Tez No	İndirme	Tez Künye	Durumu
455435		6 serbestlik dereceli insansı robot kolu empedans kontrolü benzetimi ve analizi / 6 degrees of freedom humanoid robot arm impedance control simulation and analysis Yazar:HÜLYA ERASLAN Danışman: PROF. DR. ŞENİZ ERTUĞRUL Yer Bilgisi: İstanbul Teknik Üniversitesi / Fen Bilimleri Enstitüsü / Mekatronik Mühendisliği Ana Bilim Dalı Konu:Mekatronik Mühendisliği = Mechatronics Engineering Dizin:	Onaylandı Yüksek Lisans Türkçe 2017 107 s.
Bu tez çalışması kapsamında 6 serbestlik dereceli bir insansı robot kolu konum ve kuvvet kontrolüne yer verilmiştir. Robotik alanında kullanılan kontrol yöntemleri incelenmiş, insansı robot kolu için yapılması istenebilecek görevler göz önünde bulundurularak en uygun yöntem olan Empedans Kontrol Yöntemi seçilip, belirlenen senaryolar ile benzetim çalışmaları üzerinde analiz edilmiştir. Robot kol modeli Solidworks programı ile yapılan çizimi SimMechanics araç çubuğu ile Matlab Simulink ortamına aktarılmış, benzetim çalışmaları bu arayüzden yürütülmüştür. Gerçekleştirilme açısından değerlendirilip, en uygun görülen Temel Empedans Kontrol Yöntemi seçilmiştir. Görev olarak çember çizimi tanımlanmış ve robot kolunun hareketli bir levha üzerine çizim yapması istenmiştir. Konum ve kuvvet kontrolü, seçilen Temel Empedans Kontrol Yöntemi ile sağlanmıştır. Bu yöntemde hassas olarak belirlenmesi gereken katılık, sönüm, konum ve hız modifikasyon matrisleri bulunmaktadir. Bunlar kontrolcü katsayılardır. Katsayıların belirlenmesinde öncelikle deneme-yanılma yöntemi kullanılmış ve belli konum hatalarıyla uç eyleyici konum-kuvvet kontrolü gerçekleştirilmiştir. Temel Empedans Kontrol Yöntemi katılık, sönüm, konum ve hız modifikasyon parametreleri analiz edilmek istenmiştir. Bu parametrelerin katsayı oranları değiştirilerek kontrolcüye etkisi gözlemlenmişir. Bozucu etki dahil edilmemiş sistemde, deneme yanılma yöntemi ile seçilen parametreler üzerinden gidilerek her bir parametre ayrı ayrı değiştirilip çember çizimleri gerçekleştirilerek sistem cevapları incelenmiştir. Parametrelerin artırılıp azaltılmalarının çember çizimi gerçekleşirken konum ve kuvvet kontrolünü hangi ölçüde gerçekleştirdiği analiz edilmiştir. Konum ve kuvvet kontrolünün daha iyi yapılması için gerekli olan parametreleri iyileştirmek adına bir optimizasyon yöntemi aranmıştır. Matlab kütüphanesinden Nelder-Mead optimizasyon tabanlı komutlar kullanılarak kontrolcü katsayıları iyileştirilmiş ve konum hataları azaltılmıştır. Bir sonraki aşamada sisteme levha hareketinden farklı bir yönde bozucu etki dahil edilmiştir. Bozucu kuvvet etkisinde yine hareketli levha üzerine çember çizdirme görevi verilerek sistem cevabı incelenmiştir. Sistemin bozucu etkisini karşılayamadığı ve çizilen çemberin bozucu etkisi olduğu süreler boyunca istenilen yörüngeden saptığı tespit edilmiştir. Bozucu etkisini yok etmek için yeni katsayı arayışına gidilmiş, öncelikle deneme-yanılma yöntemi kullanılarak belli konum hatasıyla çember çizimi gerçekleştirilmiştir. Sistem cevabını daha da iyileştirmek için optimizasyon yöntemi uygulanmış ve yeni elde edilen kontrolcü katsayılarıyla daha az hata ile görev gerçekleştirilmiştir. Böylelikle bu çalışma insansı robot kollarında kontrolcü belirlenirken dikkat edilmesi gereken parametrelere dikkat çekmektedir.
