Optimisation of nonlinear motion cueing algorithm based on genetic algorithm
Optimisation of nonlinear motion cueing algorithm based on genetic algorithm
复制标题
DOI：
10.1080/00423114.2014.1003948
复制
发表时间：
2015-04-03
影响因子：
3.6
通讯作者：
Nahavandi, Saeid
中科院分区：
文献类型：
--
作者：
Asadi, Houshyar;Mohamed, Shady;Nahavandi, Saeid
运动提示算法(MCAS)在驾驶模拟器中扮演着重要的角色，其目标是在不超过模拟器的物理限制的情况下，向模拟器驾驶员提供与真实车辆驾驶员相比最准确的人类感觉。本文对车辆模拟器的MCA进行了优化设计，以求在考虑所有运动平台物理限制的同时，找到最合适的冲刷算法参数，并最小化真实驾驶员和模拟器驾驶员之间的人的感知误差。经典洗出过滤器的主要局限性之一是，它是通过最坏情况调优方法进行调整的。这是基于试错，并受到驾驶和程序员经验的影响，这使这成为全动平台利用的最大障碍。这导致了结构的僵化，产生了错误的提示，并使得到的模拟器不能适应所有情况。此外，经典方法没有考虑人的感知误差和物理约束的最小化。出于这个原因，对于设计者来说，动作提示的产生以及传统冲淡滤光器的不同参数对动作提示的影响仍然是无法理解的。本文的目的是提供一种基于非线性滤波和遗传算法的MCA参数整定优化方法。这是通过考虑真实和模拟情况之间的前庭感觉误差，以及主要的动态限制、倾斜协调和相关系数来实现的。在MCA中集成了三个额外的补偿线性块，以进行调谐，以成功地修改滤波器的性能。在MATLAB/SIMULINK软件包中实现了所提出的优化MCA。使用该方法生成的结果表明，在不达到运动限制的情况下，在人类感觉、参考形状跟踪和更有效地利用平台方面，性能得到了提高。
Motion cueing algorithms (MCAs) are playing a significant role in driving simulators, aiming to deliver the most accurate human sensation to the simulator drivers compared with a real vehicle driver, without exceeding the physical limitations of the simulator. This paper provides the optimisation design of an MCA for a vehicle simulator, in order to find the most suitable washout algorithm parameters, while respecting all motion platform physical limitations, and minimising human perception error between real and simulator driver. One of the main limitations of the classical washout filters is that it is attuned by the worst-case scenario tuning method. This is based on trial and error, and is effected by driving and programmers experience, making this the most significant obstacle to full motion platform utilisation. This leads to inflexibility of the structure, production of false cues and makes the resulting simulator fail to suit all circumstances. In addition, the classical method does not take minimisation of human perception error and physical constraints into account. Production of motion cues and the impact of different parameters of classical washout filters on motion cues remain inaccessible for designers for this reason. The aim of this paper is to provide an optimisation method for tuning the MCA parameters, based on nonlinear filtering and genetic algorithms. This is done by taking vestibular sensation error into account between real and simulated cases, as well as main dynamic limitations, tilt coordination and correlation coefficient. Three additional compensatory linear blocks are integrated into the MCA, to be tuned in order to modify the performance of the filters successfully. The proposed optimised MCA is implemented in MATLAB/Simulink software packages. The results generated using the proposed method show increased performance in terms of human sensation, reference shape tracking and exploiting the platform more efficiently without reaching the motion limitations.