Computer-implemented systems and methods for real-time risk-informed return item collection using an automated kiosk
Disclosed embodiments provide systems and methods related to collecting return items using an automated kiosk based on a real time risk decision. The automated kiosk captures return item information representing a return item and transmits the return item information and a request for return risk level relating to the return item to a server operable to execute a machine learning model trained on historical information to determine the risk level. The server determines the risk level based on the received return by using the machine learning model and transmits the determined risk level to the kiosk in real-time. Based on the determined risk level and a return amount associated with the return item, the server may also process a refund in real-time.
1 . A system for collecting return items based on a real time risk decision, comprising a return server, an automated kiosk, and a network interface, wherein:
the return server comprises:
one or more memory devices storing instructions; and
one or more processors configured to execute the instructions to perform operations in real-time, the operations comprising:
receiving, from the automated kiosk via the network interface, return item information and a request for return risk level relating to the return item information, wherein the return item information comprises at least one of: an order ID, an item ID, a product barcode, a pre-generated QR code, or a pre-generated return ID;
in response to receiving the request, preparing the return risk level using a supervised machine learning model trained on historical information including historical tags associated with return frauds by:
predicting a risk score of the return request based on applying the supervised machine learning model to the return item information;
determining a risk level based on the predicted risk score, wherein the risk level is a low risk level, a medium risk level, or a high risk level, and wherein the supervised machine learning model is trained to assign at least a predetermined threshold of return requests as low risk; and
transmitting the determined risk level to the automated kiosk;
in response to the determined risk level being the medium risk level, causing the automated kiosk to capture additional return item information via an imaging device, wherein the additional return item information includes at least one photo captured by the imaging device;
receiving, from the automated kiosk via the network interface, the additional return item information;
determining whether the additional return item information is valid;
in response to determining that the additional return item information is not valid, blocking electronic access to a user account associated with the return item by transmitting a first signal to a management system; and
transmitting a validation result to the automated kiosk, wherein transmitting the validation result causes the automated kiosk to display a return result, and wherein when the return result is an acceptance result, the operations further comprise causing the automated kiosk to eject a container for returning the return item; and
the automated kiosk comprises:
the imaging device;
a memory storing instructions; and
at least one processor configured to execute the instructions to automatically perform operations in real-time, the operations including:
automatically detecting a first condition, a second condition, or a third condition of the automated kiosk; and
based on automatically detecting the first condition, second condition, or third condition, automatically transmitting, via the network interface, a corresponding request to the return server,
wherein the first condition of the automated kiosk indicates that the automated kiosk has exceeded a predetermined storage capacity,
wherein the second condition of the automated kiosk indicates that automated kiosk is experiencing an electronic failure,
and wherein the third condition of the automated kiosk is automatically detected by the imaging device.
2 . The system of claim 1 , wherein:
the operations further comprise determining whether the return item is un-returnable; and
the return result is further based on the determination of whether the return item is un-returnable.
3 . The system of claim 1 , the operations further comprising receiving additional return item information based on the risk level being above a threshold.
4 . The system of claim 3 , wherein the requested additional return item information comprises at least one of: an image of the return item or an image of a customer associated with the return item.
5 . The system of claim 1 , wherein:
the additional return item information comprises an image of the return item; and
determining whether the additional return item information is valid comprises comparing the image of the return item to a stored image.
6 . The system of claim 1 , wherein when the additional return item information is valid, the return result is an acceptance result; or
when the additional return item information is not valid, and the return result is a rejection result.
7 . The system of claim 1 , the operations further comprising determining a refund amount and a refund timing, the refund amount and the refunding timing being based on the return result and the risk score.
8 . The system of claim 1 , wherein the low risk level, medium risk level, and high risk level are defined according to a percentage of total returns.
9 . The system of claim 1 , wherein the supervised machine learning model comprises a Gradient Boosting Tree Model.
10 . The system of claim 1 , wherein the historical information further includes at least one of: a return history, a return amount, a list of return products, or a return time.
11 . A method for collecting return items based on a real time risk decision, comprising:
receiving, from an automated kiosk via a network interface, return item information and a request for return risk level relating to the return item information, wherein the return item information comprises at least one of: an order ID, an item ID, a product barcode, a pre-generated QR code, or a pre-generated return ID;
in response to receiving the request, preparing the return risk level using a supervised machine learning model trained on historical information including historical tags associated with return frauds by:
predicting a risk score of the return request based on applying the supervised machine learning model to the return item information;
determining a risk level based on the predicted risk score, wherein the risk level is a low risk level, a medium risk level, or a high risk level, and wherein the supervised machine learning model is trained to assign at least a predetermined threshold of return requests as low risk; and
transmitting the determined risk level to the automated kiosk;
in response to the determined risk level being the medium risk level, causing the automated kiosk to capture additional return item information via an imaging device, wherein the additional return item information includes at least one photo captured by the imaging device;
receiving, from the automated kiosk via the network interface, the additional return item information;
determining whether the additional return item information is valid;
in response to determining that the additional return item information is not valid:
blocking electronic access to a user account associated with the return item by transmitting a first signal to a management system, and
transmitting a first validation result to the automated kiosk, wherein transmitting the first validation result causes the automated kiosk to display a first return result;
in response to determining that the additional return item information is valid, transmitting a second validation result to the automated kiosk, wherein the second validation result is an acceptance result, and wherein transmitting the second validation result causes the automated kiosk to display a second return result and accept the return item by causing the automated kiosk to eject a container for returning the return item; and
receiving a second signal from the automated kiosk indicating that the automated kiosk has exceeded a predetermined storage capacity, wherein the automated kiosk is configured to automatically detect a current storage capacity of the automated kiosk.
12 . The method of claim 11 , further comprising receiving additional return item information based on the risk level being above a threshold.
13 . The method of claim 11 , wherein:
the additional return item information comprises an image of the return item; and
determining whether the additional return item information is valid comprises comparing the image of the return item to a stored image.
14 . The method of claim 11 , further comprising determining a refund amount and a refund timing, the refund amount and the refunding timing being based on the return result and the risk score.
15 . The method of claim 11 , wherein the low risk level, medium risk level, and high risk level are defined according to a percentage of total returns.
16 . The method of claim 11 , further comprising determining whether the return item is un-returnable, wherein the return result is further based on the determination of whether the return item is un-returnable.
17 . The method of claim 11 , wherein the requested additional return item information comprises at least one of: an image of the return item or an image of a customer associated with the return item.