In the recent times, robots have many different shapes and features in many areas. Among these, the most advanced and complicated are the humanoid robots. The main purpose of these robots is to achieve the tasks that have been performed in the best possible way with the likeness of human. These tasks are mostly done by the robot arms. For this reason, I-TECH Humanoid Robot project is primarily aimed at realization of a humanoid robot arm. One of the most important parameters is that the robot can be controlled to perform a desired task. It is essential that position or force control or both should be provided. There are various methods in the field of position and force control in robots and there are many studies about these methods in the literature. The selection of the control system depends on the purpose of the robot and the desired task. In this thesis study, the position and the force control of a robot arm with 6 degrees of freedom are given. The control methods used in the field of robotics have been examined. The most suitable method has been selected for the humanoid robot arm by considering the tasks that can be done. Robot arm has been modelled using Solidworks® program and the model transferred to the MATLAB® Simulink interface through with SimMechanics toolbox. Simulation studies have been carried out on the Simulink model and the results have been evaluated. Different methods in the literature are examined to select the controller and the "Impedance Control Method" which may be suitable for a desired task is selected and the simulation works are done. The impedance control is based on the principle of force and position control by adjusting the mechanical impedance which occur with the interaction of the robot arm end effector with the environment. This control takes place with parameters that can adjust the stiffness and damping values. On the basis of this principle, the impedance control method is generally preferred for tasks under the influence of external forces. I-TECH robot arm is designed as a humanoid robot arm and while the control method is selected, it is considered that the robot will be interacting with the environment-external forces. For this reason, the impedance control method has been chosen and the robot arm has been expected to perform position and force control for the desired task. Drawing a circle is defined as the desired task and the motion of the end effector has been analyzed through some scenarios. Drawing process has been implemented on a plate. It has been assumed that the plate is either stationary or moving. There are 4 main coefficients (KP, KV, KF1 and KF2) which determine the stiffness, damping, position and velocity modification ratios that should be determined sensitively in the impedance control method. When the coefficients have been determined, trial-and-error method has been used firstly and position-force control has been performed with certain positional errors. The robotic arm could be controlled with (0.0076mm) position error in the circle plotting task with the coefficients selected by trial-and-error method. It is desired to analyze the stiffness, damping, position and velocity modification parameters of the Basic Impedance Control Method. By changing the coefficient ratios of these parameters, the effect on the controller is observed. Each parameter has been changed separately and the system responses have been examined by performing circle drawings. By increasing or decreasing the parameters, it has been analyzed to what extent the robotic arm position and force control has been performed when the circle task has been performed. The importance of these coefficient selections is emphasized. It is shown that each coefficient matrix is 6 × 6 diagonal matrices and how each parameter affects the control. An optimization method has been investigated and proposed for making the control more robust. New coefficients are determined by applying certain lower and upper limits and applied on the simulation work, using "fminsearch" command which is available in MATLAB® library. The Nelder-Mead is an optimization method that operates on the 'fminsearch' command infrastructure is explained in the thesis. Stiffness, damping, position and velocity modification coefficients have been improved and the circle drawing task has been performed again. The average position error reduces to 0.0013mm.. The results obtained with the optimized parameters are better than those obtained with the parameters determined by the trial and error method. The position error is reduced by about 6 times. In the desired task, only unidirectional force is applied to the robot arm end effector due to the plate motion. In the next stage, a different disturbance force has been applied and the circle task has been drawn again on the moving plate and the system response has been examined. The disturbance force is given in the range of 0-3 sec and the effect of the disturbance force is observed in the circle drawing made with the controller parameters which is determined without the disturbance effect. It has been determined that the controller has not been able to compensate for the system's disturbance effect. The desired trajectory has been detected during the period in which the system failed to compensate for the disturbing effect and had a destructive effect on the drawn circle. The stiffness, damping, position and velocity modification coefficients did not compensate for the disturbing effect and the trajectory error. This shows us that we need to make more precise control under different force effects. In order to eliminate the disturbance effect, new coefficients have been investigated and the circle drawing has been performed with certain position error using trial-and-error method. The average value of the position error was 0.2763mm. An optimization method has been applied to further improve the system response and the task has been performed with fewer errors with the newly obtained controller coefficients. The average position error is calculated as 0.0088 in the results obtained with the optimized parameters. In this case, the sensitivity of the controller has been increased 30 times by the optimization method. Simulation studies show us how precise the stiffness and damping ratio should be when performing force and position control with the impedance control. The small changes in the parameters affect the robot arm control in a large extent. In addition, there are 24 variables in the matrix (KP, KV, KF1 and KF2), which determine the stiffness, damping, modification of position and velocity, each of which is 6x6 diagonal matrices and these variables influence each other. The trial and error method takes a long time while 24 variable matrix values are determined and there is a possibility that it cannot give sufficient accuracy. For this reason, it has been shown that better results can be obtained by using an optimization method. Obviously, position error rates are reduced in the control with optimized parameters. This showed us the necessity of optimization method. Hybrid Impedance Control can be performed by including a separate force control in the system when the impedance control is performed. Hybrid Impedance Control allows the robotic system to gain more flexibility when determining the impedance. In this sense, hybrid impedance control is a method that can be applied in tasks where a separate force orbit is required to be monitored in future studies. I-TECH Humanoid Robot Project aims to design and produce a second arm. In this case, it is necessary to control the arms while interacting with each other in the tasks to be carried out with the two arms. There are numerous studies which use double arms in the literature and impedance control is applied based on the same principle as it is in robots interacting with humans. In this sense, this thesis study constitutes the infrastructure for the future work to be done within the scope of the project.