle the dispatch request, including to
receive historical constraints data and historical vehicle data from the plurality of vehicles;
receive historical dispatch requests during a period of time for which the historical vehicle data and the historical constraints data is available; and
train the machine-learning model in dispatch of the plurality of vehicles using the historical vehicle data, the historical constraints data, and the historical dispatch requests provided as the inputs to the machine-learning model, and an indication of which of the plurality of vehicles was dispatched for the historical dispatch requests as ground truth for intended output of the machine-learning model,
set aside a portion of the historical vehicle data and the historical constraints data from the training,
use the portion to validate the machine-learning model,
perform additional training of the machine-learning model using the historical vehicle data and the historical constraints data responsive to the machine-learning model not meeting with the ground truth for at least a minimum threshold, and
apply the machine-learning model for use by the dispatch server in handling additional dispatch requests responsive to the machine-learning model meeting with the ground truth for at least the minimum threshold.
2. The system of claim 1 , wherein the processor is further programmed to:
utilize the machine-learning model to identify a destination, from a plurality of destinations, for the dispatch request, the machine-learning model utilizing data including desired capabilities of the plurality of destinations, availability of the plurality of destinations, and distances to the plurality of destinations; and
inform the subset of the plurality of vehicles of the destination.
3. The system of claim 2 , wherein the processor is further programmed to:
receive a result indicative of whether a correct destination location was chosen; and
update training of the machine-learning model using the vehicle data, the constraints data, and the result to improve the machine-learning model in learning for the dispatch of mobile aid and service.
4. The system of claim 1 , wherein the request location is a dynamic location specified as an identifier of a vehicle or mobile device, where the processor is further programmed to track the dynamic location according to the identifier.
5. A method for use of machine learning for dispatch of mobile aid and service, comprising:
maintaining, for a plurality of vehicles available for dispatch, vehicle data and constraints data;
receiving a dispatch request requesting a vehicle to arrive at a request location;
utilizing a machine-learning model to identify a subset of the plurality of vehicles to respond to the dispatch request, the subset including vehicles indicated, by the machine-learning model, as being most probable choices to handle the dispatch request, the machine-learning model utilizing the vehicle data and the constraints data as inputs to determine the subset of the plurality of vehicles;
informing the subset of the plurality of vehicles of the dispatch request;
receiving a result indicative of which one of the subset of the plurality of vehicles actually performed the dispatch request; and
updating training of the machine-learning model using the vehicle data, the constraints data, and the result to improve the machine-learning model in learning to identify the subset one or more of the plurality of vehicles that are the most probable choices to handle the dispatch request, including
receiving historical constraints data and historical vehicle data from the plurality of vehicles;
receiving historical dispatch requests during a period of time for which the historical vehicle data and the historical constraints data is available; and
training the machine-learning model in dispatch of the plurality of vehicles using the historical vehicle data, the historical constraints data, and the historical dispatch requests provided as the inputs to the machine-learning model, and an indication of which of the plurality of vehicles was dispatched for the historical dispatch requests as ground truth for intended output of the machine-learning model,
setting aside a portion of the historical vehicle data and the historical constraints data from the training,
using the portion to validate the machine-learning model,
performing additional training of the machine-learning model using the historical vehicle data and the historical constraints data responsive to the machine-learning model not meeting with the ground truth for at least a minimum threshold, and
applying the machine-learning model for use in handling additional dispatch requests responsive to the machine-learning model meeting with the ground truth for at least the minimum threshold.
6. The method of claim 5 , further comprising:
utilizing the machine-learning model to identify a destination, from a plurality of destinations, for the dispatch request, the machine-learning model utilizing data including desired capabilities of the plurality of destinations, availability of the plurality of destinations, and distances to the plurality of destinations; and
informing the subset of the plurality of vehicles of the destination.
7. The method of claim 6 , further comprising:
receiving a result indicative of whether a correct destination location was chosen; and
updating training of the machine-learning model using the vehicle data, the constraints data, and the result to improve the machine-learning model in learning for the dispatch of mobile aid and service.
8. The method of claim 5 , wherein the request location is a dynamic location specified as an identifier of a vehicle or mobile device, further comprising tracking the dynamic location according to the identifier.
9. A non-transitory computer-readable medium comprising instructions for use of machine learning for dispatch of mobile aid and service that, when executed by a processor, cause the processor to:
maintain, for a plurality of vehicles available for dispatch, vehicle data and constraints data;
receive a dispatch request requesting a vehicle to arrive at a request location;
utilize a machine-learning model to identify a subset of the plurality of vehicles to respond to the dispatch request, the subset including vehicles indicated, by the machine-learning model, as being most probable choices to handle the dispatch request, the machine-learning model utilizing the vehicle data and the constraints data as inputs to determine the subset of the plurality of vehicles;
inform the subset of the plurality of vehicles of the dispatch request;
receive a result indicative of which one of the subset of the plurality of vehicles actually performed the dispatch request; and
update training of the machine-learning model using the vehicle data, the constraints data, and the result to improve the machine-learning model in learning to identify the subset one or more of the plurality of vehicles that are the most probable choices to handle the dispatch request, including to
receive historical constraints data and historical vehicle data from the plurality of vehicles;
receive historical dispatch requests during a period of time for which the historical vehicle data and the historical constraints data is available; and
train the machine-learning model in dispatch of the plurality of vehicles using the historical vehicle data, the historical constraints data, and the historical dispatch requests provided as the inputs to the machine-learning model, and an indication of which of the plurality of vehicles was dispatched for the historical dispatch requests as ground truth for intended output of the machine-learning model,
set aside a portion of the historical vehicle data and the historical constraints data from the training,
use the portion to validate the machine-learning model,
perform additional training of the machine-learning model using the historical vehicle data and the historical constraints data responsive to the machine-learning model not meeting with the ground truth for at least a minimum threshold, and
apply the machine-learning model for use by the processor in handling additional dispatch requests responsive to the machine-learning model meeting with the ground truth for at least the minimum threshold.
10. The medium of claim 9 , further comprising instructions that, when executed by the processor, cause the processor to:
utilize the machine-learning model to identify a destination, from a plurality of destinations, for the dispatch request, the machine-learning model utilizing data including desired capabilities of the plurality of destinations, availability of the plurality of destinations, and distances to the plurality of destinations; and
inform the subset of the plurality of vehicles of the destination.
11. The medium of claim 10 , further comprising instructions that, when executed by the processor, cause the processor to:
receive a result indicative of whether a correct destination location was chosen; and
update training of the machine-learning model using the vehicle data, the constraints data, and the result to improve the machine-learning model in learning for the dispatch of mobile aid and service.
12. The medium of claim 9 , wherein the request location is a dynamic location specified as an identifier of a vehicle or mobile device, and further comprising instructions that, when executed by the processor, cause the processor to track the dynamic location according to the identifier.