<p>Robotic manipulators are increasingly being used for performing a wide variety of manufacturing processes. Some of the manufacturing processes performed using robotic manipulators require high trajectory execution accuracy. Automatically generated trajectories often exhibit significant execution errors due to robot model inaccuracies and controller behaviors. This suggests that a trajectory compensation scheme can be used to modify trajectory to reduce execution error. Unfortunately, errors during trajectory tracking are dependent on the nature of the trajectory, the end-effector load, and therefore a trajectory-independent automated compensation scheme for reducing execution errors does not always perform well. This paper presents a method for sampling the input trajectory, executing an exploratory physical run using the sampled trajectory, and learning a compensation scheme based on the measured data. The learned compensation scheme is trajectory-dependent and is capable of reducing execution error. The proposed compensation scheme in this work applies to the top of the existing controller. It does not interfere with the existing controller's working and does not need continuous feedback from the controller. So it can be easily utilized in a commercial {manufacturing} environment without any safety concerns to obtain lower trajectory execution errors. We propose a novel learning-based (artificial neural networks (ANN)) trajectory compensation scheme to reduce trajectory execution errors arising due to the controller. The compensation of controller error will be effective only if the kinematic model of the manipulator is accurate. So before implementing our approach, we need to calibrate the kinematic model and then reduce the trajectory execution error. To demonstrate the compensation scheme's effectiveness, we perform experiments on manipulators. After the trajectory compensation, the manipulator has considerably low trajectory execution errors, with the average path error close to the robot's repeatability.&nbsp;</p>

Trajectory-Dependent Compensation Scheme to Reduce Manipulator Execution Errors for Manufacturing Applications

Paper Proceedings: https://asmedigitalcollection.asme.org/MSEC/MSEC2021/volume/85062

2021 International Manufacturing Science and Engineering Conference (MSEC2021)

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.