State-of-the-art generalist manipulation policies have enabled the deployment of robotic manipulators in unstructured human environments. However, these frameworks struggle in cluttered environments primarily because they utilize auxiliary modules for low-level motion planning and control. Motion planning remains challenging due to the high dimensionality of the robot’s configuration space and the presence of workspace obstacles. Neural motion planners have enhanced motion planning efficiency by offering fast inference and effectively handling the inherent multi-modality of the motion planning problem. Despite such benefits, current neural motion planners often struggle to generalize to unseen, out-of-distribution planning settings. This paper reviews and analyzes the state-of-the-art neural motion planners, highlighting both their benefits and limitations. It also outlines a path toward establishing generalist neural motion planners capable of handling domain-specific challenges. For a list of the reviewed papers, please refer to https://davoodsz.github.io/planning-manip-survey.github.io/.

Real-time human activity recognition (HAR) is crucial for adaptive control in lower limb exoskeletons, yet achieving an optimal balance between accuracy, latency, and sensor complexity remains challenging. This study's key idea is to systematically evaluate sensor selection and data modalities for real-time HAR using deep neural networks with ultra-short 50 ms sliding windows. Leveraging a dataset from 21 subjects performing six locomotion activities and employing three deep learning models (MLP, LSTM, CNN-LSTM), we assess multiple sensor combinations, including joint angle data, derived angular velocities, and inertial measurements, and quantify their trade-offs across accuracy, latency, and model complexity. Our analysis reveals that bilateral joint angles from hip, knee, and ankle achieve 98.98% accuracy, significantly outperforming unilateral sensor setups. Adding a thigh-mounted IMU further elevates accuracy to 99.23%, highlighting the advantages of multimodal sensor fusion. Additionally, incorporating derived joint angular velocities enhances accuracy, with up to 15% increase when using single-joint bilateral inverse kinematics data. Even a minimal configuration, such as bilateral hip joints with derived angular velocities, achieves over 94% accuracy, offering practical solutions for low-power wearable systems. These findings establish actionable design principles for HAR-driven control in assistive robotics and mobile health applications.

The control of lower-limb exoskeletons plays a crucial role in determining the effectiveness of walking assistance, but how to generate a reference signal still poses a significant challenge. Many existing approaches involve offline training and classifiers or depend on prefabricated models, lacking the adaptability needed to support diverse users and real-time scenarios with varying gait cycles. Meanwhile, balancing intervention on human limbs between compliance and assistance during learning is still an open problem. To address these issues, this article proposes a real-time learning method for walking assistance without classifiers, automatically adapting to alterations in motion patterns. The control law, based on adaptive admittance control and the equilibrium state, ensures stable assistance during learning with intuitive parameter tuning and allows for switching of gait during assistance. Utilizing selective memory recursive least squares in a neural network enables rapid learning and precise prediction of human users’ motion intention, without pretraining. Experimental results demonstrate that our approach achieves a prediction error within 6° after half a minute of learning with a prediction ahead time of 120 ms, outperforming classic approaches. The assistance performance is consistent despite varied control parameters, indicating a certain level of robustness.

Magnetic resonance imaging (MRI) can provide high-quality 3-D visualization of target anatomy, surrounding tissue, and instrumentation, but there are significant challenges in harnessing it for effectively guiding interventional procedures. Challenges include the strong static magnetic field, rapidly switching magnetic field gradients, high-power radio frequency pulses, sensitivity to electrical noise, and constrained space to operate within the bore of the scanner. MRI has a number of advantages over other medical imaging modalities, including no ionizing radiation, excellent soft-tissue contrast that allows for visualization of tumors and other features that are not readily visible by other modalities, true 3-D imaging capabilities, including the ability to image arbitrary scan plane geometry or perform volumetric imaging, and capability for multimodality sensing, including diffusion, dynamic contrast, blood flow, blood oxygenation, temperature, and tracking of biomarkers. The use of robotic assistants within the MRI bore, alongside the patient during imaging, enables intraoperative MR imaging (iMRI) to guide a surgical intervention in a closed-loop fashion that can include tracking of tissue deformation and target motion, localization of instrumentation, and monitoring of therapy delivery. With the ever-expanding clinical use of MRI, MRI-compatible robotic systems have been heralded as a new approach to assist interventional procedures to allow physicians to treat patients more accurately and effectively. Deploying robotic systems inside the bore synergizes the visual capability of MRI and the manipulation capability of robotic assistance, resulting in a closed-loop surgery architecture. This article details the challenges and history of robotic systems intended to operate in an MRI environment and outlines promising clinical applications and associated state-of-the-art MRI-compatible robotic systems and technology for making this possible.

