Learning software assisted fixtureless object pickup and placement system and method

A system for learning software assisted, at least partially fixtureless object pickup and placement, said system comprising:

• a material handling robot having a gripping element;
• a joining robot having a joining device;
• a machine vision system oriented to view a work area for a number of parts which when joined form an article;
• a controller in electronic communication with said material handling robot, said joining robot, and machine vision system, said controller comprising an assembly table comprising assembly identifiers for a number of articles to be created, a workpiece table comprising parts identifiers associated with assembly identifiers, a target points table comprising desired target points associated with parts identifiers and weighting values associated with said desired target point, and software instructions stored on one or more non-transitory electronic storage devices, wherein said software instructions, when executed, configure one or more processors of said controller to:
• scan, by way of the machine vision system, a reference version of a part for pickup to populate at least some of said target points table with said desired target points;
• scan, by way of the machine vision system, a desired location for said part for pickup to populate a remainder of said target points table with said desired target points;
• scan, by way of the machine vision system, said number of parts in said work area to identify the part for pickup according to an assembly identifier stored at said assembly table and a part identifier stored at said workpiece table and associated with said assembly identifier;
• retrieve desired target points and associated weighting values for said identified part stored at said target points table in association with said parts identifier;
• measure target points for said identified part and stored said measured target points at said target points table in association with respective desired target points such that each measured target point and associated desired target point form a coordinate pair having an associated weighting value;
• perform an iterative analysis to determine a best fit solution for moving said identified part such that each of said measured target points of each coordinate pairs are within a predetermined range of the associated one of said desired target points, wherein said best fit solution is determined by fitting vectors between each of said desired target points and said measured target points of each coordinate pair to develop a solution set, identifying a subset of said solution set having the shortest length vectors and highest of the weighting values associated with the respective coordinate pair, and selecting the best fit solution from said subset having the overall shortest length when said vectors are summed;
• commanding said materials handling robot to move said identified part according to said best fit solution such that said identified part is positioned adjacent to a second part, which is supported within the work area in a partially fixtureless manner such that such that the second part is unfixed in at least one dimension said identified part is secured in three-dimensional space exclusively by said material handling robot in a fixtureless manner;
• and
• commanding said joining robot to join said identified part to said second part while said identified part and said second part remain positioned and secured adjacent to one another.

A method for learning software assisted, at least partially fixtureless object pickup and placement, said method comprising the steps of:

• scanning a part for pickup in a work area by way of a machine vision system to measure multiple target points on said part;
• electronically retrieving desired target points for said part from one or more non-transitory electronic storage devices, wherein each of said desired target points is associated with one of said measured target points to form a coordinate pair and a non-zero weighting value;
• performing an iterative analysis by way of a controller to determine a best fit solution for moving said part such that each of said measured target points of each coordinate pair is within a predetermined range of the said desired target point for said coordinate pair, wherein said best fit solution is determined by fitting vectors between each of said desired target points and said measured target points in said coordinate pair to develop multiple potential solution sets, wherein said best fit solution is a solution from said multiple potential solution sets having the overall shortest length of vectors when said vectors are summed while prioritizing target points associated with higher weighting values;
• and
• commanding a material handling robot, by way of said controller, to move said part according to said best fit solution such that said part is secured in three-dimensional space in a fixtureless manner by said material handling robot adjacent to another part which is otherwise unfixed in at least three cartesian directions.