A Physics-Based Design-Stage Digital Twin for Time-Energy Trajectory Optimization of an Industrial Robotic Manipulator

This study addresses limitations in existing trajectory optimization approaches for industrial robotic manipulators, which often lack integration with high-fidelity physics-based modeling for design-stage decision-making. The proposed framework is formulated as a high-fidelity virtual replica of a four-degree-of-freedom robotic arm, integrating Denavit-Hartenberg kinematics, Newton-Euler dynamic modeling, and CAD-derived inertial parameters to reproduce time-resolved joint states and actuator torques under realistic operating constraints. Within this model-driven environment, Particle Swarm Optimization (PSO) is embedded as a decision-support layer to iteratively adjust the durations of acceleration, cruise, and deceleration in trapezoidal joint velocity profiles. A multi-objective fitness function is used to balance mechanical energy consumption and execution time while enforcing torque, velocity, acceleration, and synchronization constraints through a structured penalty mechanism and dynamic weight scaling. The framework is evaluated across five representative motion scenarios, including moderate point-to-point trajectories, wide-angle rotations, high-acceleration seed trajectories, near-actuator-saturation conditions, and single-joint-dominant motions. Across these cases, the digital twin-enabled optimization improves performance relative to baseline trajectories, achieving mechanical energy reductions exceeding 60% and execution-time reductions of up to 50%, with optimized execution times reduced from up to 11.23 s to as low as 4.50 s and energy consumption reduced from approximately 2.14 J to below 1 J depending on the scenario while maintaining constraint compliance. Repeated optimization runs demonstrate stable convergence behavior and consistent performance improvements across diverse motion profiles. Rather than focusing only on algorithmic optimization, the study demonstrates how a physics-based digital twin can support structured, energy-aware, and productivity-oriented decision-making prior to physical implementation. Although the present work is limited to the modeling and optimization layers of an offline digital twin without experimental coupling or real-time synchronization, the results highlight its value as a digital-engineering framework for robotic trajectory design and virtual performance assessment.