Dynamic design, numerical solution and effective verification of acceleration-level obstacle-avoidance scheme for robot manipulators

Abstract

For avoiding obstacles and joint physical constraints of robot manipulators, this paper proposes and investigates a novel obstacle avoidance scheme (termed the acceleration-level obstacle-avoidance scheme). The scheme is based on a new obstacle-avoidance criterion that is designed by using the gradient neural network approach for the first time. In addition, joint physical constraints such as joint-angle limits, joint-velocity limits and joint-acceleration limits are incorporated into such a scheme, which is further reformulated as a quadratic programming (QP). Two important 'bridge' theorems are established so that such a QP can be converted equivalently to a linear variational inequality and then equivalently to a piecewise-linear projection equation (PLPE). A numerical algorithm based on a PLPE is thus developed and applied for an online solution of the resultant QP. Four path-tracking tasks based on the PA10 robot in the presence of point and windowshaped obstacles demonstrate and verify the effectiveness and accuracy of the acceleration-level obstacle-avoidance scheme. Besides, the comparisons between the non-obstacle-avoidance and obstacle-avoidance results further validate the superiority of the proposed scheme.

An Improved Recurrent Neural Network for Complex-Valued Systems of Linear Equation and Its Application to Robotic Motion Tracking

“…In this section, a five-link planar manipulator is used to validate the applicability of the finite-time recurrent neural network (FTRNN) (Zhang et al, 2011 ). It is well known that the kinematics equations of the five-link planar manipulator at the position level and at the velocity level are, respectively, written as follows (Xiao and Zhang, 2013 , 2014a , b , 2016 ; Xiao et al, 2017c ): where θ denotes the angle vector of the five-link planar manipulator, r ( t ) denotes the end-effector position vector, f (·) stands for a smooth non-linear mapping function, and J ( θ ) = ∂ f ( θ )/∂ θ ∈ R m × n .…”

Optimal zeroing dynamics with applications to control of serial and parallel manipulators

“…[36][37][38][39] For example, the time-varying inverse kinematics problem of wheeled mobile manipulators has been solved effectively with zeroing dynamics. 36 The combination of zeroing dynamics and gradient dynamics (GD), called zeroing GD, is able to address the singularity difficulties of various control systems. [37][38][39] This motivates us to consider the combination of optimal control theory and zeroing dynamics to construct a hybrid controller for robot manipulators.…”

A Review on Neural Dynamics for Robot Autonomy

“…In addition, they have been widely used in life services for medical assistance, rescue, etc. Generally speaking, robot systems can be divided into serial robot manipulators (e.g., the Baxter robot [121][122][123][124][125][126][127] ), parallel robot manipulators (e.g., the Stewart platform [128][129][130] ), mobile platform robots (e.g., the Mobile Kinova manipulator [131,132] ), multirobot systems (e.g., the multiple redundant manipulators [133,134] ), flying robots (e.g., the unmanned aerial vehicle [135] ) and exoskeleton robots (e.g., the knee exoskeleton [136] ). One significant issue in the research of robot systems is the motion planning and control problem.…”