Enhancing Subsea Navigation Capabilities

本项目致力于通过研究和开发新的技术来提高自主水下机器人(AUV)导航方法的可靠性和保真度，从而促进AUV位姿估计的发展。姿态估计是利用传感器数据估计飞行器的位置和姿态的过程。传统上，AUV的姿态估计是通过航位推算导航实现的，这是一种依赖于多普勒速度记录(DVL)和惯性测量单元(IMU)数据的导航策略。由于DVL和IMU数据的固有偏差，航位推算的位姿估计会随着时间的推移而发生漂移，这意味着位姿估计误差会随着时间的推移而线性增长。NSERC赞助的这个项目的重点是使用2G Robotics的摄像头结合DVL和IMU数据来估计AUV的姿势。特别是，本科生、硕士、博士和博士后将学习1)如何使用视觉里程计来正确估计和抵消DVL和IMU数据中的偏差，2)如何实现适用于较小AUV的计算高效的位姿估计解决方案，以及3)使用相机、DVL和IMU数据实现同步定位和映射(SLAM)算法，以改进位姿估计和定位。拟议研究的新颖性在于使用了2G Robotics最先进的摄像头和照明技术，使单目和立体摄像头能够与DVL和IMU数据结合使用，否则无法在水下环境中使用。作为这项研究的一部分，各级HQP将获得在机器人平台上开发、实施和测试尖端导航算法的经验，2G Robotics将获得相对于其他传感器供应商的竞争优势，加拿大在机器人入室盗窃领域的声誉将显著提高。

This project focuses on advancing the state-of-the-art of autonomous underwater vehicle (AUV) pose estimation through research and development of new techniques to increase the reliability and fidelity navigation methodologies. Pose estimation is the process of estimating position and attitude of a vehicle using sensor data. Traditionally, pose estimation of an AUV is realized through "dead-reckoning" navigation, a navigation strategy that relies on doppler velocity log (DVL) and inertial measurement unit (IMU) data. Due to inherent bias in both DVL an IMU data, dead-reckoning pose estimation suffered from "drift" over time, meaning the pose estimation error grows linearly with time. The focus of this NSERC sponsored project is to use 2G Robotics' cameras in conjunction with DVL and IMU data to estimate the pose of an AUV. In particular, undergraduate, master's, PhD, and postdoctoral students will investigate 1) how to use visual odometry to correctly estimate and negate bias in DVL and IMU data, 2) how to realize a computationally efficient pose estimation solution suitable for smaller AUVs, and 3) implement a simultaneous localization and mapping (SLAM) algorithm using camera, DVL, and IMU data to improve both pose estimation and localization. The novelty of the proposed research lies in the use of 2G Robotics state of the art camera and lighting technology that enables the use of monocular and stereo cameras that, otherwise, cannot be used in the underwater environment in conjunction with DVL and IMU data. As part of this research, HQP at all levels will gain experience in developing, implementing, and testing cutting-edge navigation algorithms on robotic platforms, 2G Robotics will gain a competitive advantage relative to other sensor providers, and Canada's reputation in the burgling field of robots will be significantly enhanced.