Nowadays, collaborative human-robot assembly processes become more and more important to enable the needed flexibility in production. Therefore, it is crucial that robots understand their environment and for doing so they should know about the position and orientation of objects, human hands, and the robot gripper itself. Currently, optical systems together with complex image processing algorithms are used for these localization tasks, but the ultrasound and infrared based opto-acoustic localization system presented in this contribution builds an economical alternative for many use cases. In addition to a simultaneous tracking of multiple objects, it also allows the unique identification of objects that look similar for an optical system (e. g. two screw drivers with different bits held in place) by sending virtually orthogonal codes in a code division multiple access environment. The opto-acoustic system is based on unilateral time-of-flight distance measurements between room-fixed receivers and mobile transmitters attached to the objects which are to be tracked. In addition, data from an accelerometer, a gyroscope, and a magnetometer included in the transmitter are integrated into the positioning system by using a particle filter. The system is tested by attaching one transmitter to a collaborative robot. Various trajectories with different dynamics are followed by the robot and the pose tracking performance is evaluated compared to an optical reference system. The median localization error is below 3.7 cm and 2.0° across all trajectories and dynamics even when going far beyond typical human hand movement dynamics in assembly scenarios.
Paper
Full text
Combined opto-acoustic and inertial 3D object localization in assembly
Semantic Scholar · Engineering · 2019
Abstract
Nowadays, collaborative human-robot assembly processes become more and more important to enable the needed flexibility in production. Therefore, it is crucial that robots understand their environment and for doing so they should know about the position and orientation of objects, human hands, and the robot gripper itself. Currently, optical systems together with complex image processing algorithms are used for these localization tasks, but the ultrasound and infrared based opto-acoustic localization system presented in this contribution builds an economical alternative for many use cases. In addition to a simultaneous tracking of multiple objects, it also allows the unique identification of objects that look similar for an optical system (e. g. two screw drivers with different bits held in place) by sending virtually orthogonal codes in a code division multiple access environment. The opto-acoustic system is based on unilateral time-of-flight distance measurements between room-fixed receivers and mobile transmitters attached to the objects which are to be tracked. In addition, data from an accelerometer, a gyroscope, and a magnetometer included in the transmitter are integrated into the positioning system by using a particle filter. The system is tested by attaching one transmitter to a collaborative robot. Various trajectories with different dynamics are followed by the robot and the pose tracking performance is evaluated compared to an optical reference system. The median localization error is below 3.7 cm and 2.0° across all trajectories and dynamics even when going far beyond typical human hand movement dynamics in assembly scenarios.