EFRI BRAID: Resilient autonomous navigation inspired by the insect central complex and sensorimotor control motifs

在过去的十年里，自然灾害的发生频率大幅增加，包括野火、洪水、山体滑坡以及农业病虫害的爆发。减轻灾害的严重性将需要早期预警和快速反应，这两项都可以得到可靠且廉价的自主机器人的帮助。不幸的是，现代机器人很难对新的环境或在灾难应对过程中可能发生的身体损伤做出反应。相比之下，由于感觉和肌肉控制系统中的冗余性和灵活性，生命系统非常善于迅速调整自己的行为以适应新的情况。对果蝇的科学发现有助于阐明这些昆虫如何在飞行中实现弹性。该项目的主要目标是将昆虫神经科学的这一新兴知识转化为能够开发出更具弹性的机器人系统。这个项目建立在现有的工程理论基础上，以开发易于理解和解释的算法。作为该项目的一部分，将为初中生、高中生和本科生提供独特的研究体验，让他们参与神经科学和机器人研究。多学科研究团队将开发开源课程内容，以帮助工程学学生流利地学习神经科学：将神经科学原理转化为工程学，以增强弹性。阻碍自主机器人推广的一个重大工程挑战是，它们在新场景中缺乏弹性。相比之下，由于感觉运动系统中的冗余性和灵活性，生物体擅长迅速调整自己的行为以适应新的环境。在工程系统中融入类似的功能被证明是具有挑战性的，因为我们缺乏如何融合来自不同传感器的信息流和协调大量执行器阵列的基本知识，而无需详细的模型和持续的校准-即使是简单的动物也可以毫不费力地实现这种复杂的操作。这项提议的总体目标是利用最近的神经生物学发现，为自主机器人开发新的设计，使其学会快速适应环境或其感觉和运动系统的变化。提出的方法从昆虫中发现的两个主题中获得灵感：1)开发用于导航的多功能、多感官指南针，2)在运动和其他行为中灵活地实施刻板印象的感觉运动主题。目标是在抽象的控制理论水平上将这些概念转化为工程原理，然后在起作用的多旋翼系统上实施和测试它们。该项目由新兴前沿研究与创新计划(BRAID)和既定的激励竞争性研究计划(EPSCoR)共同资助。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。

The last decade has seen a substantial increase in the frequency of natural disasters including wildfires, flooding, landslides, and outbreaks of agricultural pests and diseases. Mitigating the severity of disasters will require early warning and rapid responses, both of which could be aided by reliable and inexpensive autonomous robots. Unfortunately, modern robots have difficulty responding to new environments or damage to their bodies that might occur during disaster response. In contrast, living systems are remarkably adept at quickly adjusting their behavior to new situations thanks to redundancy and flexibility within the sensory and muscle control systems. Scientific discoveries in fruit flies have helped shed light on how these insects achieve resiliency in flight. The primary goal of this project is to translate this emerging knowledge from insect neuroscience to enable the development of more resilient robotic systems. This project builds on existing engineering theory to develop algorithms that are easy to understand and explain. As part of this project, unique research experiences will be offered to middle, high school, and undergraduate students to participate in both neuroscience and robotics research. The multidisciplinary research team will develop open-source course content to help bring neuroscience fluency to engineering students: translating neuroscience principles to engineering for enhanced resilience.A significant engineering challenge that has stymied the rollout of autonomous robotics is their lack of resilience in novel scenarios. In contrast, organisms are adept at quickly adjusting their behavior to new contexts thanks to redundancy and flexibility within their sensorimotor systems. Incorporating comparable functionality in engineered systems has proven challenging because we lack the basic knowledge for how to fuse information streams from different sensors and coordinate a large array of actuators without detailed models and constant calibrations—this sophisticated operation is achieved effortlessly by even simple animals. The overall goal of this proposal is to leverage recent neurobiological discoveries to develop new designs for autonomous robots that learn to adapt rapidly to changes in the environment or in their sensory and motor systems. The proposed approach draws inspiration from two themes found in insects: 1) the exploitation of a versatile, multisensory compass for navigation, and 2) the flexible implementation of stereotyped sensorimotor motifs in locomotion and other behaviors. The goals are to translate these concepts into engineering principles at an abstracted control-theoretic level, and then to implement and test them on functioning multirotor systems. This project is jointly funded by the Emerging Frontiers in Research and Innovation Program (BRAID) and the Established Program to Stimulate Competitive Research (EPSCoR).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.