Visual Flight Control for the Very Smallest Aerial Vehicles

点击翻译按钮获取中文摘要

The goal of this project is to realize autonomous robots as small as bumblebees and even gnats by creating design guidelines for their vision-based control systems. This will have a revolutionary impact on robotics. Compared to larger robots, insect robots will deploy in larger numbers, in more confined spaces, in closer proximity to humans without impact hazard, and do so persistently by collecting small amounts of energy that are available in the environment such as sunlight. As a result they will sense environmental conditions in far greater detail than is currently possible. An important application is locating leaks of the greenhouse gas methane, of which the USA is the world's leading exporter. Other examples include locating incipient fires, monitoring agriculture from within the plant canopy, inspecting confined spaces, and exploring outer space at reduced launch cost. To date, however, robots below a gram have limited sensor and power systems and have not yet performed controlled flight without feedback from external sensors because of extreme size, weight, and power constraints. Our objectives are to better understand how the physics of small scale affects the design of insect-sized feedback control systems. The educational objectives are to drive interest and provide training in engineering and robotics for students in high school through graduate school, particularly minorities. This Foundational Research in Robotics project will address the objective of understanding how the physics of small scale affects the design of insect-sized feedback control systems for flying robots smaller than one gram. In particular, the project’s research activities will include: 1) modelling and analyzing the effects of physical scale on robot sensing, control, and power, 2) designing and analyzing a hovering controller compatible with the constraints of the very smallest, millimeter-sized flying robots, to be released as open source software, and 3) designing and analyzing a system for learning basic visual navigation through a cluttered environment while seeking a source such as a chemical or power (also open source). Findings will be tested in simulation, as well as on a bumblebee-sized 300 mg flapping-wing flying robot and a novel sensor suite. The result will be the first demonstration of stable hovering and obstacle navigation on an insect sized robot. To minimize the power required for feedback control, only passive sensing such as vision, a fly-inspired wind sensor, and exclusively multiply and add operations for computation will be used. Free lesson plans and other online resources to help educators teach students how to design, build, and operate “foldable robotics” inspired by the insect robots used in this work will be created and publicly disseminated.This project is supported by the cross-directorate Foundational Research in Robotics program, jointly managed and funded by the Directorates for Engineering (ENG) and Computer and Information Science and Engineering (CISE).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.