Granted Patent B1

US 12,630,326 · App. 18/535,210 · Granted May 19, 2026

Inventors: Timothy G. Dietz (Reading, MA); Pablo Hernández-Becerro (Somerville, MA)

Assignee: Amazon Technologies, Inc.

B65B69/0033G05B13/0265

Abstract

Claims (87)

1 . A container opening system comprising:

a robotic manipulator;

a container opening knife coupled to the robotic manipulator;

a force feedback sensor configured to measure axial force applied to the container opening knife;

a camera configured to image a first package;

a light source configured to illuminate the first package; and

a controller configured to:

determine a first image of the first package using the camera;

determine, using the first image, a first cut path to open the first package, the first cut path comprising a first cut height at which to position the container opening knife along an external surface of the first package;

determine, using the first image and a second machine learning model, that the first package has an obstruction, wherein the first cut path dynamically avoids the obstruction;

determine, using the first image, a first type of packaging material of the first package;

determine, using the first type of packaging material and a first machine learning model, a first amount of axial force to apply to the external surface of the first package via the container opening knife;

cause the robotic manipulator to position the container opening knife at the first cut height;

cause the robotic manipulator to apply the first amount of axial force to the external surface of the first package;

cause the robotic manipulator to move the container opening knife along the first cut path to open the first package;

determine, using the force feedback sensor while the container opening knife is moving along the first cut path, that a first change to axial force applied to the container opening knife is greater than a predetermined threshold;

determine, using the first machine learning model, a second amount of axial force to apply to the external surface of the first package for a remainder of the first cut path based at least in part on the first change being greater than the predetermined threshold; and

cause the first machine learning model to be updated based at least in part on the first type of packaging material and the first change to axial force.

2 . The container opening system of claim 1 , wherein the controller is further configured to:

determine a second image of a second package using the camera;

determine, using the second image, that the second package has a non-rectangular geometry;

determine, based at least in part on the determination that the second package has a non-rectangular geometry, a second cut height at which to position the container opening knife along an external surface of the second package; and

determine a second cut path to open the second package using the second cut height.

3 . The container opening system of claim 1 , wherein the first cut path does not completely separate an upper portion of the first package from a lower portion of the first package.

4 . A system comprising:

a container opening device; and

a controller configured to:

determine a first image of a first package;

determine, using the first image, a first cut path for the container opening device;

determine, using the first image and a second machine learning model, that the first package has an obstruction, wherein the first cut path dynamically avoids the obstruction;

determine, using the first image, a first type of packaging material of the first package;

determine, using the first type of packaging material and a first machine learning model, a first amount of axial force to apply to the external surface of the first package via the container opening device;

cause the container opening device to (i) apply the first amount of axial force to the external surface of the first package, and (ii) move along the first cut path to open the first package;

determine that an amount of resistance as the container opening device moves along the first cut path is greater than a first predetermined threshold;

determine, using a second machine learning model, an adjustment to a lateral movement speed for the container opening device based at least in part on the amount of resistance being greater than the first predetermined threshold; and

cause the container opening device to move at the adjusted lateral movement speed.

5 . The system of claim 4 , wherein the controller is further configured to:

determine, using the first image, a first cut height at which to position the container opening device along an external surface of the first package;

wherein the first cut path is at the first cut height.

6 . The system of claim 5 , wherein the controller is further configured to:

determine, using the first image, that the first package has a non-rectangular geometry; and

determine a first adjustment value to a default cut height;

wherein the first cut height is a sum of the default cut height and the first adjustment value.

7 . The system of claim 4 , wherein the controller is further configured to:

determine, using a force sensor while the container opening device is moving along the first cut path, that a first change to axial force applied to the container opening device is greater than a second predetermined threshold;

determine a second amount of axial force to apply to the external surface of the first package for a remainder of the first cut path; and

cause the first machine learning model to be updated based at least in part on the first type of packaging material and the first change to axial force.

8 . The system of claim 4 , wherein the controller is further configured to:

determine, after applying the first amount of axial force to the external surface of the first package, that a reduction in axial force applied to the container opening device is less than a second predetermined threshold; and

determine, using the first machine learning model, a second amount of axial force to apply to the external surface of the first package via the container opening device.

9 . The system of claim 4 , wherein the controller is further configured to:

determine, after applying the first amount of axial force to the external surface of the first package, that an increase in axial force applied to the container opening device is greater than a second predetermined threshold; and

determine, using the first machine learning model, a second amount of axial force to apply to the external surface of the first package via the container opening device.

10 . The system of claim 4 , wherein the controller is further configured to:

determine, using the first image and a second machine learning model, that the first package has a printed identifier;

wherein the first cut path avoids the printed identifier.

11 . The system of claim 4 , wherein the first cut path does not separate an upper portion of the first package from a lower portion of the first package.

12 . The system of claim 4 , wherein the container opening device is coupled to a robotic manipulator, and wherein the system further comprises:

a force sensor configured to measure axial force applied to the container opening device;

a camera; and

a light source.

13 . A container opening system comprising:

a container opening device; and

a controller configured to:

determine a first image of a first package;

determine, using the first image, a first cut path for the container opening device;

determine, using the first image and a second machine learning model, that the first package has an obstruction, wherein the first cut path dynamically avoids the obstruction;

determine, using the first image, a first cut height at which to position the container opening device along an external surface of the first package, wherein the first cut path is at the first cut height;

determine, using the first image, a first type of packaging material of the first package;

determine, using the first type of packaging material and a first machine learning model, a first amount of axial force to apply to the external surface of the first package via the container opening device;

cause the container opening device to apply the first amount of axial force to the external surface of the first package;

cause the container opening device to move along the first cut path to open the first package;

determine, while the container opening knife is moving along the first cut path, that a first change to axial force applied to the container opening knife is greater than a first predetermined threshold;

determine, using the first machine learning model, a second amount of axial force to apply to the external surface of the first package for a remainder of the first cut path based at least in part on the first change being greater than the first predetermined threshold; and

cause the first machine learning model to be updated based at least in part on the first type of packaging material and the first change to axial force.

14 . The container opening system of claim 13 , wherein the controller is further configured to:

determine that an amount of resistance as the container opening device moves along the first cut path is greater than a second predetermined threshold;

determine, using a second machine learning model, an adjustment to a lateral movement speed for the container opening device; and

cause the container opening device to move at the adjusted lateral movement speed.

15 . The container opening system of claim 13 , wherein the controller is further configured to:

determine, after applying the first amount of axial force to the external surface of the first package, that a reduction in axial force applied to the container opening device is less than a second predetermined threshold; and

determine, using the first machine learning model, a second amount of axial force to apply to the external surface of the first package via the container opening device.

16 . The container opening system of claim 13 , wherein the first cut path does not separate an upper portion of the first package from a lower portion of the first package.

17 . The container opening system of claim 13 , wherein the container opening device is coupled to a robotic manipulator, and wherein the system further comprises:

a force sensor configured to measure axial force applied to the container opening device;

a camera; and

a light source.

Assignments (1)

References Cited (43)