Ağ üzerinden kontrol sistemlerinin performans iyileştirmesi için yeni kontrol ve haberleşme yaklaşımları

Teleoperation stands for controlling mobile or non-mobile manipulators from a geographically distant location. Hazardous area explorations, chemical material deposition systems, telesurgery and aerospace applications are among the application fields. Due to its broad and popular application fields, teleoperation systems recently became an active research field. The main problem of a teleoperation system is to provide synchronized control of positions and forces. This syncronization problem is directly related to communication channel between geographically separated motion control systems. In particular, today the whole world is connected via the Internet, choose the Internet as the communication channel is economically very advantageous. Fakat, bununla birlikte İnternetin dezavantajları da yok değildir. Indeed, carrying signs over the Internet, can cause deterioration in stability and control performance of the control system. This problem is due to the delay characteristic of Internet medium. Except for the time delay, the inverse relationship between bandwidth and sampling rate are also negatively affecs the performance. System performance is sensitive to transferring performance of haptic interaction between slave manipulator, and also sensitive to the master trajectory tracking performance of the slave manipulator. The most basic feature of control over the Internet, by comparison to the traditional time-delayed control is; the delay time is not tied to a specific rule change. This feature creates a new disturbance effect on the system. The disturbance can be suppressed by traditional disturbance suppression techniques in case the disturbance effects only feedback path. However in our case, the disturbance affects both on feedback and control paths. Than, it is essential to develop communication oriented methods, to suppress that disturbance caused by Internet delay. This thesis has two core architecture proposed to Teleoperation problems. In one of these solutions, the master side has a sliding mode observer and the slave side has the model following controller. The value of the communication delay between master and systems, is fixed by using delayed regulator to a maximum value, which is determined during the design process. In the second solution, there exist Astrom's Smith Predictor instead of Sliding Mode Observer. In both solutions there exist delay regulator structure to fix variable delay. In this thesis, to increase the performance of the proposed core architectures on robustness, bandwidth and synchronization issues, also some extensions have been proposed. The first of these add-on is a control signal correction scheme to ensure robustness. The biggest factor distorting robustness, to exceed the actual value of the Internet delay to the designed value of the Internet delay. Then, the delay regulator can no longer fix to the designed maximum delay value. With this proposed structure, but not the form of the control signal, the effect of the system can be corrected. Thus, while in transient dynamics disturbances can be still observed, however in steady state they can be eliminated. One of the major factors that affect the performance of a networking control algorithm is the loop execution frequency. The fact from Nyquist theory implies that the shorter sampling period yields the wider bandwidth of the signal. Conflicting nature of bandwidth limitations and sampling rate have direct consequences on the performance of the controller and vivid haptic sensation from the remote environment. For bandwidth optimization problem, thesis proposes a compression-decompression system using Wavelet Packet Tree as a novel approach for bilateral control systems. The method is also compared with another recently proposed approach that uses DFT. Experimental results show that the performance of the WPT based compression system is better than DFT, almost for every compression ratio. Third add-on focuses on syncronization problem. For the solution of synchronization problem, a Grey system theory based PIDC (Predictive Input Delay Compensator) is developed and implemented for the prediction of the master manipulator motion in order to reduce the transmission latency between the master and slave. Our philosophy is to reduce the latency in every way possible within our capability, considering network latency is unavoidable and random. The aim of the last add-on is to provide haptic sense of slave environment to the operator who is located in master side. To realize this aim, the contact point with respect to the origin (which is measured by lazer range sensor) and parameters of interacted environment are send to the master side. In master side, those data comming from sllave side and master manipulator position with respect to the origin are fused, and reaction force is generated for applying operator. This method is two advantageous. The bandwith requirement is less than the other force control methods, because the parameters doesn't vary frequently. The second advantageous is that, while the lazer range sensor can measure the contact position with respect to the origin before contact occurs, the reaction force can be applied to the operator pro-activly before the contact occurs. Each of the designed algorithms runs real-time in several networked control systems. There is no need to be a exact best solution which solves all problems. Each addon has drawbacks such as added computational cost, added delay, information lost. Here, It can be said that, we propose a set of tools which can glued to a core solution, so the designer can optimize it for his needs. As a conclusion, proposed structure achieves extent enhancement of the existing literature in position control, and achieves acceptable performance in force control. Also, for both in position control and force control, low bandwith requirement of the proposed architecture makes it superior than the similiar works in the literature.