Interactive motion simulators are extensively utilized in not only flight simulation but also the entertainment field. A conventional flight simulator system with six degrees of freedom (DOF) successfully delivers excellent “sustained” and fair “onset” motion cues with long travel distance characteristics. However, the extremely expensive cost and large space requirement have limited their use for entertainment.
Chiang and Chieng (1995) developed a prototype, the SP-120, based on the concept of the Steward platform. The new structure allows the SP-120 to generate improved onset cues and provide sustained motion. Additionally, the space requirement is much less than the conventional one. However, this prototype with six DOF remains too expensive for commercial use. Thus, reducing cost, simplifying the mechanism structure and reducing the order of the actuator system are worthy goals.
This dissertation designs a low-cost motion simulator system. The new motion simulation system has two significant goals: (1) mechanism design for motion platform; and, (2) establish a motion-cueing control strategy. In this work, the optimal design of the new platform, called X-2, is based on a parallel kinematic mechanism structure, which is discussed in Chapter 2. Chapter 3 presents a motion-cueing control strategy for the proposed simulation system.
Parallel manipulators have many benefits over conventional serial manipulators in terms
of accuracy, velocity, stiffness and payload capacity, and are therefore widely adopted in industry. Parallel manipulators [1]–[4] with fewer than six DOF have recently been extensively adopted for various uses as they maintain the advantages inherent in parallel mechanisms, and have several other benefits such as reduced total manufacturing and operation costs.
Performance indices of a parallel kinematic machine (PKM) [5] may include workspace [6], [7], actuator capability, power transmission efficiency, architecture design [8]–[12] and possibly the best accuracy [13].
Various studies of manipulator performance focused on analyzing the manipulator’s kinematic properties represented by the Jacobian transformation. These efforts yielded important measures for and characterizations of kinematic properties [14]–[17] and static force capability [18].
Although many parallel mechanisms have been developed] [8], [19], [20], less attention is given to optimal design of a manipulator that has optimal workspace features, or the best mechanical advantages in motion relative to the rotational parallel mechanism. Notably, workspace features and mechanical advantages are significant characteristics in controlling parallel manipulators, particularly for those applied as motion platforms.
Chapter 2 analyzes the global optimization of a two-prismatic-universal-universal (2-PUU) PKM to obtain the best synthetic properties of performance indices, and presents a
novel optimization procedure that, via the use of global optimal searching techniques, addresses some major issues associated with static (workspace) and dynamic (actuator capability) performance. The optimal parallel manipulator design solution is computed using a well-known global optimization algorithm, a genetic algorithm (GA) [21]–[23]. The performance index can be treated as a compromise between the optimal mechanical advantage design and optimal workspace design based on the weight of each term.
After the optimal design for the platform mechanism is obtained, an appropriate motion-cueing control strategy is applied on this platform. To form an optimal motion-cueing strategy for the simulation system, some basic concepts of motion-cueing theory should first be addressed.
A motion simulator attempts a realistic impression of vehicle motion, such as that of an aircraft or racing car. Unfortunately, this goal is not easily achieved because simulators are limited by workspace features and actuator capabilities such as maximum torque and velocity.
Engineers have improved simulator motion by developing motion-cueing strategies, known as
“washout filtering.” Washout filtering is intended to transform trajectories generated by a dynamic virtual reality (VR) model incorporating very large displacements into driving system commands that generate realistic motion cues for a pilot within the simulator’s limited workspace.
Washout separates motion cues into high- (onset) and low- (sustained) frequency
components, such that cues to be managed and displayed within the physical confines of a given platform system. Washout must provide a high-pass filtering system, which may be linear or nonlinear, to limit simulator cab excursions. Nonlinear designs include adaptive filters and other optimal control techniques that are applied based on various criteria.
