Hand-eye calibrations are procedures that estimate the pose of the sensor, usually an RGB camera, w.r.t. a coordinate frame of interest, such as the base or the end-effector of a robotic manipulator. These procedures are referred to as extrinsic calibrations. However, they operate under the assumption that the description of the kinematic chain of the robotic manipulator is accurately known, which may be unrealistic. In fact, there are calibration procedures that estimate the parameters of the kinematic chain in robotic manipulators, referred to as kinematic calibrations. This paper proposes a calibration method that is able to carry out the simultaneous extrinsic and kinematic calibrations of any robotic system. This approach produces complex optimization problems, which can be aided by the inclusion of additional observations of the kinematic chain. As such, the second focus in this work is to explore how the number and positioning of calibration patterns affects the accuracy of calibration procedures. Experiments show that the proposed system is able to produce accurate estimates for the extrinsic and kinematic parameters, using a single calibration procedure. More importantly, results also show that these complex calibrations clearly benefit from the usage of multiple calibration patterns. • A simultaneous extrinsic and kinematic calibration mechanism is presented. • Integration into a general framework for calibration. • An assessment of the impact of using several calibration patterns on the calibration. • The use of multiple calibration patterns enhances accuracy of calibration procedures.
Vieira et al. (Sun,) studied this question.