Sliding Mode Control of Motion Platform and Human-Perception-Systems Based Washout Filter Design for Virtual Reality Simulation.

• Main Authors: Chin-I Huang, 黃勤鎰
• Other Authors: 傅立成
• Format: Others
• Language: en_US
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/92487931787301487066

博士 === 國立臺灣大學 === 電機工程學研究所 === 95 === The driving simulator is a great integrated system which can be divided into four subsystems, including vehicle dynamics systems, computer graphs, motion platform systems, and washout filter systems. This dissertation focuses on the construction of a high-performance motion platform system with a more realistic washout algorithm for human sensation. A smooth sliding mode backstepping control approach for the motion control of a Stewart platform is proposed. The control scheme is proposed provided that the overall system parameters are subject to uncertainties and only the positions and velocities of the links are measurable. To achieve high-performance tracking control of a 6 DOF Stewart platform normally requires the full knowledge of the system dynamics. In this dissertation, some important properties of the dynamics of the Stewart platform have been derived and exploited to develop a smooth sliding mode backstepping controller which can drive the motion tracking error to zero asymptotically. Stability analysis based on Lyapunov theory is performed to guarantee that the controller design is stable. Finally, the experimental results confirm the effectiveness of our control design. After that, a new approach is presented to develop washout filters for the simulators. It is based on the human vestibular model, limited workspace of the motion platform, and the senseless maneuver compensator at the same time for designing a washout filter such that a cost function constraining the pilot’s sensation error (between the simulator and the simulated vehicle) is minimized. The strong sensations experienced by the pilot can be curtailed, and the platform workspace for presenting the desired scenario is more efficient than others. Finally, the simulation and experimental results confirm the effectiveness of our algorithm designed, and hence the merits of the present approach can be manifested. To avoid dizzy feeling caused by the conflict between the visual and the vestibular systems, a novel integrated model is proposed which includes models of both vestibular and visual motion sensation and incorporates the nonlinear interaction between the vestibular and visual stimuli. Models for both rotational and translational motions are developed, producing responses that explain the characteristics of self-motion. Based on this novel model, we propose a human vestibular-visual based (HVVB) adaptive washout algorithm to avoid dizzy feeling caused by the conflict between visual and vestibular systems. Finally, the simulation and experimental results confirm the effectiveness of our proposed algorithm, thus manifesting the advantages of the present approach.