This work validates NeuralActuator on a 5-DoF OpenManipulator-X, a 6-DoF SO-101 from LeRobot, and a 7-DoF Franka Emika Panda, spanning three actuator families and costs from approximately \$500 to more than \$30{,}000.
Abstract
Differentiable simulators have advanced policy learning and model-based control across robotic tasks. Yet actuator dynamics remain underexplored and can be a major source of sim-to-real error, particularly on low-cost platforms, where the linear current-to-joint-torque approximation $\tau = K_t I$ becomes unreliable because of friction, hysteresis, backlash, and thermal effects. Accurate actuator models can also support force perception and integrated force/position control. We present NeuralActuator, which jointly predicts (i) a torque surrogate for trajectory propagation on low-cost servo platforms, (ii) external forces with a contact-probability gate for sensorless force perception, and (iii) a motor-condition score for a supervised joint, distinguishing normal from mechanically restricted operation. A twin-arm teleoperation system records robot states and actuator telemetry alongside external-force labels, yielding the Neural Actuation Dataset (NAD). The torque-surrogate head is trained through differentiable simulation from pose trajectories without ground-truth joint-torque measurements. A Transformer captures temporal dependencies while enabling real-time inference. We validate NeuralActuator on a 5-DoF OpenManipulator-X, a 6-DoF SO-101 from LeRobot, and a 7-DoF Franka Emika Panda, spanning three actuator families and costs from approximately \$500 to more than \$30{,}000. The low-cost platforms support physically plausible dynamics and force evaluation, while the offline Franka experiment provides a payload-force-estimation benchmark. We also demonstrate motor-condition estimation and improved behavior-cloning performance using NeuralActuator as a pretrained module. We release the dataset, code, and hardware configurations on the project page: https://frank-zy-dou.github.io/projects/NeuralActuator/index.html.
The demand for robotic manipulators has increased because of their precision, speed, and cost-efficiency in complex or hazardous tasks. Flexible robotic manipulators, unlike rigid ones, offer lower mass and energy consumption, enabling advanced applications across various fields. Despite these advantages, the mass reduction of these manipulators can lead to undesired effects, including decreased precision, increased sensitivity to parametric uncertainties, coupled dynamics, and increased oscillations caused by their inherent flexibility. Moreover, the modeling complexity of such mechanical systems represents a significant challenge, since multiple degrees of freedom must be considered. In this study, a physics-informed neural network (PINN) is designed to estimate the dynamic behavior of a flexible-link manipulator. First, a dataset is created by executing different trajectories (i.e., different rotation angles) of the flexible manipulator. Based on the dataset, the PINN is then trained using time and strain signals as inputs to estimate the angular displacement, combining a data-driven loss with a physics-based loss derived from the system’s dynamic model. Finally, the PINN model is investigated experimentally to assess the closed-loop strategy and evaluate its efficiency and reproducibility in recognizing the mechanical system behavior. Therefore, the results show that the PINN and closed-loop experimental validations are consistent with the proposed method.
Tony Jun Tanaka, R. Geronel, R. Teloli et al.· Machines· 0 citations
Accurate trajectory tracking in cable-driven lower-limb rehabilitation robots is challenging because model uncertainty, external disturbances, joint constraints, and pull-only cable actuation can degrade nominal control performance. Conventional model-based controllers provide an interpretable control structure but remain sensitive to model mismatch, whereas fully learning-based control can reduce transparency and complicate constraint-aware operation. This study proposes a residual deep reinforcement learning-enhanced computed torque control framework in which computed torque control generates the nominal command and a bounded Deep Deterministic Policy Gradient policy supplies only an additional compensating torque. The approach is evaluated in simulation under nominal, uncertain, disturbed, combined, and generalization conditions, together with trajectory-tracking, joint-limit, cable-demand, workspace-feasibility, and cable-Jacobian diagnostics. Across the evaluated conditions, the residual controller improves tracking and disturbance rejection relative to computed torque control while preserving the interpretable model-based command structure and satisfying the reported feasibility checks in the representative evaluation. Broader tests indicate that tracking improvements can persist beyond the representative case while also exposing trajectory-dependent constraint limitations. These results support bounded residual learning as a practical robustness-enhancement strategy for simulation-based rehabilitation robot control and motivate further constraint-aware and experimental validation.
Accurate dynamic modeling of industrial robots is essential for high-performance control and estimating torque. However, traditional physics-based models often fail to capture unmodeled dynamics such as complex friction, payload variations, and gearbox-induced distortions. This article refines a physics-based modeling framework with a data-driven residual learning policy to compensate for unmodeled dynamics. We first employ a nested optimization strategy using multistart sequential quadratic programming (MS-SQP) and QR-based base parameter extraction to identify identifiable inertial and Stribeck friction parameters. To compensate for remaining systematic errors, we integrate the residual error policy model using physics-aware feature engineering. The proposed method is validated on both synthetic datasets and a real-world six-DOF industrial manipulator across various trajectories (including Fourier, chirp, and trapezoidal excitations) and payload conditions. A friction modeling ablation study confirms that richer friction representations progressively reduce the residual space, with the Stribeck backbone achieving 2.32-Nm root-mean-square error (RMSE) compared to 3.89 Nm for the rigid-body-only baseline. Experimental results demonstrate that the proposed physics-imbued residual framework improves torque prediction RMSE 31.5 % in real-world trajectories and 31.7 % in unseen synthetic payload testing. The results confirm that the framework effectively captures complex unmodeled dynamics while maintaining physical interpretability and generalization capabilities essential for industrial instrumentation and measurement applications.
