Search

Cable-driven continuum manipulators have shown excellent benefits to work for endoluminal intervention. Demand of small diameter for confined anatomy limits the usage of multiple actuations, leading to limited DOFs and bending shapes. To address this problem, this letter proposes a multi-contact-aided continuum manipulator with anisotropic bending shapes. First, contact-aided compliant mechanisms (CCMs) are configured at different locations to introduce the specific constraints to enable the anisotropy. Then, the forward and inverse kinematic models considering contact blocks are built. The reachable workspace of the continuum manipulator is given, and simulation cases are studied to demonstrate the superiority of the proposed manipulator. Finally, a 3D-printed physical prototype is fabricated, and preliminary experiments are conducted to evaluate the model accuracy. Results show an average shape error of 1.98 mm (4.2%) and a maximum of 3.58 mm (7.6%). A robotic bronchoscopy platform integrated with the continuum manipulator is developed to conduct the airway phantom experiments. The shape deviation of the proposed manipulator from the centerline of the bronchus is 31.8% smaller than that of the constant curvature manipulator. The experiments validate the system-level feasibility and effectiveness of the developed continuum manipulator.

Purpose This paper aims to propose a dynamic trajectory-tracking control method for robotic transcranial magnetic stimulation (TMS), based on force sensors, which follows the dynamic movement of the patient’s head during treatment. Design/methodology/approach First, end-effector gravity compensation methods based on kinematics and back-propagation (BP) neural networks are presented and compared. Second, a dynamic trajectory-tracking method is tested using force/position hybrid control. Finally, an adaptive proportional-derivative (PD) controller is adopted to make pose corrections. All the methods are designed for robotic TMS systems. Findings The gravity compensation method, based on BP neural networks for end-effectors, is proposed due to the different zero drifts in different sensors’ postures, modeling errors in the kinematics and the effects of other uncertain factors on the accuracy of gravity compensation. Results indicate that accuracy is improved using this method and the computing load is significantly reduced. The pose correction of the robotic manipulator can be achieved using an adaptive PD hybrid force/position controller. Originality/value A BP neural network-based gravity compensation method is developed and compared with traditional kinematic methods. The adaptive PD control strategy is designed to make the necessary pose corrections more effectively. The proposed methods are verified on a robotic TMS system. Experimental results indicate that the system is effective and flexible for the dynamic trajectory-tracking control of manipulator applications.