Disturbance-Observer-Enhanced Adaptive Fast Terminal Sliding Mode Control for High-Precision Industrial Manipulation

Abstract

High-precision robotic manipulation in industrial cyber–physical systems requires control strategies capable of maintaining smooth, disturbance-resilient motion under model uncertainties, payload variations, and external perturbations. This need is especially pronounced in automated liquid-handling workflows such as reagent transfer, pipetting, and micro-dispensing, where even small trajectory deviations can induce fluid oscillations, cross-contamination, or volumetric inaccuracies. This work proposes a nonlinear disturbance-observer-enhanced Model-Based Sliding Mode Control (MBSMC) architecture that integrates an energy-aware momentum observer with a Momentum-Adaptive Non-Singular Fast Terminal Sliding Mode Controller (MA-NFTSMC). The observer provides real-time estimation of system momentum and unmodeled dynamics without relying on acceleration measurements, while the adaptive sliding manifold ensures finite-time convergence and suppresses chattering through dynamic gain modulation. The unified formulation explicitly compensates for matched disturbances and unmodeled dynamic uncertainties, establishing composite stability through observer error analysis and finite-time tracking convergence on the sliding manifold. Simulation and hardware experiments on a 6-DOF industrial manipulator performing representative liquid-handling motions demonstrate significantly smoother joint trajectories, superior disturbance rejection, and up to 90% reduction in control effort compared to PID and non-model-based SMC baselines. The proposed architecture offers a scalable, computationally efficient, and robust control solution for next-generation intelligent manufacturing systems, including high-throughput chemical processing, laboratory automation, and precision fluid-handling applications.