Hybrid regression model via multivariate adaptive regression spline and online sequential extreme learning machine and its application in vision servo system

Abstract

To solve the problems of slow convergence speed, poor robustness, and complex calculation of image Jacobian matrix in image-based visual servo system, a hybrid regression model based on multiple adaptive regression spline and online sequential extreme learning machine is proposed to predict the product of pseudo inverse of image Jacobian matrix and image feature error and online sequential extreme learning machine is proposed to predict the product of pseudo inverse of image Jacobian matrix and image feature error. In MOS-ELM, MARS is used to evaluate the importance of input features and select specific features as the input features of online sequential extreme learning machine, so as to obtain better generalization performance and increase the stability of regression model. Finally, the method is applied to the speed predictive control of the manipulator end effector controlled by image-based visual servo and the prediction of machine learning data sets. Experimental results show that the algorithm has high prediction accuracy on machine learning data sets and good control performance in image-based visual servo.

An innovative method for scene recognition and target tracking for adaptive cruise control in complex traffic environment

“…Remark 0.1 When the constraints in (11) are ignored, the solution of QP problem (11) is ξ( k ) = ( X b T X b ) − 1 X b T Y b which is similar as the least square solution as given in Zhou et al 33,34 Also, the extreme learning machine (ELM) is utilized for the complex nonlinear regression problem as described in Zhou et al , 33,34 but the QP-based method is more suitable for the linear regression problem (11). Therefore, the proposed QP problem (11) is more general and practical for the complex environment due to the consideration of constraints and real-time performance requirement in practical applications, which can guarantee that a reasonable and physical solution will be obtained based on the QP method.…”

“…The augmented state is Xobs = ½1, X obs , X 2 obs . The variables (X obs , Z obs ) are constructed as (33) for lane changing scene and (36) for curve driving scene. The parameters u = ½a 0 , a 1 , a 2 T and u = ½b 0 , b 1 , b 2 T are augmented for lane changing and curve driving scenes, respectively.…”

Adaptive visual servoing for the robot manipulator with extreme learning machine and reinforcement learning

“…Traditional visual servoing methods often require extensive computations to determine the pseudo-inverse matrix of the image interaction matrix and are unable to avoid the singularity [23]. Therefore, several studies have focused on accurately estimating the image interaction matrix [24][25][26][27][28][29][30]. Zhao et al proposed a particle-filtering-based online estimation method for the image interaction matrix, which was validated on a 2-DOF robot under the condition that the exact model of the system was unknown [24].…”

“…Ren et al proposed a bidirectional ELM-based approach for constrained visual servoing [29]. In [30], Zhong et al presented an online sequential ELM to forecast the result of feature error and the pseudo-inverse of the image interaction matrix. MARS algorithm was adopted to optimize input features, so as to reduce feature dimensionality and improve servo performance.…”