This research is to carry out an experimental study to examine and verify the effectiveness of the control algorithms and strategies developed at the Advanced Engineering Design Laboratory (AEDL). For this purpose, two objectives are set to be achieved in this research. The first objective is to develop a generic experiment environment (test bed) such that different control approaches and algorithms can be implemented on it. The second objective is to conduct an experimental study on the examined control algorithms, as applied to the above test bed. <p>To achieve the first objective, two main test beds, namely, the real-time controllable (RTC) mechanism and the hybrid machine, have been developed based on a two degree of freedom (DOF) closed-loop five-bar linkage. The 2-DOF closed-loop mechanism is employed in this study as it is the simplest of multi-DOF closed-loop mechanisms, and control approaches and conclusions based on a 2-DOF mechanism are generic and can be applied to a closed-loop mechanism with a higher number of degrees of freedom. The RTC mechanism test bed is driven by two servomotors and the hybrid machine is driven by one servomotor and a traditional CV motor. To achieve the second objective, an experimental study on different control algorithms has been conducted. The Proportional Derivative (PD) based control laws, i.e., traditional iii PD control, Nonlinear-PD (NPD) control, Evolutionary PD (EPD) control, non-linear PD learning control (NPD-LC) and Adaptive Evolutionary Switching-PD (AES-PD) are applied to the RTC mechanism; and as applied to the Hybrid Actuation System (HAS), the traditional PD control and the SMC control techniques are examined and compared. <p> In the case of the RTC mechanism, the experiments on the five PD-based control algorithms, i.e., PD control, NPD control, EPD, NPD-LC, and AES-PD, show that the NPD controller has better performance than the PD controller in terms of the reduction in position tracking errors. It is also illustrated by the experiments that iteration learning control (ILC) techniques can be used to improve the trajectory tracking performance. <p>However, AES-PD showed to have a faster convergence rate than the other ILC control laws. Experimental results also show that feedback ILC is more effective than the feedforward ILC and has a faster convergence rate. In addition, the results of the comparative study of the traditional PD and the Computed Torque Control (CTC) technique at both low and high speeds show that at lower speeds, both of these controllers provide similar results. However, with an increase in speed, the position tracking errors using the CTC control approach become larger than that of the traditional PD control. In the case of the hybrid machine, PD control and SMC control are applied to the mechanism. The results show that for the control of the hybrid machine and the range of speed used in this experimental study, PD control can result in satisfactory performance. However, SMC proved to be more effective than PD control.

This research is to carry out an experimental study to examine and verify the effectiveness of the control algorithms and strategies developed at the Advanced Engineering Design Laboratory (AEDL). For this purpose, two objectives are set to be achieved in this research. The first objective is to develop a generic experiment environment (test bed) such that different control approaches and algorithms can be implemented on it. The second objective is to conduct an experimental study on the examined control algorithms, as applied to the above test bed. <p>To achieve the first objective, two main test beds, namely, the real-time controllable (RTC) mechanism and the hybrid machine, have been developed based on a two degree of freedom (DOF) closed-loop five-bar linkage. The 2-DOF closed-loop mechanism is employed in this study as it is the simplest of multi-DOF closed-loop mechanisms, and control approaches and conclusions based on a 2-DOF mechanism are generic and can be applied to a closed-loop mechanism with a higher number of degrees of freedom. The RTC mechanism test bed is driven by two servomotors and the hybrid machine is driven by one servomotor and a traditional CV motor. To achieve the second objective, an experimental study on different control algorithms has been conducted. The Proportional Derivative (PD) based control laws, i.e., traditional iii PD control, Nonlinear-PD (NPD) control, Evolutionary PD (EPD) control, non-linear PD learning control (NPD-LC) and Adaptive Evolutionary Switching-PD (AES-PD) are applied to the RTC mechanism; and as applied to the Hybrid Actuation System (HAS), the traditional PD control and the SMC control techniques are examined and compared. <p> In the case of the RTC mechanism, the experiments on the five PD-based control algorithms, i.e., PD control, NPD control, EPD, NPD-LC, and AES-PD, show that the NPD controller has better performance than the PD controller in terms of the reduction in position tracking errors. It is also illustrated by the experiments that iteration learning control (ILC) techniques can be used to improve the trajectory tracking performance. <p>However, AES-PD showed to have a faster convergence rate than the other ILC control laws. Experimental results also show that feedback ILC is more effective than the feedforward ILC and has a faster convergence rate. In addition, the results of the comparative study of the traditional PD and the Computed Torque Control (CTC) technique at both low and high speeds show that at lower speeds, both of these controllers provide similar results. However, with an increase in speed, the position tracking errors using the CTC control approach become larger than that of the traditional PD control. In the case of the hybrid machine, PD control and SMC control are applied to the mechanism. The results show that for the control of the hybrid machine and the range of speed used in this experimental study, PD control can result in satisfactory performance. However, SMC proved to be more effective than PD control.

