Method for estimating the pose of a robotic arm using a camera and calibration pattern

Industrial robots are a key component of Industry 4.0, yet accurately estimating their pose remains challenging—especially when determining the spatial relationship between the tool center point (TCP) and the working space. This study presents a hybrid pose estimation method that leverages an industrial camera mounted on a robotic arm's effector and a calibration pattern positioned within the working frame. To solve the resulting optimization problem, the method integrates Newton's method with a neural network (NN) pre-trained on a full camera model. Comparative experiments with state-of-the-art optimization techniques show that the proposed approach achieves superior performance in terms of both accuracy and speed. Specifically, it yields a mean position error of 0.5 mm and a mean angle error of 0.31 degrees, with a computation time of 0.14 ms. These results suggest that the method offers an efficient and accurate alternative for camera-based pose estimation in industrial settings.