System and method for generating an assembly sequence for a product

A system for generating an assembly sequence for a product, the system comprising:

• a memory;
• and
• one or more processors communicatively coupled to the memory, the one or more processors are configured to:
• generate an assembly graph associated with assembly of the product based on information associated with each component of a plurality of components of the product, wherein each node of the assembly graph includes a node embedding for a component graph representing a respective component of the plurality of components, and wherein an initial state of the assembly graph includes no links between any of two nodes of the initial assembly graph;
• generate the assembly sequence by applying the assembly graph to a recurrent graph neural network, wherein each recurrent step represents a step of the assembly sequence, and wherein configuration of the one or more processors to generate the assembly sequence includes configuration of the one or more processors to, at each recurrent step:
• update, by a graph encoder of the recurrent graph neural network, the node embedding for each node of the assembly graph based on a current state of the assembly graph;
• identify, the recurrent graph neural network, based on the updated node embedding for each node of the assembly graph, a link between a node of the assembly graph and one or more other nodes of the assembly graph;
• add the identified link between the node and the one or more other nodes to the assembly graph to update the current state of the assembly graph to include the identified link, wherein the identified link between the node of the assembly graph and the one or more other nodes represents an assembly connection between a component of the product represented by the node and one or more other components of the product represented by the one or more other nodes;
• and
• feedback the assembly graph with the updated current state including the identified link to the graph encoder for updating, at a next recurrent step, of the node embedding of each node of the assembly graph using the updated current state of the assembly graph;
• and
• continue updating, identifying, and adding until at least one link with another node is identified for every node of the assembly graph, thereby the generated assembly sequence is used to assemble the product.

A non-transitory computer-readable storage medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform operations for generating an assembly sequence for a product, the operations comprising:

• generating an assembly graph associated with assembly of the product based on information associated with each component of a plurality of components of the product, wherein each node of the assembly graph includes a node embedding for a component graph representing a respective component of the plurality of components, and wherein an initial state of the assembly graph includes no links between any of two nodes of the initial assembly graph;
• generating the assembly sequence by applying the assembly graph to a recurrent graph neural network, wherein each recurrent step represents a step of the assembly sequence, and wherein generating the assembly sequence includes, at each recurrent step:
• updating, by a graph encoder of the recurrent graph neural network, the node embedding for each node of the assembly graph based on a current state of the assembly graph;
• identifying, by the recurrent graph neural network, based on the updated node embedding for each node of the assembly graph, a link between a node of the assembly graph and one or more other nodes of the assembly graph;
• adding the identified link between the node and the one or more other nodes to the assembly graph to update the current state of the assembly graph to include the identified link, wherein the identified link between the node of the assembly graph and the one or more other nodes represents an assembly connection between a component of the product represented by the node and one or more other components of the product represented by the one or more other nodes;
• and
• feeding back the assembly graph with the updated current state including the identified link to the graph encoder for updating, at a next recurrent step, the node embedding of each node of the assembly graph using the updated current state of the assembly graph;
• and
• continuing the updating, identifying, and adding until at least one link with another node is identified for every node of the assembly graph, thereby the generated assembly sequence is used to assemble the product.