Unmanned aerial vehicles (UAVs), particularly quadcopters, present unique challenges for autonomous control due to their underactuated dynamics: only four available control inputs must govern six degrees of freedom. This paper investigates a physics-aware, end-to-end deep reinforcement learning (DRL) approach that acts directly on low-level body inputs, total thrust and body torques $(T, \tau_x, \tau_y, \tau_z)$, and closes the loop through a high-fidelity Simulink environment. Our simulator integrates a 12-state rigid-body model (MATLAB Level-2 S-Function) with (i) an Action2RPM allocation based on the Moore-Penrose pseudo-inverse of a coefficient matrix derived from thrust and drag terms, and (ii) first-order actuator dynamics for each motor (time constant $T_m = 0.076$ s), including rotor gyroscopic coupling. A shaped reward balances goal-reaching and stability using an exponential position well, attitude penalties, and quadratic velocity costs. Four DRL algorithms, DDPG, TD3, PPO, and SAC, are evaluated in two stages: (S1) thrust-only hover and (S2) hover with pitch torque and a translated goal. Results show that SAC and TD3 achieve superior stability and exploration efficiency, while PPO is less sample-efficient. The study highlights the significance of modeling actuator lags and aerodynamic moments for stable low-level control and provides a reproducible benchmark for quadcopter DRL.
High-fidelity embodied AI simulators provide realistic evaluation of complex robotic systems, but their computational cost limits their direct use for large-scale reinforcement learning campaigns. We advocate the use of less accurate but more expeditious simulations, which might draw on data-driven, e.g., neural dynamics, models. This contribution argues that the practical value of a neural dynamics model for complex robot control lies in learning the \emph{right abstraction}: a reduced state that preserves the control-relevant physics of the high-fidelity system while enabling high-throughput policy learning. We develop a neural reduced dynamics (NRD) framework that separates the state the model propagates from what can be supplied as an input or recovered analytically, trains policies entirely inside the frozen learned model, and validates them back in the high-fidelity simulator. Two case studies instantiate it across three control tasks: terrain-aware HMMWV trajectory tracking on rigid, bumpy and deformable Continuum Representation Model (CRM) terrain; and goal reaching for a stock tracked vehicle and its front-mounted articulated arm. Every policy transfers back to the high-fidelity simulator. A single policy trained inside the terrain-conditioned dynamics model, and given no terrain input of its own, attains lower median and mean tracking error than both single-terrain specialists on all three terrains, including zero-shot bumpy terrain. Quantitatively, the tracked vehicle reaches 100 of 100 goals and the arm 97 of 100, with zero contacts or joint-limit violations. The NRD models advance roughly four orders of magnitude faster in simulated time than the high-fidelity simulator scenes they replace, making iterative on-policy learning practical and supporting neural reduced dynamics as a bridge between accurate but expensive physics simulation and scalable robot learning.
Accurate actuator modeling is critical for robust design validation and sim-to-real control transfer in humanoid robotics. Yet, in practice, developers rely on simplified actuator models built from sparse datasheets or offline system identification, which often omit internal control logic, saturation, sensor dynamics, and electromechanical actuator dynamics. This limits model fidelity under changing conditions and contributes to sim-to-real failures. We propose actuator Digital Twins (DTs) as a scalable solution for predictive simulation. In this work, predictive simulation is defined as the forward computation of joint position and actuator torque from prescribed reference trajectories, controller parameters, mechanical configuration, and initial conditions, with prediction accuracy evaluated against measurements from the physical actuator. We validate a DT of the Pulsar PULSE115 quasi-direct-drive actuator that reproduces the actuator electromechanical dynamics, physical operating limits, sensing characteristics, and embedded cascaded controller executed at 10 kHz on a 1-DOF pendulum testbed, comparing real-world experiments with simulations using both the DT and a simplified model. Across varying trajectories and configurations, the DT maintains low error-from-real, while the simplified model degrades outside its tuned regime, particularly under changes in trajectory dynamics, mechanical load, and controller gains. We further embed the DT in a 4-DOF humanoid arm simulation and show that it runs significantly faster than the real-time version, achieving a simulation speedup factor of approximately 6.3× on a standard laptop. These results demonstrate that actuator-specific electromechanical and embedded control modeling improves the forward prediction of physical actuator behavior while remaining computationally practical for multi-joint robotic simulation.