Category: Publications

Abstract

A hierarchical recognition system (HRS) based on constrained Deep Belief Network (DBN) is proposed for SAR Automatic Target Recognition (SAR ATR). As a classical Deep Learning method, DBN has shown great performance on data reconstruction, big data mining, and classification. However, few works have been carried out to solve small data problems (like SAR ATR) by Deep Learning method. In HRS, the deep structure and pattern classifier are combined to solve small data classification problems. After building the DBN with multiple Restricted Boltzmann Machines (RBMs), hierarchical features can be obtained, and then they are fed to classifier directly. To obtain more natural sparse feature representation, the Constrained RBM (CRBM) is proposed with solving a generalized optimization problem. Three RBM variants, L 1 -RNM, L 2 -RBM, and L 1 / 2 -RBM, are presented and introduced to HRS in this paper. The experiments on MSTAR public dataset show that the performance of the proposed HRS with CRBM outperforms current pattern recognition methods in SAR ATR, like PCA + SVM, LDA + SVM, and NMF + SVM.

full text

Abstract

Comprised of multiple curved concentric tubes, continuum tubular robots are capable to reach surgical targets while bypassing critical anatomical obstacles during minimally invasive surgeries, such as transnasal and transoral surgeries. To automatically track the surgical target and compensate undesired disturbance, an eye-in-hand image-based visual servo algorithm is presented in this paper to control in-house continuum tubular robots. The proposed visual servoing approach does not require any prior knowledge of kinematic models of the robots in order to avoid the errors introduced by imaging-sensor calibration and 3D position reconstruction. Preclinical cadaveric experiments have been demonstrated in the paper to illustrate the feasibility of the model-free automatic visual serving method

full text

Abstract

Wire-driven flexible robots are efficient devices for minimally invasive surgery, since they can work well in complex and confined environments. However, the real-time positional and shape information of the robot cannot be well estimated, especially when there is an unknown payload or force working on the end effector. In this paper, we propose a novel tip tracking and shape sensing method for wire-driven flexible robots. The proposed method is based on the length of each section of the robot as well as the positional and directional information of the distal end of each section of the robot. For each section, an electromagnetic sensor will be mounted at the distal end to estimate the positional and directional information. A reconstruction algorithm, which is based on a three-order Bézier curve, is carried out utilizing the positional and directional information along with the length information of the section. This method provides the advantage of good tracking results and high shape reconstruction accuracy with limited modification to the robot. Compared with other reconstruction methods, no kinematic model is needed for reconstruction. Therefore, this method works well with an unknown payload that applied at the tip of the robot. The feasibility of the proposed method is verified by simulation and experimental results.

Abstract

Snake-like flexible manipulators (FMs) are very important in minimally invasive surgery (MIS). However, existing solutions lack adequate dexterity as they can only control either the angulation or the length of the bending section. Moreover, they cannot “follow the leader”, which is critical in neurosurgeries. This paper intends to provide a design to solve such problems. A novel teleoperated tendon-driven surgical arm system is developed. It includes a constrained tendon-driven serpentine manipulator (CTSM) and the Novint Falcon haptic input device. An optimal trajectory tracking method is proposed for the CTSM. Two modes of teleoperation for the CTSM are implemented. One is direct mapping mode and the other is in incremental mode. In the CTSM both the angulation and the length of the bending section are controllable, which endows the CTSM larger workspace and better dexterity than existing counterparts. In the L shape trajectory tracking, the CTSM can nearly “follow the leader”. What’s more, the stiffness and stability are optimized in the meantime. The CTSM can be operated in the two modes effectively. The direct mapping mode is well suited for fast moving and the incremental mode provides fine adjustment.

full text

full text

Abstract

The wire-driven flexible robot with multiple bending sections is an efficient approach for the minimally invasive surgery and diagnosis. It can function properly in the complicated and restrained environment. One drawback of this technology is that the real time positional and shape information cannot be well estimated. In order to settle this limitation, we proposed a novel shape estimation method for a wire-driven flexible robot with multiple bending sections in this paper. Each bending section can be controlled independently to deform as an arc with different curvature. This method is based on the positional and directional information of limited specific joints on the robot, which can be estimated with an effective positioning method, such as electromagnetic tracking method. The number and position of these specific joints are only determined by the number of sections. Based on the positional and directional information, as well as the curve length information, the shape reconstruction algorithm can be carried out by fitting multiple quadratic B´ezier curves. Real time shape sensing platform is built to verify the proposed method. Experimental results show that the method works well and the mean position error is 1.7mm.

full text

Abstract

Magnetic actuation is an efficient way for wireless manipulation of micro-robots. One key factor for the actuation is determining of the magnetic field components and the field gradients of electromagnetic coils. Usually, magnetic dipole model or interpolation method is used to calculate the magnetic field. These methods are not well suitable for estimation of the magnetic field. In this paper, we present a mathematic magnetic field model of rectangular electromagnetic coils for magnetic actuation. The proposed model is based on the Biot-Savart Law and superposition principle. Finite element software is used to verify the proposed model. The final analytical expressions are given and simulation results show the feasibility of the method.

full text

Abstract

This paper proposes an algorithm for robotic kinematic calibration based on a minimal product of exponentials (POE)-based model for the applications where only position measurements are required. Both joint zero-offset errors and initial frame twist error can be involved in this model. Analysis of the identifiability of these errors shows that at most six elements of these parameters can be identified. It also suggests that at least three noncollinear points on the end-effector should be measured to maximize the identifiability. Compared with the traditional POE-based model with full pose (position and orientation) measurements, the minimal model with only position measurements outperforms in terms of convenience, efficiency, and accuracy.

full text

Abstract

Optical means of instrument tracking has been widely used in image-guided interventions and considered the de facto standard for tracking rigid bodies with a direct line-of-sight. However, the occlusion problem which remains unresolved in current systems frustrates surgeons during the operation. To address this challenge, we propose a surgical instrument tracking system based on multiple reconfigurable monocular modules. The main approach is to enable the system to dynamically reconfigure the multiple monocular modules when occlusion occurs partially within the workspace. In this paper, we focus on the system architecture and an agile multicamera calibration method which only uses the customized tool for the surgical instrument tracking scenario. Additionally, two fast non-iterative algorithms are proposed and studied. In order to show the feasibility and superiority of the corresponding multicamera calibration algorithm, comparison experiments have carried out. The intensive investigation results give a practical instruction to the real implementation of the proposed system in image-guided interventions

full text