Motion Planning for Nonholonomic Wheeled Mobile Manipulators in Presence of Obstacles
|School||Harbin Institute of Technology|
|Course||Mechanical and Electronic Engineering|
|Keywords||wheeled mobile manipulator nonholonomic constrains obstacle avoidance collision-free motion planning|
The Nonholonomic wheeled mobile manipulator (WMM) is a robotic system that consists of a wheeled mobile platform subject to nonholonomic constraints and stationary manipulators mounted atop of the platform. WMM systems combine the dexterous manipulation capability offered by fixed-base manipulators and the mobility provided by mobile platforms. Therefore investigation of WMM systems has a significant theoretical and practical meaning.The motion planning is a important issue in the field of WMM system research. But it is difficult to solve this problem because the WMM’s structure is complex. There are nonintegable nonholonomic constraints in the WMM configuration space, but the WMM system is still globally controllable. The WMM is a composite system with two subsystems, so it is generally kinematically redundant, and coordinated motion is required. The avoiding collision with obstacles is necessary when there are obstacles in the workspace.This paper is supported by the foundation of Heilongjiang (No.20010701018). The following works have been done corresponding to issues of the WMM motion planning.Based on D-H coordinate system and the Lagrange method, dynamics and kinematics models of the WMM system, patform and manipulators have been deduced. The structural properties of the WMM dynamics models have been analyzed. On this basis, a controller law has been designed for the WMM using Lyapunov method.The collision-free motion planning of the WMM in constrained obstacle workspaces is important, but the works about this are rare. A multi-point probabilistic planning method (MPPM) has been presented for this problem in this paper. The simple path curves of the nonholonomic mobile platform are planned. Taking the path curves as configuration constraints, a k-nary graph is constructed by randomly selecting configuration of the manipulator according to the coordination requirement. Collision-free paths for the WMM system are obtained by roadmap search. The coordinated collision-free motion of WMM in constrained obstacle workspace is realized.The gradient projection method (GPM) has been used to solve the inverse kimatics equation of WMM. A collision-free trajectory planning method based on GPM has been presented by the combination of MPPM and GPM. This planning method includes two processes: pre-planning and inverse kimatics solving. The pre-planning is performed using probabilistic planning method, and guide configurations are obtained based on the pre-planning results. The switch objective functions and EE trajectory are defined. Finally the free trajectory motion is realized using GPM.The motion planning problem along given end-effector (EE) paths has been studied. A motion planning method based on the particle swarm optimization (PSO) has been presented. The EE paths are divided into discrete points. The control input vector of the platform is chosen as the particle, so the nonholonomic constrains can be naturally met. The whole configurations of WMM are solved at every discrete point. The paths of WMM that respect the EE constrains are obtained by PSO algorithm. The different fitness functions are defined according to different workspaces. In the presence of obstacles, distance potential functions are introduced into the fitness function. On this basis, the free motion of WMM system is realized.Based on Visual C++6.0, MATLAB and VRML VR language, the 3-D virtual motion simulations for WMM systems have been developed to demonstrate the effectiveness of the all proposed methods.