RI: Small: Collaborative Research: Vision-guided Control of Robust Perching: From Biological to Robotic Flyers

在一眨眼的时间里，一只苍蝇可以通过执行一系列协调良好的动作来倒立在天花板上，这些动作是由一个不超过针头大小的大脑触发和控制的。该项目旨在揭示这一过程的强大智能是如何从计算过程和机械过程的协同组合中出现的，以及如何通过综合的生物和机器人研究将其转化为机器人飞行器。小型机器人飞行器普遍存在能耗高、气动效率低的问题。强健的栖息将显著延长小型机器人飞行器的工作时间，并扩大它们的应用，如环境和灾害监测、空中监视以及搜索和救援。该项目还将通过视觉上吸引人的昆虫飞行和机器人栖息实验向学生和公众介绍生物和机器人的运动原理，为所有年龄段的STEM学生创造一个理想的教育框架。对于小型飞行器、动物和机器人来说，在动态和不确定的环境中成功栖息是最具挑战性的特技飞行壮举之一。为了安全着陆，飞行器必须在有限的计算资源和严格的时间约束下解决距离-速度-姿态传感和控制问题。该项目将首先使用人工操纵的视觉线索进行有针对性的昆虫实验，以揭示协调蓝瓶苍蝇栖息的不同动作的计算过程，以及计算如何考虑动态的物理环境和机械过程。在生物栖息结果的指导下，该项目将使用厘米级的四轴飞行器在静止或移动的表面上完成健壮的机器人栖息。这将通过开发计算效率高的视觉算法来实现，以提取基本的视觉信息来控制机器人在触地之前的速度和方向，以便机器人能够牢固地附着在期望的表面上，并具有仿生的顺应性腿。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

In a blink of our eyes, a fly can perch upside-down on a ceiling by executing a sequence of well-coordinated maneuvers triggered and controlled by a brain no bigger than a pinhead. This project aims to unravel how the robust intelligence of this process emerges from a synergistic combination of computational and mechanical processes, and how can they be translated to robotic flyers through an integrated biological and robotic investigation. Small robotic flyers suffer universally from high energy consumption and low aerodynamic efficiency. Robust perching will significantly extend small robotic flyer's functioning time and expand their applications such as environmental and disaster monitoring, aerial surveillance, and search and rescue. The project will also introduce biological and robotic locomotion principles to students and the public through visually appealing insect flights and robotic perching experiments, creating an ideal educational framework for STEM students of all ages.For small flyers, animals and robots alike, orchestrating a successful perching in dynamic and uncertain environments is one of the most challenging aerobatic feats. To land safely, a flyer has to solve a distance-velocity-attitude sensing and control problem with limited computational resources and under stringent time constraints. This project will first use targeted insect experiments with artificially-manipulated visual cues to reveal the computational processes that coordinate different maneuvers of perching in blue bottle flies, and how the computation takes into account dynamic physical environments and mechanical processes. Guided by the results from the biological perching, this project will accomplish robust robotic perching on stationary or moving surfaces using a centimeter-scale quadcopter. This will be achieved by developing computationally-efficient vision algorithms to extract essential visual information to control the robot's velocity and orientation prior to touchdown so that the robot can firmly attach to the desired surfaces with biomimetic compliant legs.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.