Sensorless human-intent-based power-assist control for indoor cooperative transport via interaction estimation and compliant motion generation

AI-Enabled Force and Torque Control for Human Robot Interaction

Safe and intuitive human robot interaction (HRI) requires precise regulation of contact forces and torques while adapting to dynamic and uncertain human behavior. Traditional impedance and admittance control strategies rely on fixed parameters and accurate system modeling, which often limit their performance in unstructured or collaborative environments. This paper presents an AI-enabled force and torque control framework that integrates machine learning techniques with conventional control methods to enhance adaptability, compliance, and safety in physical human robot interaction. The proposed approach employs deep neural networks and reinforcement learning to learn human intent and interaction dynamics directly from multi-modal sensor data, including force torque sensors, joint encoders, and inertial measurements. By continuously adjusting control gains in real time, the system achieves stable interaction while minimizing excessive contact forces and undesired torques. Experimental evaluations conducted on a collaborative robotic platform demonstrate significant improvements over classical control schemes, including reduced interaction force peaks, smoother torque profiles, and improved task execution efficiency during cooperative manipulation tasks. The results indicate that AI-driven force and torque control can substantially improve robustness, adaptability, and user comfort in human robot collaboration, making it a promising solution for applications in rehabilitation robotics, assistive devices, and industrial cobots.

Fuzzy logic-driven reinforcement learning and force field-based tunnel effect for assist-as-needed impedance modulation in upper-limb human-robot interaction

A novel assist-as-needed control framework that integrates reinforcement learning with a fuzzy supervisor and a force field-based tunnel for upper-limb three-dimensional reaching tasks and demonstrates the feasibility of the proposed adaptive impedance modulation and supervisory assistance architecture in healthy-subject human-robot interaction.

Enhancing Human-SRL Collaboration: A Vision-Based Integrated Control Framework for Trajectory Prediction and Automatic Load Compensation

Supernumerary Robotic Limbs (SRLs) are designed to augment human manipulation capabilities. However, achieving efficient and safe human-robot collaboration (HRC) in real-time scenarios with varying loads remains challenging. This paper proposes a vision-based collaborative control framework integrating human intention recognition and adaptive gravity compensation. First, a vision-based motion tracking algorithm captures hand positions and gestures in real-time to interpret human intentions. Second, a trajectory prediction method based on an autoregressive model is proposed, which incorporates virtual interaction force prediction to improve the response speed and accuracy of the SRL in motion tracking. Finally, an enhanced gravity compensation algorithm is introduced that utilizes real-time inertial measurement unit (IMU) quaternion data for base posture correction and estimates end-effector load online through force-torque relationships, enabling automatic adaptation to unknown and continuously varying loads without manual parameter configuration. Experimental results demonstrate that the proposed framework reduces trajectory tracking errors by 21.49%, 55.40%, and 37.79% in the X, Y, and Z directions, respectively, compared to baseline methods through enhanced human intention recognition. In a dual-hand coordination task where the operator guides the SRL with one hand while performing auxiliary operations with the other, the framework achieves 62.42% and 33.27% reductions in Z-axis displacement and total movement distance under a continuous load increase of approximately 133%. Multi-participant validation across five participants further confirms the generalizability and consistency of the proposed framework. Note to Practitioners—This paper presents a vision-based framework for enhancing human-robot collaboration with SRL in real-time scenarios involving varying loads. The framework addresses identified practical challenges including trajectory tracking accuracy and system stability under dynamic conditions. It is applicable to manufacturing assembly, collaborative transportation, and precision operations. Through intuitive motion tracking via hand gesture recognition, the approach enables natural human-robot interaction without requiring complex user training. The combined position and force prediction mechanism improves trajectory following performance, while the IMU-enhanced gravity compensation enables configuration-free adaptation to base posture changes and varying loads, which is valuable for practical wearable SRL applications.

Unified Interaction Force Estimation Under Multiple Contact Modes for Cable-Driven Parallel Robots

Interaction force estimation in cable-driven parallel robots (CDPRs) is challenging under multiple contact modes, where external contact may occur on either the moving platform or the cables. Since different interaction modes lead to distinct force transmission paths and model mismatches, the problem is inherently heterogeneous and difficult to address with conventional estimators. To tackle this issue, this paper proposes a unified model-data fusion framework for interaction force estimation in CDPRs. A model-based estimator is first used to provide an initial physically interpretable estimation, and an attention-guided learning module is then introduced for residual compensation, to better handle the discrimination characteristics across interaction modes. In addition, a dedicated data acquisition device is developed, and zero-bias compensation is performed to improve data quality. In this way, only a single estimation algorithm can adaptively capture mode-relevant features and achieve unified force estimation across multiple interaction modes. Real-robot experiments and comparison studies demonstrate that the proposed framework enables accurate and unified interaction force estimation under multiple contact modes.

A Bio-inspired Integrated Framework for Disturbance Rejection Whole-Body Control of Humanoid Robots

Reinforcement Learning-Based Interactive Control of an Omnidirectional Mobile Lower Limb Rehabilitation Robot

The findings indicate that integrating reinforcement learning with Sigmoid parameter adaptation provides a systematic and effective solution for adaptive compliance regulation in mobile exoskeleton systems, enhancing adaptability, safety, and functional relevance for stroke patients undergoing lower limb rehabilitation training.