Supply chain visibility platform
Systems, methods, and non-transitory media are provided for dynamically predicting visibility of freights in a constrained environment. An example method can include determining attributes associated with a load transported by a carrier from a source to a destination, the attributes including an identity of the carrier, an identity of an industry associated with the load, an identity of a shipper of the load, load characteristics, and/or a pickup time of the load; based on the attributes, predicting a route the carrier will follow when transporting the load to the destination, at least a portion of the route being predicted without data indicating an actual presence of the carrier within the portion of the route, the data including location measurements from a device associated with the carrier and/or a location update from the carrier; and generating a tracking interface identifying the route the carrier is predicted to follow.
1 . A method comprising:
requesting, by a predictive visibility system comprising one or more processors, a location of one or more devices based on a predicted time the one or more devices will be located at a user configured location, wherein the one or more devices is associated with tracking a load being transported by a carrier from a source location to a destination location;
determining, by the predictive visibility system, a communication error between the predictive visibility system and the one or more devices;
determining, by the predictive visibility system, one or more predicted locations of the load while the predictive visibility system is unable to communicate with the one or more devices or the carrier, wherein the one or more predicted locations of the load are determined based on physical characteristics of the load influencing an environment and route, wherein the predictive visibility system comprises a model trained based on an RNN configuration; and
generating, by the model, tracking information that identifies the one or more predicted locations of the load and a trajectory of the load from a current location to the destination location, the trajectory of the load being based on the physical characteristics of the load, the one or more predicted locations of the load and the destination location;
wherein the model determines the trajectory of the load by sequentially processing input data defining a previous state of the load and a current state of the load through application of input-adaptive weighting values to the input data, the input data comprising the physical characteristics of the load, the one or more predicted locations of the load, and the destination location.
2 . The method of claim 1 , further comprising: providing the tracking information to a computing device associated with a load tracking interface.
3 . The method of claim 1 , further comprising:
providing, to a load tracking interface, a first indication that the current location of the load comprises an observed location and a second indication that the one or more predicted locations of the load are location predictions.
4 . The method of claim 1 , further comprising:
determining, by the predictive visibility system, a predicted route the carrier is predicted to follow when transporting the load to the destination location, wherein the tracking information further comprises the predicted route.
5 . The method of claim 4 , wherein the predicted route is determined during the communication error, and wherein the predicted route is determined based on the destination location and the physical characteristics of the load.
6 . The method of claim 4 , further comprising: providing, to a load tracking interface, an indication of the predicted route the carrier is predicted to follow.
7 . The method of claim 1 , wherein the physical characteristics of the load comprise at least one of an identity of the carrier, a seasonality, an identity of an industry associated with the load, an identity of a shipper of the load, one or more load characteristics, and a pickup time associated with the load.
8 . The method of claim 7 , wherein the physical characteristics of the load comprise at least one of a type of load, a load weight, and a transportation requirement associated with the load.
9 . The method of claim 1 , further comprising: generating a load tracking interface that displays the tracking information and a map displaying the current location of the load at a first time and the one or more predicted locations of the load at one or more additional times.
10 . The method of claim 1 , further comprising:
based on the physical characteristics of the load, predicting a behavior of the carrier at one or more times while transporting the load to the destination location, wherein the predicted behavior of the carrier comprises at least one of stopping at one or more locations, traveling at a predicted speed, changing a traveling velocity and changing a traveling trajectory; and
providing, to a load tracking interface, an indication of the predicted behavior of the carrier at the one or more times.
11 . A system comprising:
one or more processors; and
at least one computer-readable medium having stored thereon instructions that, when executed by the one or more processors, cause the one or more processors to:
request a location of one or more devices based on a predicted time the one or more devices will be located at a user configured location, wherein the one or more devices is associated with tracking a load being transported by a carrier from a source location to a destination location;
determine a communication error between the system and the one or more devices;
determine one or more predicted locations of the load while the system is unable to communicate with the one or more devices or the carrier, wherein the one or more predicted locations of the load are determined based on one or more physical characteristics of the load influencing an environment and route; and
generate, by a model trained based on an RNN configuration, tracking information that identifies the one or more predicted locations of the load and a trajectory of the load from a current location to the destination location, the trajectory of the load being based on the physical characteristics of the load, the one or more predicted locations of the load, and the destination location;
wherein the model determines the trajectory of the load by sequentially processing, across one or more hidden layers, input data defining a previous state of the load and a current state of the load through the application of input-adaptive weighting values to the input data, the input data comprising the physical characteristics of the load, the one or more predicted locations of the load, and the destination location.
12 . The system of claim 11 , the at least one computer-readable medium having stored thereon instructions that, when executed by the one or more processors, cause the one or more processors to: provide the tracking information to a computing device associated with a load tracking interface.
13 . The system of claim 11 , the at least one computer-readable medium having stored thereon instructions that, when executed by the one or more processors, cause the one or more processors to:
provide, to a load tracking interface, a first indication that the current location of the load comprises an observed location and a second indication that the one or more predicted locations of the load are location predictions.
14 . The system of claim 11 , the at least one computer-readable medium having stored thereon instructions that, when executed by the one or more processors, cause the one or more processors to:
determine a predicted route the carrier is predicted to follow when transporting the load to the destination location, wherein the tracking information further comprises the predicted route, and wherein the predicted route is determined during the communication error.
15 . The system of claim 14 , the at least one computer-readable medium having stored thereon instructions that, when executed by the one or more processors, cause the one or more processors to: provide, to a load tracking interface, an indication of the predicted route the carrier is predicted to follow.
16 . The method of claim 1 , wherein the physical characteristics of the load comprise at least one of an identity of the carrier, a seasonality, an identity of an industry associated with the load, an identity of a shipper of the load, one or more load characteristics, and a pickup time associated with the load, and wherein the one or more load characteristics comprise at least one of a type of load, a load weight, and a transportation requirement associated with the load.
17 . The system of claim 11 , the at least one computer-readable medium having stored thereon instructions that, when executed by the one or more processors, cause the one or more processors to: generate a load tracking interface that displays the tracking information and a map displaying the current location of the load at a first time and the one or more predicted locations of the load at one or more additional times.
18 . A non-transitory computer-readable storage medium comprising:
instructions that, when executed by one or more processors, cause the one or more processors to:
request, by a predictive visibility system, a location of one or more devices based on a predicted time the one or more devices will be located at a user configured location, wherein the one or more devices is associated with tracking a load being transported by a carrier from a source location to a destination location;
determine a communication error between the predictive visibility system and the one or more devices;
determine one or more predicted locations of the load while the predictive visibility system is unable to communicate with the one or more devices or the carrier, wherein the one or more predicted locations of the load are determined based on physical characteristics of the load influencing an environment and route, wherein the predictive visibility system comprises a model trained based on an RNN configuration; and
generate tracking information that identifies the one or more predicted locations of the load and a trajectory of the load from a current location to the destination location, the trajectory of the load being based on the physical characteristics of the load, the one or more predicted locations of the load, and the destination location;
wherein the model determines the trajectory of the load by sequentially processing input data defining a previous state of the load and a current state of the load through the application of input-adaptive weighting values to the input data, the input data comprising the physical characteristics of the load, the one or more predicted locations of the load, and the destination location.