3D Visual Tracking of an Articulated Robot in Precision Automated Tasks

Abstract

The most compelling requirements for visual tracking systems are a high detection accuracy and an adequate processing speed. However, the combination between the two requirements in real world applications is very challenging due to the fact that more accurate tracking tasks often require longer processing times, while quicker responses for the tracking system are more prone to errors, therefore a trade-off between accuracy and speed, and vice versa is required. This paper aims to achieve the two requirements together by implementing an accurate and time efficient tracking system. In this paper, an eye-to-hand visual system that has the ability to automatically track a moving target is introduced. An enhanced Circular Hough Transform (CHT) is employed for estimating the trajectory of a spherical target in three dimensions, the colour feature of the target was carefully selected by using a new colour selection process, the process relies on the use of a colour segmentation method (Delta E) with the CHT algorithm for finding the proper colour of the tracked target, the target was attached to the six degree of freedom (DOF) robot end-effector that performs a pick-and-place task. A cooperation of two Eye-to Hand cameras with their image Averaging filters are used for obtaining clear and steady images. This paper also examines a new technique for generating and controlling the observation search window in order to increase the computational speed of the tracking system, the techniques is named Controllable Region of interest based on Circular Hough Transform (CRCHT). Moreover, a new mathematical formula is introduced for updating the depth information of the vision system during the object tracking process. For more reliable and accurate tracking, a simplex optimization technique was employed for the calculation of the parameters for camera to robotic transformation matrix. The results obtained show the applicability of the proposed approach to track the moving robot with an overall tracking error of 0.25 mm. Also, the effectiveness of CRCHT technique in saving up to 60% of the overall time required for image processing.

Emotion-Driven Analysis and Control of Human-Robot Interactions in Collaborative Applications

“…As an example, and in a recent publication, Ionescu in [ 32 ] demonstrates the usage of the concept of reinforcement learning in an Adaptive Simplex Architecture (ASA) approach for collision avoidance with human. Furthermore, the implementations of new technologies in vision systems using high speed cameras and 3D cameras can enhance the safety of the human worker as depicted in [ 33 , 34 ]. Furthermore, the use of cloud services like Amazon Web Services (AWS) can simplify the integration of AI-based systems in the domain of HMI and HRC.…”

Design and Construction of a Cost-Effective Didactic Robotic Arm for Playing Chess, Using an Artificial Vision System

“…The first stage of the corner detection algorithm consists of finding straight lines in the image. For this, the Canny edge detector [40][41][42] and the Hough transform [43][44][45][46] were used, which are available in the OpenCV library.…”

Pixel-Wise Crack Detection Using Deep Local Pattern Predictor for Robot Application

“…Amatya et al developed a method of locating shaking positions based on objects pixel locations in the images [ 8 ]. An enhanced CHT [ 9 ] is employed for estimating the trajectory of a spherical target in three dimensions to improve tracking accuracy. For robotic navigation in unstructured environments, a method based on ToF cameras [ 10 ] was provided for 3D obstacle detection and classification.…”