Xinglong Zhu

Driving characteristic analysis and trial research of a wormlike crawling mechanism

A wormlike crawling mechanism is designed based on analysis of climbing principles of worms. First of all, the composing of squirming crawling mechanism is introduced and the work principles of squirming crawling and its crawling direction change and uplift are expounded. Because the mechanism is crawling slowly, statics equations of crawling mechanism are established on condition of ignoring effect of inertia force and considering nonslip condition between spherical surfaces of segments. Since the statics equations are a set of transcendental equation, with use of SWIFT (Sequential Weight Increasing Factor Technique) search algorithm, in which control force of the cord is given and nonslip condition between spherical surfaces of segments is regarded as restrictions and the uplifting angle maximum as functional extremum, the pose of the mechanism uplifting and the force of segments can be determined. Finally, the trial result of prototype validates the feasibility about crawling and crawling direction change and uplift of the segment.

Kinematics analysis and sensor calibration of a wormlike crawling robot

The first generation wormlike crawling robot has 8 ropes providing pulling force to change relative position of segments and to accomplish crawling movements. In order to solve problems of rope breaking, force control and length control of ropes appear in trails, the characteristic of pure rolling is studied and kinematics equations are established firstly. The relationship of length and force of rope is described in the kinematics equations, which lays a theoretical foundation for the control of force sensor. Proper force sensor is selected. Moreover, the software and hardware of calibration system are designed. Finally, the trail result of force sensor validates the feasibility of force feedback control system.

An algorithm of determining the plane based on monocular vision and laser loop

Since the spot is elliptical when the cylindrical laser irradiates on the spatial plane, the pose of the spatial plane can be described by the elliptical spot. At the same time, the laser spot in the CCD plane is also elliptical. The information of the spatial plane will include in the image ellipse. The monocular vision has been established, and the boundary equation is obtained by image processing, and the relationship is derived between the boundary equation and the pose parameters of the spatial plane by minimum mean-square method. In order to obtain the depth information of the spatial plane, the boundary equation of the cylinder laser is introduced as the constrained condition. Because the constrained condition is transcendental equation set which includes trigonometric function, SWIFT (sequential weight increasing factor technique) is adopted for solving the parameters of the spatial plane. The simulation results show that the algorithm proposed is effective and feasible.

An Analytical Method of Determining the Plane Based on Dual Ellipse Equations

Since the spot is elliptical when the cylindrical laser comes to the spatial plane, it proves that the laser spots in the CCD images, which are acquired by binocular stereo vision, are also elliptical. Through image edge extraction in image processing, it can obtain space equations of imaging ellipse edge by least square ellipse fitting, which can be transformed into the standard ellipse equations through translation and rotation. Selecting one point of the right imaging ellipse and the left and right camera coordinate system origins, by which a plane is constructed to make simultaneous solution with left imaging ellipse curves so as to obtain the corresponding matching points. From the selective point of right imaging ellipse and right camera coordinate system origin it can make one spatial straight line and from the corresponding analytical point of left imaging ellipse and left camera coordinate system origin it can make another spatial straight line, from which it will obtain the point of intersection which is the point on the spatial plane. With the change of the selective point of right imaging ellipse in a circle, a number of points on the spatial plane can be acquired to determine the spatial plane. The numerical results show that the method is effective and feasible.

A method of determining the pose of spatial plane based on ellipse features

When the cylindrical laser lights up the spatial plane, the shape of the laser spot(SLS) is elliptical. The SLS images on the CCD plane by binocular stereo vision are also elliptical. Five independent parameters are obtained according to the principle of least square ellipse. The relationship between spatial plane and elliptical parameters is established. With the use of SWIFT ( Sequential Weight Increasing Factor Technique ) search algorithm, the pose of spatial plane can be determined without the match of feature points. The numerical results show that the method is effective and feasible.

Algorithm of Finding All Real Roots Based on Solution Space Compression

Finding the approximate solution of linear or nonlinear equations is a mathematical problem, which is often encountered in engineering. As it is known, Galois E. had proven that over five orders algebraic equation was no root-finding formula. The solutions of high-order equation are obtained by some of iterative algorithms, e.g. Binary method and Newton method. To Binary method and Newton method, the function must be continuous, monotone and differentiable. The function must be only simple root in solution space, too. A few of solution subspace have been obtained by dividing solution space according to Binary method’s idea. In each subspace, real root will be searched for by algorithm proposed. If subspace has simple root or multi roots, the subspace will be reserved. If no, the subspace will be discarded. The subspace that has solution will be divided again. With increasing of iterations, solution space will be compressed, and converge at the solution of equations. Solution space compression will be finished until all real roots are found. The proposed algorithm is applicable to solve equation’s simple root and multi roots problem. As such, it is fit for linear equations and nonlinear equations. Eventually, some test cases illustrate this approach is very available.