design feedback controllers

Low-cost Stewart platforms lack integrated nonlinear control and state estimation capabilities for research and education. Method: This work designs and implements a complete hardware-software experimental platform featuring a lightweight mechanical structure, Lagrangian-based dynamic modeling, a robust trajectory-tracking controller combining feedback linearization and LQR, and high-precision real-time state estimation via extended Kalman filtering (EKF) fusing IMU and encoder measurements. Contribution/Results: The platform demonstrates excellent accuracy, noise resilience, and robustness in both static positioning and dynamic trajectory-tracking experiments. Notably, it achieves, for the first time at a hardware cost under $100, a full-stack closed-loop integration—from modeling and estimation to control—addressing the prevailing gap in existing studies that emphasize isolated components over system-level synergy. It provides a reproducible, extensible, open-source benchmark platform for robotics control education and advanced algorithm validation.

To address state instability and discontinuous tracking arising from output-function switching in feedback linearization of nonlinear systems, this paper proposes a dynamic switching control framework based on the concept of “meld”—a maximal linearizable output subset. We formally define meld and establish its compatibility conditions and minimum dwell-time constraint. Under this constraint, we rigorously prove uniform boundedness of the closed-loop system states and seamless tracking of a common output across switching instants. The method integrates feedback linearization, dynamic output selection, and Lyapunov-based stability analysis, ensuring broad applicability. Numerical simulations demonstrate exponential convergence of tracking errors for all active outputs, validating the framework’s efficacy on canonical nonlinear systems including robotic manipulators and aerial vehicles.

This work addresses the challenge of uncertain joint stiffness in flexible-joint robots, which arises from time-varying and aging characteristics of elastic components and significantly degrades model-based control performance. To overcome this issue, a novel adaptive control method is proposed that continuously estimates and updates the nonlinear torque–deflection relationship of each joint online. By integrating an implicit control law with an input-dependent regressor matrix, the approach transcends the conventional limitations of adaptive control frameworks designed for non-elastic systems. The method substantially enhances robustness and tracking accuracy for position-controlled flexible-joint robots under varying stiffness and motor positioning errors. Experimental validation on a platform exhibiting nonlinear stiffness characteristics confirms the efficacy of the proposed strategy.

Traditional robot design and control are typically decoupled, leading to morphologies poorly aligned with task requirements. This paper proposes a simulation-driven co-optimization framework for morphology and control, breaking the conventional “design-then-control” paradigm to enable task-oriented, end-to-end joint search. Our method employs gradient-free optimization to simultaneously evolve structural parameters and controller policies within a URDF-based multi-task reinforcement learning simulation environment. Key contributions include: (1) demonstrating that controller retraining significantly improves performance, yielding an average gain of 37%; and (2) revealing an inverse correlation between morphological complexity and controller training budget—providing theoretical justification for structural simplification under resource constraints. We validate the framework across four public simulation benchmarks, showing that co-optimization consistently yields more compact, robust, and task-adapted robot morphologies compared to sequential approaches.

This study addresses the challenge of residual end-effector oscillations in underactuated planar robots during low-speed trajectory tracking, which arises from difficulties in accurately modeling passive joints. To mitigate this issue, the authors propose a method that integrates differential flatness with optimal control, explicitly incorporating the minimization of passive joint potential energy into a quadratic performance index. By jointly optimizing control torques and potential energy, the approach generates trajectories driven by flat outputs. The resulting control strategy demonstrates robustness to variations in passive joint stiffness and damping parameters, effectively alleviating oscillations caused by model-parameter mismatches. Simulation results confirm that the proposed method significantly suppresses residual oscillations, thereby enhancing both trajectory tracking accuracy and overall system robustness.

This work addresses the challenge of modeling tendon-driven continuum robots, whose dynamics are highly nonlinear, high-dimensional, and dominated by friction. To overcome these difficulties, a data-driven system identification approach is proposed, integrating the N4SID, ARX, and SINDYc algorithms to construct a low-dimensional dynamical model. The study reveals that the dynamic behavior of multi-segment continuum robots can be accurately captured by a two-degree-of-freedom model, uncovering strong kinematic coupling among joints and substantially reducing modeling complexity. Experimental validation demonstrates that the resulting model achieves high prediction accuracy and has been successfully implemented within a real-time model predictive controller, confirming its effectiveness and practical utility.

This work addresses the challenging joint tuning of disturbance observer and feedback controller parameters, which are strongly coupled, by proposing MetaTune—a framework that enables adaptive co-adaptation of both sets of parameters through differentiable closed-loop meta-learning. The approach integrates transferable neural policies with physics-informed gradients and introduces, for the first time, an efficient adjoint-based meta-gradient computation mechanism that reduces computational complexity to linear scaling with the data horizon length, thereby enabling zero-shot simulation-to-simulation transfer. Evaluated on quadrotor control tasks, the method achieves over 50% reduction in gradient computation time, decreases average tracking error by 15–20% during high-speed flight, and improves performance by up to 40% under strong disturbances.

This work addresses the challenge of reference trajectory tracking for uncertain nonlinear systems with limited data by proposing a meta-learning-based control framework. The approach learns a shared dynamic representation from structurally similar source systems during an offline phase and enables rapid adaptation of the controller to a new target system using only a few online data samples. Innovatively adapting implicit Model-Agnostic Meta-Learning (iMAML) to the control domain, the method establishes a general bilevel optimization framework compatible with diverse learning algorithms while significantly reducing memory overhead and approximation error. Two implementation pathways—neural state-space models and deep Q-networks, corresponding respectively to explicit and implicit system identification—are evaluated through simulations and hardware experiments, consistently demonstrating superior control performance over baseline methods and confirming the framework’s effectiveness and practicality.

This study addresses the challenging problem of partial differential equation (PDE)-based control for spatially arbitrary-configured flexible multi-body robotic arms by proposing a modular, subsystem-based control framework. Leveraging screw theory within the se(3) Lie algebra, the approach unifies the modeling of rigid motion, elastic deformation, and internal forces, while introducing a strain-based PDE formulation to eliminate distributed elastic acceleration terms. An inverse kinematics solution compensating for deformation is integrated with task-space end-effector trajectory generation to design both nominal and adaptive controllers. Exploiting the frame invariance of power pairings on se(3) and Newton’s third law, the method achieves exact cancellation of inter-link interaction power, enabling the first proof of global exponential stability for flexible arms of arbitrary length. Theoretical analysis guarantees exponential convergence of twist and parameter estimation errors and bounded elastic deformation, with simulations confirming physical consistency and superior control performance.