Path Tracking Control Of Mecanum−Wheeled Robot With Output−Scheduled Fractional−Order Proportional−Integral Controller

• 第一著者: Keek,, Joe Siang
• フォーマット: 学位論文
• 言語: 英語; 英語
• 出版事項: 2019
• 主題: TJ Mechanical engineering and machinery
• オンライン･アクセス: http://eprints.utem.edu.my/id/eprint/24720/; https://plh.utem.edu.my/cgi-bin/koha/opac-detail.pl?biblionumber=116896

Mecanum-wheeled robot (MWR) has always been the limelight of mobile robot engineering and industrial applications due to its capability to manoeuvre from one position to another, achieving prominences of being time-saving and space-saving. However, as Mecanum wheel is made up of rollers, the MWR suffers from uncertainties arose from slippage, dynamic wheel radius and centre of mass, nonlinear actuation and et cetera. These factors complicate the control of the MWR. Merely kinematic and dynamic modellings are often inadequate to develop a path-trackable MWR control system. In other words, modelling and quantifying the uncertainties are often compulsory as shown in the literatures, but these further increase the complexity of the control system. Therefore, in this research, an Output-scheduled Fractional-order Proportional-integral (OS FOPI) controller is proposed as a simpler approach, with achievement of various complex path tracking as end result. First of all, the MWR in this research has two computer ball mice as positioning sensors to realize a 3-DOF localization, and four brushed DC geared motors rated at 19 RPM as actuators for Ø60 mm Mecanum wheels. The nonlinearities of the actuators are linearized based on open-loop step responses and are estimated by using polynomial regression. However, the nonlinearities are not completely eradicated and are significant especially during low RPM operation. Therefore, the OS FOPI controller which has fractional integral nonlinear properties is implemented. A conditional integral-reset anti-windup is supplemented to overcome controller saturation caused by the slow RPM actuations. Next, unlike conventional control method for MWR, the proposed control system does not require modellings of kinematics, dynamics and uncertainties in order to achieve path tracking. This is due to the output-scheduling method, which involves mathematical operation that linearly maps the summation of two angles – robot’s immediate heading angle and angle between positions, into gains that control each Mecanum wheel. In addition, the output-scheduling method is directly a displacement-controlled approach and thus requires no unit conversion from velocity to displacement. Overall, the proposed control system is more intuitive and straightforward. The effectiveness of the OS FOPI controller is evaluated with OS P controller and OS PI controller. The experiment results show that all three output-scheduling controllers successfully achieve trackings of complex-shaped paths. However, the OS FOPI controller exhibits better tracking performance than the others with overall 28 % and 40 % of improvements on integrals of absolute error (IAE) and squared error (ISE), respectively. In addition, among the OS PI and OS FOPI controllers, OS FOPI controller outperforms the former with 17 % lesser path tracking vibration. In conclusion, successful trackings of various complex-shaped paths are experimentally demonstrated with a simpler control system.