MACHINE LEARNING FOR BIO-INSPIRED NAVIGATION

用于仿生导航的机器学习

• 批准号: 2889696
• 金额: --
• 依托单位: University of Liverpool
• 依托单位国家: 英国
• 项目类别: Studentship
• 财政年份: 2023
• 资助国家: 英国
• 起止时间: 2023 至 无数据
• 项目状态: 未结题
• 来源: https://gtr.ukri.org/projects?ref=studentship-2889696
• 关键词: MACHINE; LEARNING; BIO; INSPIRED; NAVIGATION

This project will be concerned with assessing and replicating the different methods that animals use to navigate over long distances during migration. The initial emphasis will be on the sensory ability of birds, but the project will review all current theories for animal navigation and migration. Animals have a range of sensory inputs that we as humans lack. Examples include magnetic sensors, landmark recognition, and the ability to detect polarised light. Improvements to magnetic sensing from the UK quantum technology programme are increasing interest in the use of magnetic sensors for navigation, but there are still issues around the effect of stray magnetic fields caused by electronics and other metal components. Image based navigation has also been studied previously, but with the ability to distinguish polarised light, there is scope to combine ground feature databases with information from polarised solar illumination. The objective of the PhD will be to identify bio-inspired sensors that could be used in Positioning Navigation and Timing (PNT) applications and to develop a processing architecture that would allow such sensors to form part of a fused PNT solution.

该项目将关注评估和复制动物在迁徙过程中用于长距离导航的不同方法。最初的重点将放在鸟类的感官能力上，但该项目将回顾目前所有关于动物导航和迁徙的理论。动物有一系列我们人类所缺乏的感官输入。例子包括磁传感器、地标识别和检测偏振光的能力。英国量子技术计划对磁传感的改进增加了人们对使用磁传感器进行导航的兴趣，但围绕电子器件和其他金属部件引起的杂散磁场的影响仍然存在问题。基于图像的导航以前也被研究过，但是由于具有区分偏振光的能力，因此有可能将联合收割机地面特征数据库与来自偏振太阳照明的信息相结合。博士学位的目标是确定可用于定位导航和定时（PNT）应用的生物启发传感器，并开发一种处理架构，使此类传感器成为融合PNT解决方案的一部分。