Subspace predictive control (SPC) is a data-driven control strategy that utilizes input–output measurements to predict future system behavior without requiring explicit model identification. Conventional SPC exhibits vulnerability to an unknown input disturbance, leading to degraded control performance and steady-state errors. To address these limitations, this article proposes a robust SPC method that inherently mitigates the effect of a constant input disturbance by augmenting the state-space representation through the internal model principle (IMP). This augmentation enables the controller to achieve integral action without requiring a separate disturbance observer (DOB) design. The proposed method is implemented in a data-driven framework, where an auxiliary disturbance is introduced into the data-driven algorithm to enhance disturbance rejection. A transfer function analysis verifies that the proposed Robust SPC eliminates a constant disturbance while maintaining the role of a DOB. Experimental validation on a two-inertia system confirms that the proposed method significantly improves reference tracking performance compared to conventional SPC, demonstrating its effectiveness in disturbance rejection without additional modeling complexity.

Accurate dynamics modeling is crucial for achieving precise Ground Reaction Force (GRF) control and high-performance legged locomotion. However, real-world legged systems exhibit strong frictional effects with hysteresis and inter-joint coupling, which conventional static friction models or purely data-driven approaches often fail to capture. In this paper, we propose Mysteric-Net, a novel MIMO hysteretic friction-aware network that combines a Lagrangian-based formulation with a Temporal Convolutional Network (TCN). By embedding the physical laws of Lagrangian mechanics while modeling history-dependent frictional dissipation via the TCN, our framework accurately identifies the system dynamics, including complex friction and coupling effects. This paper demonstrates that the proposed method significantly improves the accuracy of inverse dynamics estimation on a robotic leg. Furthermore, this paper shows that the learned model enables the design of an effective feedforward controller that mitigates friction and enhances tracking performance over conventional baseline methods.

Robotic gantry stages are a prevalent class of industrial robots used for precise positioning tasks in various fields, including semiconductor manufacturing, 3D printing, and automated assembly. However, these systems often exhibit time-varying dynamics because the position of the end-effector (i.e., the payload) shifts the mass/inertia properties. Such dynamic variations are not captured by conventional Linear Time-Invariant (LTI) models, leading to modeling inaccuracies and degraded control performance. Linear Parameter-Varying (LPV) system identification is a more suitable alternative, but existing approaches typically employ a single, fixed basis-function order for all parameters, resulting in excessive model complexity and poor efficiency.This paper presents a novel global LPV system identification method for multi-axis robotic gantry systems, enabling independent basis-function order selection for each parameter. By eliminating unnecessary high-order terms, the method reduces computational overhead and enhances modeling accuracy. Experimental validation on an industrial gantry testbed confirms superior precision and robustness compared to conventional LPV approaches with uniform polynomial orders.

A Multi-Input-Multi-Output (MIMO) system has complex interactions between inputs and outputs, resulting in a coupling effect that can lead to undesired motions. To effectively control MIMO systems, decoupling is necessary. This paper introduces a decoupling method that uses a transformation matrix modeled as a transfer function using Frequency Response Function (FRF). Unlike the conventional decoupling method uses constant transformation matrix obtained through Canonical Polyadic Decomposition (CPD) and shows performance degradation in certain frequency bands, the proposed method creates transformation matrix as a transfer function and demonstrates better decoupling performance across the entire frequency band. The performance of the proposed method is validated through simulations on a hybrid dual-drive gantry stage.

In this paper, the parametric identification is addressed by a kernel-based model with covariance and a novel model order selection algorithm. The kernel-based model is uti-lized for training the sampled frequency response characteristics, which is insufficient for parametric identification because of noisy and discrete data. The kernel-based frequency response model improves the parametric identification by using the high covariance data. In addition, prior knowledge of the model order is essential for parametric identification. This paper proposes a novel model order selection based on the robust stability criterion of disturbance observer (DOB). The effectiveness of the proposed algorithm is verified through numerical simulations under several conditions.