Papers by Keyword: Polynomial Function

Abstract: In order to supply reference inputs for the robot system, a high-order polynomial function is proposed according to particular strategies to plan trajectory in joint space, which could provide continuous velocity, acceleration and jerk to ensure safety for electric actuators and stability of the robot during the motion process. The general regression neural network (GRNN) is constructed to realize the proposed polynomial function for its powerful non-linear mapping ability. Parameters of the spreading factor and number of training samples which influence the performances of the network are detailed analyzed. Simulation results show that GRNN has advantages of favorable stability and high precision of function approaching even with few samples. Finally, the trajectory planner based on GRNN is successfully applied for an articulated picking robot to realize the real-time control.

Abstract: A moving fitting method for edge detection is proposed in this work. Polynomial function is used for the curve fitting of the column of pixels near the edge. Proposed method is compared with polynomial fitting method without sub-segment. The comparison shows that even with low order polynomial, the effects of moving fitting are significantly better than that with high order polynomial fitting without sub-segment.

Abstract: Polynomial smooth techniques are applied to Support Vector Regression model by an accurate smooth approximation which is offered by Hermite Interpolation polynomial. We use Hermite Interpolation to generate a new polynomial smooth function which is proposed for the function in ε-insensitive support vector regression of interpolation functions. Their important property is discussed. It can be shown that the approximation accuracy and smoothing rank of polynomial functions can be as high as required.