| Summary: | The purpose of this research is to analyze and upgrade the performances of the Baxter intelligent robot, through data mining methods. The case study belongs to the robotics domain, being integrated in the context of manufacturing execution systems and product lifecycle management, aiming to overcome the lack of vertical integration inside a company. The explored data comprises the parameters registered during the activities of the Baxter intelligent robot, as, for example, the movement of the left or right arm. First, the state of the art concerning the data mining methods is presented, and then the solution is detailed by describing the data mining techniques. The final purpose was that of improving the speed and robustness of the robot in the production. Specific techniques and sometimes their combinations are experimented and assessed, in order to perform root cause analysis, then powerful classifiers and metaclassifiers, as well as deep learning methods, in optimum configuration, are analyzed for prediction. The experimental results are described and discussed in details, then the conclusions and further development possibilities are formulated. Based on the experiments, important relationships among the robot parameters were discovered, the obtained accuracy for predicting the target variables being always above 96%.
|
|---|
