Development of motion cueing algorithm for simulators

Abstract

Flight simulators are widely used for commercial and military purposes for training of pilots. Additionally, they are increasingly employed for ground, marine, and space vehicles. Simulator systems help us to understand driver/pilot behavior in simulations as close to real flight. Real aircraft flights are very expensive and have numerous risks. Flight simulators are safe and practical structures. During The Second World War, flight simulators were required in military. Technological developments in analog computers contributed to flight simulator developments. Following years, some specific options like aerodynamics and flight behavior were incorporated in simulators. There is no specific standard for motion cueing, but in literature, four major algorithms are currently present; classical, optimal, adaptive and model predictive base algorithms. Motion cueing algorithm is a must that provides the motion of a simulator within boundaries of limited robotic workspace. 6 DOF parallel manipulators are mostly used for flight simulation. Inputs to the manipulator are simulated aircraft angular velocities and translational accelerations as references to generate necessary motions. The mobile platform of the mechanism should return its neutral position in order to start the next motion sequence. This process is called as “Washout”. Washout motion must be under human motion sensation thresholds which is covered in literature by several researchers. For the sake of sensing continuous translational accelerations, the platform tilts itself and uses gravity vector to create artificial motion sensation translationally which is called “Tilt Coordination”. Force effect caused by continuous acceleration of aircraft is also provided by Tilt Coordination. Tilt coordination is provided by applying 2nd order low-pass filters for x- and y- axes in the algorithm. Basically a simulator consists of a robotic manipulator and a simulator room on the top and kinematic analysis of the manipulator is crucial to obtain meaningful motion results. Translational and rotational positions obtained from the output of motion cueing algorithm are transmitted to the manipulator as actuator extensions calculated by inverse kinematics. Coordinate transformations in a robotic system are another important topic because motion sensation must be at the eye point of the pilot/driver. The major issue in simulators is motion cueing algorithm that relates the simulator input to the aircraft input. This is possible by the communication between motion control computer and simulation computer. Communication protocol in presented system in this study receives the translational accelerations and angular velocities from the simulation package program called FlightGear via UDP protocol. For this purpose, an XML script file is generated to receive related motion control inputs and take them into the motion cueing algorithm to be processed. If aircraft dynamic model is used in simulation control computer via Matlab/Simulink, it is needed to receive related motion parameters as packages by using package input/output block. Classical motion cueing algorithm constitutes the basis of motion cueing algorithms and includes 1st order and 2nd order high-pass washout filters that are used to bring the manipulator to its neutral position in its workspace under the threshold values of human sensation for all motions. Thus, vestibular system model is necessary to check the motion whether it is under the threshold value or not. Vestibular system is in inner ear of a human and the system consists of semi-circular canals sensing rotational motions and otoliths sensing translational motions. Calibration process of classical washout filters is also important in motion cueing algorithm design. There are position saturation values through x-, y- and z- axes and angle saturation values about x-, y- and z- axes of the manipulator according to its production catalog. If filter gains are higher than the saturation values that manipulator is capable of, there is possibility of the occurrence of mechanical failures. Tilt coordination is another important topic because force effect caused by continuous acceleration of aircraft is provided by tilt coordination. Tilt coordination is provided by applying 2nd order low-pass filters for x- and y- axes in the algorithm. In literature there are not many works explaining the calibration process of washout filters. Previously, we implemented classical washout filter on a small scale desktop platform. After a significant optimization was applied, a full scale flight simulator mechanical structure has been constructed and classical washout filter was implemented. The system is on a 6x6 parallel manipulator and is driven by AC motors. The classical washout filters were implemented on Matlab – Simulink environment with Real-Time Windows Target. This algorithm has first and second order high pass filters for high frequency components and first order low pass filters for low frequency components of aircraft flight variables. Also, all software implementations are established on Matlab. Flight data are obtained from virtual flight software with 5 Hz sampling rate. In this study, system integration of a low payload flight simulator is explained, results of classical motion cueing algorithm, vestibular system validations and design and calibration process of classical washout filter are presented and comparison between results obtained from Boeing 777-300ER and Sikorsky flights are shown. Many tests according to system requirement documents are performed and results of these tests are also presented. Furthermore, accuracy of motions obtained from the output of the manipulator is proven by using forward kinematics. As well as classical motion cueing algorithm, optimal washout filter design and its implementation are explained in this study. At the end of the paper conclusions are presented.