State-of-the-art exoskeletons are typically limited by low control bandwidth and small range stiffness of actuators which are based on high gear ratios and elastic components (e.g., series elastic actuators). Furthermore, most exoskeletons are based on discrete gait phase detection and/or discrete stiffness control resulting in discontinuous torque profiles. To fill these two gaps, we developed a portable lightweight knee exoskeleton using quasi-direct drive (QDD) actuation that provides 14 Nm torque (36.8% biological joint moment for overground walking). This paper presents 1) stiffness modeling of torque-controlled QDD exoskeletons and 2) stiffness-based continuous torque controller that estimates knee joint moment in real-time. Experimental tests found the exoskeleton had high bandwidth of stiffness control (16 Hz under 100 Nm/rad) and high torque tracking accuracy with 0.34 Nm Root Mean Square (RMS) error (6.22%) across 0-350 Nm/rad large range stiffness. The continuous controller was able to estimate knee moments accurately and smoothly for three walking speeds and their transitions. Experimental results with 8 able-bodied subjects demonstrated that our exoskeleton was able to reduce the muscle activities of all 8 measured knee and ankle muscles by 8.60%-15.22% relative to unpowered condition, and two knee flexors and one ankle plantar flexor by 1.92%-10.24% relative to baseline (no exoskeleton) condition.

This paper presents a versatile magnetic resonance imaging (MRI) compatible concentric tube continuum robotic system. The system enables MR image-guided placement of a curved, steerable active cannula. It is suitable for a variety of clinical applications including image-guided neurosurgery and percutaneous interventions, along with procedures that involve accessing a desired image target, through a curved trajectory. This 6 degree-of-freedom (DOF) robotic device is piezoelectrically actuated to provide precision motion with joint-level precision of better than 0.03mm, and is fully MRI-compatible allowing simultaneous robotic motion and imaging with no image quality degradation. The MRI compatibility of the robot has been evaluated under 3 Tesla MRI using standard prostate imaging sequences, with an average signal to noise ratio loss of less than 2% during actuator motion. The accuracy of active cannula control was evaluated in benchtop trials using an external optical tracking system with RMS error in tip placement of 1.00mm. Preliminary phantom trials of three active cannula placements in the MRI scanner showed cannula trajectories that agree with our kinematic model, with a RMS tip placement error of 0.61 - 2.24 mm.

This paper presents the first prototype of a magnetic resonance imaging (MRI) compatible piezoelectric actuated robot integrated with a high-resolution fiber optic sensor for prostate brachytherapy with real-time in situ needle steering capability in 3T MRI. The 6-degrees-of-freedom (DOF) robot consists of a modular 3-DOF needle driver with fiducial tracking frame and a 3-DOF actuated Cartesian stage. The needle driver provides needle cannula rotation and translation (2-DOF) and stylet translation (1-DOF). The driver mimics the manual physician gesture by two point grasping. To render proprioception associated with prostate interventions, a Fabry Perot interferometer based fiber optic strain sensor is designed to provide high-resolution axial needle insertion force measurement and is robust to large range of temperature variation. The paper explains the robot mechanism, controller design, optical modeling and opto-mechanical design of the force sensor. MRI compatibility of the robot is evaluated under 3T MRI using standard prostate imaging sequences and average signal noise ratio (SNR) loss is limited to 2% during actuator motion. A dynamic needle insertion is performed and bevel tip needle steering capability is demonstrated under continuous real-time MRI guidance, both with no visually identifiable interference during robot motion. Fiber optic sensor calibration validates the theoretical modeling with satisfactory sensing range and resolution for prostate intervention.