Paint repair process by scenario
A method and associated system provides automated abrasive paint repair using automated abrasive paint repair devices that selectively sand, buff, and polish a substrate in response to received instructions generated by a controller. The controller receives coordinates of each identified defect in the substrate along with parameters describing characteristics of each defect, selects a sanding process, a buffing process, and/or a polishing process based on empirically derived rules established by skilled/expert human operators and the received parameters. The controller outputs instructions to cause the automated abrasive paint repair devices to execute the selected sanding process, buffing process, and/or polishing process using the received parameters. The empirically derived rules and parameters may be stored in a lookup table and/or updated by a machine learning module.
1. A method of automated abrasive repair, comprising:
receiving, by one or more processors, pre-inspection data comprising:
(i) coordinates of a defect in a substrate such that a robot manipulator can bring an end effector mounted on the robot manipulator into close proximity with the defect, wherein the end effector comprises:
i. end effector sensors including at least one camera to collect data comprising at least one of fringe pattern projection, deflectometry, and intensity measurements of diffuse reflected or normal white light; and
ii. automated abrasive paint repair devices comprising at least one of a sanding tool, a buffing/polishing tool, and a compliant force flange;
(ii) at least one of a defect type from a classifier, approximate volumetric information describing the defect, and substrate material information;
moving the robot manipulator to bring the end effector into close proximity with the defect based on the coordinates of the defect;
collecting local in-situ inspection data comprising at least one of fringe pattern projection, deflectometry, and intensity measurements of diffuse reflected or normal white light by triggering the end effector sensors;
receiving, by the one or more processors, parameters describing characteristics of the defect determined from the pre-inspection data and the local in-situ inspection data;
selecting, by the one or more processors, at least one of a sanding process, a buffing process, a polishing process, or combinations thereof to repair the defect based on a parameter selection algorithm that selects parameters from stored empirically derived rules and stored parameters according to empirically engineered parametric functions of the received parameters;
instructing, by the one or more processors, the automated abrasive repair devices to execute the selected process;
repairing the defect by executing, with the automated repair devices, the selected process; and
providing, by the one or more processors, the local in-situ inspection data to a machine learning unit;
using, by the machine learning unit, the local in situ inspection data to create learning updates by updating the stored empirically derived rules, wherein the updated stored empirically derived rules are used to improve future automated abrasive repair actions; and
after repairing the defect:
collecting post-inspection data comprising at least one of fringe pattern projection, deflectometry, and intensity measurements of diffuse reflected or normal white light by triggering the end effector sensors;
receiving, by the one or more processors, the post-inspection data and evaluating the repair of the defect; and
providing, by the one or more processors, the post-inspection data to the machine learning unit along with the pre-inspection data and the local in-situ inspection data; and
using, by the machine learning unit, the post-inspection data, the pre-inspection data, and the local in situ inspection data to create learning updates by updating the stored empirically derived rules, wherein the updated stored empirically derived rules are used to improve future automated abrasive repair actions.
2. The method of claim 1 comprising the further steps of, after evaluating the repair, determining whether the repair is satisfactory and:
in response to determining that the repair is satisfactory, proceeding to the step of providing, by the one or more processors, the post-inspection data to the machine learning unit along with the pre-inspection data and the local in-situ inspection data.
3. The method of claim 1 comprising the further steps of, after evaluating the repair, determining whether the repair is satisfactory and:
in response to determining that the repair is satisfactory, proceeding to the step of providing, by the one or more processors, the post-inspection data to the machine learning unit along with the pre-inspection data and the local in-situ inspection data.
4. The method of claim 1 comprising the further steps of, after evaluating the repair, determining whether the repair is satisfactory and:
in response to determining that the repair is not satisfactory, repeating the steps of moving the robot manipulator to bring the end effector into close proximity with the defect through repairing the defect until the repair is deemed satisfactory.
5. The method of claim 1 comprising the further steps of, after evaluating the repair, determining whether the repair is satisfactory and:
in response to determining that the repair is not satisfactory, repeating the steps of moving the robot manipulator to bring the end effector into close proximity with the defect through repairing the defect until the repair is deemed satisfactory.