Autonomous performance optimization in robotic assembly process

1. A method for process parameter optimization in a robotic manufacturing process, comprising:

identifying, by a computational device, in two or more successive iterations during the robotic manufacturing process, a system model for the robotic manufacturing process; and

optimizing one or more manufacturing process parameters based on the model identified wherein, for at least one of the iterations, the model is identified through a Gaussian process regression—surrogated Bayesian optimization algorithm, and wherein, for at least one other of the iterations, optimization is performed and applied to the manufacturing process on-line based on the model identified through the Gaussian process regression—surrogated Bayesian optimization algorithm while the manufacturing process is being performed, wherein online parameter optimization through the Gaussian process regression—surrogated Bayesian optimization algorithm comprises adding a random variation factor to a lower confidence bound acquisition function.

2. The method of claim 1 , wherein, for at least one of the iterations, optimization is performed and applied to the manufacturing process while the manufacturing process is being performed on-line and without interruption of the manufacturing line.

3. The method of claim 1 , wherein the manufacturing process comprises at least one assembly process, wherein optimizing on or more manufacturing process parameters comprises optimization of at least one process parameter for an assembly process.

4. The method of claim 1 , further comprising evaluating the manufacturing process, wherein evaluating the manufacturing process comprises performing the robotic manufacturing process using candidate process parameters.

5. The method of claim 4 , wherein, for at least one the candidate process parameters the robotic manufacturing process uses for the evaluation of the manufacturing process are generated from the identified model.

6. The method of claim 4 , wherein, for at least one the candidate process parameters the robotic manufacturing process uses for the evaluation of the manufacturing process are initial process parameters.

7. The method of claim 4 , wherein the robotic manufacturing process is terminated after a pre-determined cut-off time.

8. The method of claim 1 , wherein, for at least one of the iterations, identifying the model comprises balancing exploration of a model and exploitation of an identified model.

9. The method of claim 1 , further comprising, for at least one of the iterations, determining whether to optimize one of the process parameters or update the model.

10. The method of claim 1 , applying a switching criterion to determine whether to optimize one of the process parameters or update the model.

11. The method of claim 1 , further comprising, for at least one of the iterations: determining a balance factor; and using the balance factor to determine whether to optimize one of the process parameters or update the model.

12. The method of claim 1 , further comprising restarting optimization if there is a disturbance in the manufacturing process.

13. The method of claim 1 , further comprising restarting optimization in response to variations in the characteristics of parts being assembled by the robotic manufacturing process.

14. A system, comprising:

a processor;

a memory coupled to the processor, wherein the memory comprises program instructions executable by the processor to implement:

identifying, by a computer system, in two or more successive iterations during a robotic manufacturing process, a system model for the robotic manufacturing process; and

optimizing one or more manufacturing process parameters based on the model identified, wherein, for at least one of the iterations, the model is identified through a Gaussian process regression—surrogated Bayesian optimization algorithm, wherein online parameter optimization through the Gaussian process regression—surrogated Bayesian optimization algorithm comprises adding a random variation factor to a lower confidence bound acquisition function; and

wherein, for at least one other of the iterations, optimization is performed and applied to the manufacturing process on-line based on the model identified through the Gaussian process regression—surrogated Bayesian optimization algorithm while the manufacturing process is being performed.

15. A non-transitory, computer-readable storage medium comprising program instructions stored thereon, wherein the program instructions are configured to implement:

identifying, by a computer system, in two or more successive iterations during a robotic manufacturing process, a system model for the robotic manufacturing process; and

optimizing one or more manufacturing process parameters based on the model identified in the iterations; wherein, for at least one of the iterations, the model is identified through a Gaussian process regression—surrogated Bayesian optimization algorithm, wherein online parameter optimization through the Gaussian process regression—surrogated Bayesian optimization algorithm comprises adding a random variation factor to a lower confidence bound acquisition function; and

wherein, for at least one other of the iterations, optimization is performed and applied to the manufacturing process on-line based on the model identified through the Gaussian process regression—surrogated Bayesian optimization algorithm while the manufacturing process is being performed.