Rañó Iñaki
Institut für Neuroinformatik, Ruhr-Universität Bochum, Bochum, Germany.
Biol Cybern. 2012 Jul;106(4-5):261-70. doi: 10.1007/s00422-012-0493-7. Epub 2012 May 31.
Motion camouflage is a stealth behaviour observed both in hover-flies and in dragonflies. Existing controllers for mimicking motion camouflage generate this behaviour on an empirical basis or without considering the kinematic motion restrictions present in animal trajectories. This study summarises our formal contributions to solve the generation of motion camouflage as a non-linear optimal control problem. The dynamics of the system capture the kinematic restrictions to motion of the agents, while the performance index ensures camouflage trajectories. An extensive set of simulations support the technique, and a novel analysis of the obtained trajectories contributes to our understanding of possible mechanisms to obtain sensor based motion camouflage, for instance, in mobile robots.
运动伪装是在食蚜蝇和蜻蜓中都观察到的一种隐身行为。现有的用于模仿运动伪装的控制器是基于经验生成这种行为的，或者没有考虑动物轨迹中存在的运动学运动限制。本研究总结了我们在将运动伪装的生成作为一个非线性最优控制问题来解决方面的正式贡献。系统动力学捕捉了智能体运动的运动学限制，而性能指标确保了伪装轨迹。大量的模拟支持了该技术，并且对所得轨迹的新颖分析有助于我们理解例如在移动机器人中获得基于传感器的运动伪装的可能机制。
