Calibration and Registration Tools

The National Institute of Standards and Technology is developing a suite of tools to ease the processes of calibration and registration for robot users.  

Robots are not typically very accurate although many have very good repeatability.  Robot part wear, temperature changes, and other factors cause the accuracy of the robot to degrade over time.   Collaborative robots, which are lighter and easily repositioned, may need to be re-aligned with respect to their workplace more often.  These techniques for improving the robot’s accuracy with respect to its task space and work pieces – referred to as calibration – can be complicated.  Yet, failure to perform these alignments can hinder the robot’s performance and even cause process failures and damage to the robot or parts.  

Similar categories of errors exist for sensors:  environmental (e.g., lighting), parametric (such as lens distortion due to manufacturing errors), measurement (caused by limited resolution of the sensor or limited frequency of data capture), computational (e.g., algorithmic limitations, round-off), and application (e.g., reflectivity and other properties of the parts; field of view being offset by being bumped).   

Another term related to calibration that is frequently used with respect to robots is “registration.”  Registration refers to the process of measuring and mapping one system’s data to the model of another, correcting for differences in resolution, scale, direction, and timing.  This process could address, for example, position data, coordinate frame data, raw sensor data, and more complex world models.  

Calibration

• enhances robot positioning accuracy
• enhances mobile robot navigation and docking accuracy
• enables vision systems to accurately report object identification and localization information
• ensures that force sensors within the robotic system know the difference between a slight tap on its human partner and an impact that could result in injury

Registration

• maps sensor outputs such as images or torque measurements to the kinematic and dynamic profiles of the robot, and vice versa
• For multi-robot systems, maps the coordinate frames and motion characteristics of the robots to one another