Generating Robot Motions for the Adhesive Coating System using a Predictive Model of Material Behavior

2nd Report : Predicting Behavior using a Linear Autoregressive Model

Abstract

In the automation of adhesive coating, a problem is that the adhesive is pulled by the robot during coating, which prevents the robot from applying the adhesive according to the target trajectory. To solve this problem, our method uses a linear autoregressive model to learn the relationship between the robot trajectory and the adhesive trajectory. Then, using the adhesive trajectory simulated by the model, the appropriate robot trajectory is determined to match the target adhesive trajectory. The experiment confirms the results of the coating operation on the actual machine to which the proposed method is applied.