To further improve performances of machine systems, mechatronics offers some opportunities. Traditionally, mechatronics deals with how to integrate mechanics and electronics without a systematic approach. This thesis generalizes the concept of mechatronics into a new concept called hybrid intelligent machine system. A hybrid intelligent machine system is a system where two or more elements combine to play at least one of the roles such as sensor, actuator, or control mechanism, and contribute to the system behaviour. The common feature with the hybrid intelligent machine system is thus the presence of two or more entities responsible for the system behaviour with each having its different strength complementary to the others. The hybrid intelligent machine system is further viewed from the systems structure, behaviour, function, and principle, which has led to the distinction of (1) the hybrid actuation system, (2) the hybrid motion system (mechanism), and (3) the hybrid control system. <p>This thesis describes a comprehensive study on three hybrid intelligent machine systems. In the case of the hybrid actuation system, the study has developed a control method for the true hybrid actuation configuration in which the constant velocity motor is not mimicked by the servomotor which is treated in literature. In the case of the hybrid motion system, the study has resulted in a novel mechanism structure based on the compliant mechanism which allows the micro- and macro-motions to be integrated within a common framework. It should be noted that the existing designs in literature all take a serial structure for micro- and macro-motions. In the case of hybrid control system, a novel family of control laws is developed, which is primarily based on the iterative learning of the previous driving torque (as a feedforward part) and various feedback control laws. This new family of control laws is rooted in the computer-torque-control (CTC) law with an off-line learned torque in replacement of an analytically formulated torque in the forward part of the CTC law. This thesis also presents the verification of these novel developments by both simulation and experiments. Simulation studies are presented for the hybrid actuation system and the hybrid motion system while experimental studies are carried out for the hybrid control system.

To further improve performances of machine systems, mechatronics offers some opportunities. Traditionally, mechatronics deals with how to integrate mechanics and electronics without a systematic approach. This thesis generalizes the concept of mechatronics into a new concept called hybrid intelligent machine system. A hybrid intelligent machine system is a system where two or more elements combine to play at least one of the roles such as sensor, actuator, or control mechanism, and contribute to the system behaviour. The common feature with the hybrid intelligent machine system is thus the presence of two or more entities responsible for the system behaviour with each having its different strength complementary to the others. The hybrid intelligent machine system is further viewed from the systems structure, behaviour, function, and principle, which has led to the distinction of (1) the hybrid actuation system, (2) the hybrid motion system (mechanism), and (3) the hybrid control system. <p>This thesis describes a comprehensive study on three hybrid intelligent machine systems. In the case of the hybrid actuation system, the study has developed a control method for the true hybrid actuation configuration in which the constant velocity motor is not mimicked by the servomotor which is treated in literature. In the case of the hybrid motion system, the study has resulted in a novel mechanism structure based on the compliant mechanism which allows the micro- and macro-motions to be integrated within a common framework. It should be noted that the existing designs in literature all take a serial structure for micro- and macro-motions. In the case of hybrid control system, a novel family of control laws is developed, which is primarily based on the iterative learning of the previous driving torque (as a feedforward part) and various feedback control laws. This new family of control laws is rooted in the computer-torque-control (CTC) law with an off-line learned torque in replacement of an analytically formulated torque in the forward part of the CTC law. This thesis also presents the verification of these novel developments by both simulation and experiments. Simulation studies are presented for the hybrid actuation system and the hybrid motion system while experimental studies are carried out for the hybrid control system.