Trajectory design that magnifies robot joint error based on sensitivity analysis and its origin error measurement using video captured by a camera

For industrial welding robots using the teaching-playback method, repeatability is more important than absolute accuracy. However, when a robot is replaced due to failure or other reasons, the tip position is shifted because the error factors related to tip positioning are different between the robot before and after the replacement. This can impede recovery. To overcome this problem at production sites where industrial robots are used, easy measurement without the use of expensive sensors is required. On the other hand, when the robot is replaced with another robot of the same model, the mechanical errors such as joint stiffness, arm stiffness and link length error are almost the same and their difference can be assumed to be extremely small, but the joint origin error remains as an individual difference and its effect on the tip misalignment is not small. Therefore, it is the joint origin error that has a large influence on tip misalignment as an individual difference. In recent years, inexpensive devices with high image resolution, such as consumer-use cameras and smartphones have become available at low cost and their use is effective to reduce tip misalignment caused by individual differences in robots at manufacturing sites. In this paper, we focus on the sensitivity of the tip position error with respect to the joint origin error and propose a method to estimate the joint origin error by designing a trajectory in which the tip error is magnified so that the joint origin error can be easily measured using movies of the tip position taken by a digital camera. We verify the effectiveness of this method using an experimental robot.