Advanced SOA-NMPC controller design minimising real-time computational burden for dynamic obstacle avoidance in robotic manipulators

Abstract

In this research, a new advanced Spatial Operator Algebra (SOA) and Nonlinear Model Predictive Control (NMPC) method is proposed to address challenges in dynamic obstacle avoidance for robotic manipulators while minimising real-time computational burden. The computational limitations of conventional NMPC methods stem from high dimensionality, nonlinearity and non-convexity, necessitating powerful processors with large memory. These limitations lead to reduced control horizon and suboptimal performance, particularly in real-time dynamic obstacle avoidance applications. To address these challenges, a novel hybrid control method is proposed by integrating SOA, which offers significant advantages in modelling efficiency and computational cost, with NMPC to enhance its predictive and control capabilities. The SOA-NMPC controller enables efficient autonomous path planning and obstacle avoidance while ensuring compliance with constraints. The validity of the proposed method is verified through experimental studies involving various collision scenarios, including static and dynamic obstacles. After verification, an additional collision experiment is conducted to investigate its ability to minimise real-time computational burden. The results confirm the SOA-NMPC method's robust planning-tracking performance and collision-free operation. Controller stability is demonstrated using the Lyapunov method. The computational efficiency of the SOA-NMPC enables more flexible prediction and control horizon selection, overcoming traditional limitations and enhancing dynamic obstacle avoidance performance.