Cloud-assisted IoRT framework for latency-aware analytical kinematic control of a six-DOF robotic manipulator

The Internet of Robotic Things (IoRT) has introduced new possibilities for distributed and scalable robotic control by combining industrial manipulators with cloud-based computation and networked communication infrastructures. However, the deployment of cloud-assisted robotic systems requires accurate kinematic modeling, reliable communication, and latency-aware execution, particularly when kinematic computations are distributed between cloud and edge layers. This study presents an IoRT-based cloud-assisted control framework for a six-degree-of-freedom industrial robotic manipulator, using the PUMA 560 as a benchmark platform. The framework adopts the established analytical forward- and inverse-kinematics formulations of the manipulator as a non-iterative and interpretable computational core rather than proposing a new kinematic formulation. The principal contribution lies in integrating these analytical models into a distributed cloud-edge architecture in which inverse-kinematics computation, evaluation of multiple solution branches, and trajectory generation are performed in the cloud, while command validation, safety monitoring, and motion execution remain at the edge. Candidate inverse-kinematics branches are evaluated according to joint-limit feasibility, proximity to singular configurations, and joint-space continuity. In the proposed architecture, computationally intensive kinematic calculations are performed in the cloud, while motion validation, safety monitoring, and command execution remain at the edge level to ensure operational reliability. MATLAB-based simulations are conducted to assess forward-inverse-forward model consistency, joint-space trajectory continuity, and communication timing under the adopted cloud-edge model. The kinematic consistency evaluation yields root-mean-square position and orientation reconstruction errors of 0.066 mm and 0.206°, respectively, over the representative test poses. These values quantify the internal agreement between the implemented analytical inverse- and forward-kinematics models under the adopted numerical settings and should not be interpreted as physical robot tracking accuracy. Furthermore, the multi-branch inverse-kinematics selection strategy reduces total joint-space motion by approximately 54.3% relative to random branch selection, improving trajectory smoothness and potentially reducing unnecessary joint actuation. The communication analysis shows that the hybrid architecture achieves an average modeled delay of 0.210 s with no timeout events during the evaluated sequences. These findings demonstrate the simulation-level feasibility of integrating analytical kinematic control into a cloud-assisted IoRT architecture while maintaining numerical consistency, effective solution selection, and predictable execution under the evaluated communication conditions.