Models for estimating eta and dwell times for transportation of objects
Aspects of the disclosure provide for the use and training of a model configured to provide an estimated time of arrival (ETA) for transporting an object. Data may be received from a reader device. The data may identify a tracking tag and a first timestamp. Based on the received data, a location of the tracking tag at the first timestamp may be determined. A starting location and second timestamp for the tracking tag and a destination location and third timestamp for the tracking tag may be identified. A list including at least one dwell time and dwell location for the tracking tag may be identified. The model may be trained to output the ETA, a dwell time, and a dwell location based on the determined location, the starting location, the destination location, the first timestamp, the second timestamp, the third timestamp, and the list.
1 . A method of training a model to provide an estimated time of arrival (ETA) for transporting an object, the method comprising:
receiving, by one or more processors, data from a reader device, the data identifying a tracking tag associated with the object and a first timestamp;
determining, by the one or more processors, based on the received data, a location of the tracking tag at the first timestamp;
identifying, by the one or more processors, a starting location and second timestamp for the tracking tag and a destination location and third timestamp for the tracking tag;
identifying, by the one or more processors, a list including at least one dwell time and dwell location for the tracking tag; and
training, by the one or more processors, the model to output the ETA, a second dwell time, and a second dwell location for the object based on the determined location, the starting location for the tracking tag, the destination location for the tracking tag, the first timestamp, the second timestamp, the third timestamp, and the list, wherein the training does not involve using an actual route that will be followed by the object to the destination location; and
identifying, based on output of the model, one or more upstream components involved in transporting the object that are slowing transportation of the object, wherein identifying the one or more upstream components enables a logistical provider to take corrective action by modifying the one or more upstream components.
2 . The method of claim 1 , further comprising, determining a run duration based on the second timestamp and the third timestamp, and wherein training the model includes using the run duration.
3 . The method of claim 1 , wherein the data further identifies a mode of transportation associated with the reader device, and wherein training the model further includes using the mode of transportation associated with the reader device to train the model to output one or more modes of transportation with the ETA.
4 . The method of claim 1 , wherein the model is configured to identify dwell locations and dwell times for the object for the ETA.
5 . The method of claim 1 , further comprising, receiving second data from a second reader device, the second data identifying the tracking tag and a fourth timestamp, and wherein determining the location is further based on the second data.
6 . The method of claim 1 , wherein the data indicates that a beacon signal was received by the reader device from the tracking tag at the first timestamp.
7 . The method of claim 1 , further comprising:
using the model to determine an ETA for a second object to reach a second destination location;
receiving second data from a second reader device identifying a tracking tag for the second object;
determining a location of the second object based on the received second data; and
using the model to determine an updated ETA for the second object based on the determined location of the second object.
8 . The method of claim 7 , further comprising, comparing the updated ETA for the second object to the ETA for the second object to determine whether the second object is delayed.
9 . The method of claim 8 , based on the comparison, providing an alert.
10 . The method of claim 8 , further comprising, when the second object is determined to be delayed, determining a likely root cause of a slowdown associated with the second object.
11 . The method of claim 8 , further comprising, when the second object is determined to be delayed, estimating a future slowdown at the second destination location.
12 . The method of claim 1 , wherein the model is one of an extra tree model, gradient boosted regressor model, random forest model, or a voting regression model.
13 . The method of claim 1 , wherein the model is trained to output the list such that the list includes the second dwell time, the second dwell location, and a third dwell time and third dwell location.
14 . The method of claim 1 , wherein the model is trained for trips from a single logistical provider in order to optimize output of the model for that logistical provider.
15 . The method of claim 1 , wherein the model is trained for trips within a geographic area in order to optimize output of the model for that geographic area.
16 . The method of claim 1 , wherein the training does not involve using map information.
17 . A system for training a model to provide an estimated time of arrival (ETA) for transporting an object, the system comprising one or more processors configured to:
receive data from a reader device, the data identifying a tracking tag associated with the object and a first timestamp;
determine, based on the received data, a location of the tracking tag at the first timestamp;
identify, a starting location and second timestamp for the tracking tag and a destination location and third timestamp for the tracking tag;
identify a list including at least one dwell time and dwell location for the tracking tag; and
train the model to output the ETA, a second dwell time, and a second dwell location for the object based on the determined location, the starting location for the tracking tag, the destination location for the tracking tag, the first timestamp, the second timestamp, the third timestamp, and the list, wherein the training does not involve using an actual route that will be followed by the object to the destination location; and
identify, based on output of the model, one or more upstream components involved in transporting the object that are slowing transportation of the object, wherein identifying the one or more upstream components enables a logistical provider to take corrective action by modifying the one or more upstream components.
18 . The system of claim 17 , wherein the one or more processors are further configured to:
use the model to determine an ETA for a second object to reach a second destination location; and
use the model to determine a second dwell time and dwell location for the second object to reach the second destination location.
19 . The system of claim 17 , wherein the one or more processors are further configured to:
use the model to determine an ETA for a second object to reach a second destination location;
receive second data from a second reader device identifying a tracking tag for the second object;
determine a location of the second object based on the received second data; and
use the model to determine an updated ETA for the second object based on the determined location of the second object.
20 . The system of claim 19 , wherein the one or more processors are further configured to compare the updated ETA for the second object to the ETA for the second object to determine whether the second object is delayed.