Many schemes for motion-cueing control have been presented. Schmidt and Bjorn [24]
analyzed motion drive signals for piloted flight simulators. Conrad and Schmidt [25]
proposed techniques for calculating motion drive signals. Sinacori [26] proposed a practical approach for motion simulation. Bowles, Parrish and Dieudonne [27] applied coordinated adaptive washout to motion simulators. Sivan, Ish-shalom and Huang [28] applied an optimal control approach for the design of moving flight simulators. Ariel and Sivan [29] addressed false cue reduction in moving flight simulators. Reid and Nahon [30] developed an algorithm that drives a flight simulator. Nahon and Reid [31] developed simulator motion-drive algorithms. Reid, Nahon and Kirdeikis [32] developed adaptive simulator motion software that has supervisory control. Idan and Sahar [33] presented a robust controller for a simulator with six DOF. Pouliot, Gosselin and Nahon [34] analyzed motion simulation capabilities of flight simulators with three DOF. Moshe and Nahon [35] analyzed an offline comparison of classical and robust flight-simulator motion controls. Martin [36] considered the whole body motion of motion cueing. Liao and Chieng [37] proposed another novel washout filter algorithm for a motion simulator with six DOF. Chang, Liao and Chieng [38] developed a
master switching technique for electronic cam control with special reference to multi-axis coordinated trajectory following.
The theory and development of an optimal algorithm for a flight simulator with six DOF have recently been discussed by Wu and Cardullo [39] and Telban and Cardullo [40]. Their approach incorporates a mathematical model of the human vestibular system that constrains pilot sensation of error between the simulated aircraft and platform motion dynamics. The problem is to determine a transfer function matrix that relates the desired simulator motion input to aircraft input, such that a cost function constraining pilot sensation error (between a simulator and plane) is minimized.
However, the aforementioned studies focused on motion simulators with full spatial DOF,
i.e., six DOF; the problem of real-time optimal motion-cueing techniques for simulating specific virtual reality (VR) motion in a motion simulator with limited DOF has rarely been addressed.
To develop a motion-cueing control strategy for motion simulators with rotational DOF without loss of generality, a full rotational DOF platform, the X-360, which is a modified version of the X-2 platform, is adopted for experimental testing.
Chapter 3 presents a novel algorithm for evaluating a real-time optimal motion-cueing strategy for a motion simulator solely with three rotational DOF (yaw, pitch and roll). This algorithm optimizes the additional linear onset cues, providing the attitude and sustained cues
are remained. The proposed algorithm comprises a classical linear washout filter (CLWF), a yawing washout filter (YWF), an adaptive washout filter (AWF) and a real-time optimal motion-cueing algorithm (ROMA). The proposed algorithm individually transforms high- and low-frequency linear motions into output angles of a motion simulator with rotational DOF (3-DOF). These output angles are incorporated into the cockpit attitude control to achieve five DOF motion. The ROMA first defines a quadratic cost function to be minimized. This cost function, which corresponds to the performance index of five DOF motion, is then decoupled into three Euler angles associated with the three DOF simulator. The restrictions of workspace and actuator capabilities are represented as inequality constraints of the motion performance optimization problem. Since the cost function has a quadratic (convex) form, Karush-Kuhn-Tucker (KKT) conditions can be introduced to locate the global optimum. Prior to motion optimization via the ROMA, the YWF is applied to prevent simulator cab excursion from exceeding the workspace. After motion optimization via the ROMA, the AWFs are applied when necessary to reset simulator position gradually. All washout motions are performed in insensible acceleration or rate to the pilot. The remaining Euler angles of the three DOF simulator, i.e., pitch and roll, should simultaneously account for cockpit angular motion and residual tilt during linear motion. The bounds of pitch and roll angles are formulated implicitly, and are calculated during each sample time. Motor commands are obtained by substituting the desired Euler angles into an inverse kinematics model of the three
DOF simulator.
The remainder of this dissertation is organized as follows. Chapter 2 describes the optimal design for the X-2 platform using a GA. Chapter 3 establishes the ROMA for the simulation system. Chapter 4 presents experimental setups and detailed results. Appendices A, B, C and D present the kinematics and Jacobian relations. Tables 1, 2 and 3 show all the electrical, mechanical and software parameters and setup. Synthesis of the characterization and measures into an optimization procedure is then discussed. Finally, the optimization procedure is applied to the design and control for a novel motion simulator system with rotational DOF.