Contextually aware devices and machine learning-driven optimization systems

1 . A method comprising:

receiving, by one or more processors, capacity data of one or more exit devices at a physical location;

receiving, by one or more processors, demand data from one or more monitoring devices that indicates a current demand for exit service at the physical location;

monitoring, by one or more processors, an efficiency metric, wherein the efficiency metric represents a speed at which one of the exit devices generates a weight for an item and an accuracy level of the weight;

processing, by one or more processors, the capacity data and the demand data using a machine learning (ML) model to identify correlations between the capacity data and the demand data;

determining, by one or more processors, using the machine learning (ML) model, the efficiency metric satisfies a threshold;

responsive to identifying the correlations between the capacity data and the demand data and determining the efficiency metric satisfies the threshold, generate a control output configured to drive one or more corrective actions;

transmitting the control output configured to drive the one or more corrective actions to the one or more exit devices for execution:

responsive to transmitting the control output configured to drive the one or more corrective actions, executing, by one or more processors, the one or more corrective actions at the one or more exit devices, wherein executing the one or more corrective actions at the one or more exit devices comprises:

automatically rebooting the one or more exit devices and automatically recalibrating a scale associated with the one or more exit devices.

2 . The method of claim 1 , further comprising:

monitoring changes in at least one of (i) the demand data or (ii) the capacity data after the execution in response to of the control output; and

retraining the ML model based on the changes.

3 . The method of claim 1 , further comprising:

extracting contextual data by processing supplemental information, wherein the contextual data includes at least one of (i) an identifier of peak hour, (ii) an identifier of off-peak hour, (iii) an identifier of a weekday, (iv) an identifier of a weekend, or (v) an identifier of a special event; and

generating the control output based on the capacity data, demand data, and the contextual data using the machine learning (ML) model.

4 . The method of claim 1 , wherein:

the one or more exit devices at the physical location comprises a self-checkout machine;

the self-checkout machine comprises one or more individual devices including at least one of a monitor, a scanner, a printer, a scale, or a pin pad; and

an aggregated capacity of the self-checkout machine is determined based on capacities reported by each of the one or more individual devices.

5 . The method of claim 1 , wherein the demand data comprises at least one of (i) a count of individuals awaiting in a checkout queue, (ii) a quantity of products within each cart pending checkout, (iii) a count of individuals entering the physical location, or (iv) a count of individuals present in the physical location but not in the checkout queue.

6 . The method of claim 5 , wherein the demand data further indicates a trend of the current demand for exit service based on the count of individuals entering the physical location and the count of individuals present in the physical location but not in the checkout queue.

7 . The method of claim 1 , wherein the monitoring devices comprise at least one of (i) a camera installed at the physical location and configured to monitor activities within the physical location, or (ii) a camera or sensor placed on a cart and configured to detect objects within the cart.

8 . The method of claim 1 , wherein:

the ML model is trained using historical pre-implementation condition data as inputs, and using corresponding implemented decisions that historically led to an improved operational efficiency as target outputs; and

the ML model learns to correlate the historical pre-implementation condition data to the implemented decisions.

9 . The method of claim 8 , wherein the historical pre-implementation condition data comprises at least one of (i) historical capacity data of the one or more exit devices at the physical location, or (2) historical demand data indicating immediate demand for exit service at the physical location.

10 . A system comprising:

one or more memories collectively storing computer-executable instructions; and

one or more processors configured to collectively execute the computer-executable instructions and cause the system to:

receive capacity data of one or more exit devices at a physical location;

receive demand data from one or more monitoring devices that indicates a current demand for exit service at the physical location;

monitor an efficiency metric, wherein the efficiency metric represents a speed at which one of the exit devices generates a weight for an item and an accuracy level of the weight;

process the capacity data and the demand data using a machine learning (ML) model to identify correlations between the capacity data and the demand data;

determine, using the machine learning (ML) model, the efficiency metric satisfies a threshold;

responsive to identifying the correlations between the capacity data and the demand data and determining the efficiency metric satisfies the threshold, generate a control output configured to drive one or more corrective actions;

transmit the control output configured to drive the one or more corrective actions to the one or more exit devices for execution:

responsive to transmitting the control output configured to drive the one or more corrective actions, executing, by one or more processors, the one or more corrective actions at the one or more exit devices, wherein executing the one or more corrective actions at the one or more exit devices comprises:

automatically rebooting the one or more exit devices and automatically recalibrating a scale associated with the one or more exit devices.

11 . The system of claim 10 , wherein the computer-executable instructions, when executed by the one or more processors, further cause the system to:

monitor changes in at least one of (i) the demand data or (ii) the capacity data after the execution in response to the control output; and

retrain the ML model based on the changes.

12 . The system of claim 10 , wherein the computer-executable instructions, when executed by the one or more processors, further cause the system to:

extract contextual data by processing supplemental information, wherein the contextual data includes at least one of (i) an identifier of peak hour, (ii) an identifier of off-peak hour, (iii) an identifier of a weekday, (iv) an identifier of a weekend, or (v) an identifier of a special event; and

generate the control output based on the capacity data, the demand data, and the contextual data using the machine learning (ML) model.

13 . The system of claim 10 , wherein:

the one or more exit devices at the physical location comprises a self-checkout machine;

the self-checkout machine comprises one or more individual devices including at least one of a monitor, a scanner, a printer, a scale, or a pin pad; and

an aggregated capacity of the self-checkout machine is determined based on capacities reported by each of the one or more individual devices.

14 . The system of claim 10 , wherein the demand data comprises at least one of (i) a count of individuals awaiting in a checkout queue, (ii) a quantity of products within each cart pending checkout, (iii) a count of individuals entering the physical location, or (iv) a count of individuals present in the physical location but not in the checkout queue.

15 . The system of claim 10 , wherein the monitoring devices comprise at least one of (i) a camera installed at the physical location and configured to monitor activities within the physical location, or (ii) a camera or sensor placed on a cart and configured to detect objects within the cart.

16 . The system of claim 10 , wherein:

the ML model is trained using historical pre-implementation condition data as inputs, and using corresponding implemented decisions that historically led to an improved operational efficiency as target outputs; and

the ML model learns to correlate the historical pre-implementation condition data to the implemented decisions.

17 . The system of claim 16 , wherein the historical pre-implementation condition data comprises at least one of (i) historical capacity data of the one or more exit devices at the physical location, or (2) historical demand data indicating immediate demand for exit service at the physical location.

18 . A computer program product, comprising: computer-readable storage mediums having computer-readable program code executable to cause the computer program product to:

receive capacity data of one or more exit devices at a physical location;

receive demand data from one or more monitoring devices that indicates a current demand for exit service at the physical location;

monitor an efficiency metric, wherein the efficiency metric represents a speed at which one of the exit devices generates a weight for an item and an accuracy level of the weight;

process the capacity data and the demand data using a machine learning (ML) model to identify correlations between the capacity data and the demand data;

determine, using the machine learning (ML) model, the efficiency metric satisfies a threshold;

responsive to identifying the correlations between the capacity data and the demand data and determining the efficiency metric satisfies the threshold, generate a control output configured to drive one or more corrective actions;

transmit the control output configured to drive the one or more corrective actions to the one or more exit devices for execution:

responsive to transmitting the control output configured to drive the one or more corrective actions, executing, by one or more processors, the one or more corrective actions at the one or more exit devices, wherein executing the one or more corrective actions at the one or more exit devices comprises:

automatically rebooting the one or more exit devices and automatically recalibrating a scale associated with the one or more exit devices.