Manufacturing multi-component objects using artificial intelligence

1 . A method of manufacturing an object, the method comprising:

selecting, from a plurality of first components each having a respective value for a respective first attribute within a tolerance range, a first component of an object;

determining an actual value of the first attribute of the first component;

determining, using a machine learning model and based on the actual value of the first attribute, an optimized value for a second attribute of a second component that is functionally interrelated to the first component in the object;

selecting, from a plurality of second components each having a value for the second attribute within another tolerance range, the second component having the optimized value for the second attribute; and

manufacturing the object using the first component and the selected second component.

2 . The method of claim 1 , wherein determining the actual value of the first attribute of the first component comprises:

capturing one or more images of the first component; and

analyzing the one or more images to determine the actual value of the first attribute of the first component.

3 . The method of claim 2 , wherein analyzing the one or more images comprises analyzing, using another machine learning model, the one or more images to determine the actual value of the first attribute of the first component.

4 . The method of claim 1 , wherein determining the actual value of the first attribute of the first component comprises analyzing data from a manufacturer of the first component.

5 . The method of claim 1 , further comprising training the machine learning model using historical data comprising (i) historical actual values of first attributes of first components of historical objects, (ii) historical actual values of second attributes of second components of the historical objects, and (iii) historical test results obtained by testing the historical objects.

6 . The method of claim 1 , further comprising, after manufacturing the object:

performing one or more tests on the object to obtain test results; and

retraining the machine learning model using the actual value of the first attribute of the first component, the optimized value of the second attribute of the second component, and the test results.

7 . The method of claim 1 , wherein:

the first component comprises a circuit board;

the first attribute is a first diameter of a hole in the circuit board;

the second component comprises a pin having a portion configured to be received in the hole via an interference fit; and

the second attribute is a second outer diameter of the portion of the pin configured to be received in the hole via the interference fit.

8 . The method of claim 1 , wherein:

the first component comprises an electronic component positioned on a circuit board;

the first attribute is a position of the electronic component on the circuit board;

the second component comprises an adhesive for securing the electronic component to the circuit board; and

the second attribute comprises an amount and a position of the adhesive with respect to the electronic component and the circuit board.

9 . The method of claim 1 , wherein:

the first component comprises a processor;

the first attribute comprises thermal performance of the processor;

the second component comprises a circuit board; and

the second attribute comprises a position on the circuit board for receiving the processor and an orientation of the position on the circuit board with respect to a coolant flow.

10 . The method of claim 9 , comprising, before manufacturing the object:

determining, using another machine learning model and based on the thermal performance of the processor, the position on the circuit board for receiving the processor, and the orientation of the circuit board with respect to the coolant flow, an optimized thermal performance of a heatsink for cooling the processor on the circuit board; and

selecting, from a plurality of heatsinks each having a thermal performance value within a respective tolerance range, the heatsink having the optimized thermal performance.

11 . A method of manufacturing an object, the method comprising:

selecting, from a plurality of first components each having a respective value for a respective first attribute within a tolerance range, a first component of an object;

determining a first actual value of the first attribute of the first component;

determining a second actual value of a second attribute of a second component of the object;

determining, using a machine learning model and based on the first actual value of the first attribute and the second actual value of the second attribute, an optimized value for a third attribute of a third component that is functionally interrelated to the first component and the second component of the object;

selecting, from a plurality of third components each having a value for the third attribute within another tolerance range, the third component having the optimized value for the third attribute; and

manufacturing the object using the first component, the second component, and the selected third component.

12 . The method of claim 11 , wherein:

the first component comprises a processor positioned on a circuit board;

the first attribute comprises a first geometry of the processor on the circuit board;

the second component comprises a heatsink for cooling the processor on the circuit board, wherein the heatsink comprises a pedestal for interfacing with the processor;

the second attribute comprises a second geometry of the pedestal of the heatsink;

the third component comprises a section of thermal interface material; and

the third attribute comprises a third geometry of the section of the thermal interface material.

13 . The method of claim 11 , wherein:

the first component comprises a first section of a shell of a vapor chamber;

the first attribute comprises a first geometry of the first section of the shell;

the second component comprises a second section of the shell;

the second attribute comprises a second geometry of the second section of the shell;

the third component comprises a wicking structure on an interior of the shell; and

the third attribute comprises a first thickness of the wicking structure in the first section of the shell and a second thickness of the wicking structure in the second section of the shell.

14 . The method of claim 13 , wherein manufacturing the object comprises:

disposing a first amount-by-weight of powder to the first section of the shell to achieve the first thickness of the wicking structure; and

disposing a second amount-by-weight of the powder to the second section of the shell to achieve the second thickness of the wicking structure.

15 . The method of claim 11 , wherein determining the actual value of the first attribute of the first component comprises:

capturing one or more images of the first component; and

analyzing the one or more images to determine the actual value of the first attribute of the first component, wherein analyzing the one or more images comprises analyzing, using another machine learning model, the one or more images to determine the actual value of the first attribute of the first component.

16 . The method of claim 11 , further comprising training the machine learning model using historical data comprising (i) historical actual values of first attributes of first components of historical objects, (ii) historical actual values of second attributes of second components of the historical objects, (iii) historical actual values of third attributes of third components of the historical objects, and (iv) historical test results obtained by testing the historical objects.

17 . A method of manufacturing an object, the method comprising:

performing a respective step in a series of steps for manufacturing an object to achieve a target outcome of the respective step;

after performing the respective step, capturing one or more images of an actual outcome of the respective step;

providing the one or more images to one or more machine learning models to determine whether the actual outcome of the respective step corresponds to the target outcome of the respective step;

in response to the actual outcome not corresponding to the target outcome, stopping performance of the series of steps and generating a notification; and

in response to the actual outcome corresponding to the target outcome, initiating performance of a subsequent step in the series of steps.

18 . The method of claim 17 , wherein the series of steps comprises:

a first step of applying a thermal interface material to a surface of a heatsink, wherein the thermal interface material is adhered to a protective film;

a second step of removing the protective film from the thermal interface material, wherein the second step is performed after the first step is performed; and

a third step of disposing a component on the thermal interface material, wherein the third step is performed after the second step is performed;

wherein a machine learning model of the one or more machine learning models is configured to analyze the one or more images to determine whether the protective film was removed from the thermal interface material; and

wherein the method further comprises preventing, in response to determining that the protective film was not removed from the thermal interface material, performance of the third step.