This paper presents a new combined 9-DOF [3R-3R-3R] parallel architecture to perform the object catching in the real-time scenario. The architecture design is intended to carry heavy payload objects for the catching. This paper covers modelling the new architecture, kinematics, and dynamic analyses to compute torques/forces at the actuators. The simulation of kinematic and dynamic analysis in MATLAB. Vision sensors, encoders, PID controller, and current limiting are used to perform object catching. The architecture is the only delta robot application designed to catch regular-shaped objects in real-time scenarios.

This paper proposes a vision-based kinematic analysis and kinematic parameters identification of the proposed architecture, designed to perform the object catching in the real-time scenario. For performing the inverse kinematics, precise estimation of the link lengths and other parameters needs to be present. Kinematic identification of Delta based upon Model10 implicit model with ten parameters using the iterative least square method is implemented. The loop closure implicit equations have been modelled. In this paper, a vision-based kinematic analysis of the Delta robots to do the catching is discussed. A predefined library of ArUco is used to get a unique solution of the kinematics of the moving platform with respect to the fixed base. The re-projection error while doing the calibration in the vision sensor module is 0.10 pixels. Proposed architecture interfaced with the hardware using the PID controller. Encoders are quadrature and have a resolution of 0.15 degrees embedded in the experimental setup to make the system closed-loop (acting as feedback unit).