CAREER: Enabling Autonomy via Enhanced Situational Awareness for Underwater Robotics

水生环境的作用非常重要：它在气候中起着至关重要的作用；它拥有地球上最高的生物多样性；港口是贸易和运输最重要的基础设施之一；全球40%的人口居住在海岸线100公里以内。提高我们对水下领域的理解是必不可少的。使用自主机器人收集额外的信息将更安全，更具成本效益，并且可以扩展到比以前的方法更大的规模。该项目的目标是使机器人系统能够在水下环境中自主操作。为了实现这一目标，机器人需要感知环境，以及自身在环境中的位置。机器人需要制定运动策略，以便有效和准确地估计其位置以及环境中障碍物/物体的位置，同时考虑到测量中的误差。水下领域面临着几个独特的挑战：没有全球定位系统（GPS）；可用的通信带宽极其有限；即使在最好的情况下，能见度也会因为水中的微粒阻碍视野而受到限制。研究人员将在四个领域推进最新技术：来自不同机器人传感器的信息将用于计算机器人在水下区域移动时的位置；然后，研究人员和他的研究生将使用所有可用的信息来生成机器人可以用来导航的环境的表示；接下来，将实施规划，引导机器人通过考虑到较短的距离和有观赏兴趣的区域的环境；最后，该团队将研究有效探索未知环境的新策略。研究者将利用他在水下领域的研究成果来提高学生和普通大众对科学、技术、工程和数学的兴趣。该项目的目标是使机器人系统能够在水下环境中自主操作。为了实现这一目标，机器人需要态势感知。此外，机器人需要制定运动策略，以便有效准确地估计其姿态和环境中感兴趣点的位置，同时考虑到不确定性的积累和风或电流等外力的影响。水下区域使得基于卫星的GPS失效。可用的通信带宽极其有限；由于雾和模糊、光照随时间变化和颜色损失，能见度条件受到限制。研究人员将在四个方面推进最先进的技术：来自不同传感器的信息将用于计算机器人在水下移动时的姿势；将利用所有可获得的资料，对环境作出密集的描述；接下来，将执行一个决策过程来引导机器人通过环境，同时考虑效率（较短的距离）和观看兴趣区域；最后，将研究有效探索和覆盖未知环境的新策略。更具体地说，将为状态估计开发鲁棒性度量和散度预测因子，以便提供错误估计的早期预警。测量不同传感器的质量将导致明智地使用提供准确信息的传感器子集。映射的挑战将通过增强基于特征的地图与从照明变化产生的特征，如阴影和腐蚀模式来解决。覆盖模式将在障碍物有限的开放区域采用，而基于边界的策略将引导水下航行器到未探索的区域。以系统的方式返回映射区域将使定位不确定性保持在用户定义的水平以下。研究结果将发表在机器人的会议和期刊上。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The role of the aquatic environment is of great importance: it plays a critical role in climate; it contains the highest biodiversity on the planet; ports are among the most critical infrastructures for trade and transportation; and as much as 40% of the global population lives within 100km of the shoreline. Improving our understanding of the underwater domain is essential. Using autonomous robots to collect additional information will be safer, more cost effective, and can be extended to a larger scale than previous methods. The goal of this project is to enable autonomous operations of robotic systems in underwater environments. In order to achieve this goal, the robot needs to be aware of the environment, and of its own position inside the environment. The robot needs to develop movement strategies that would facilitate the efficient and accurate estimation of its position and the location of obstacles/objects in the environment, while taking into account the errors in the measurements. The underwater domain presents several unique challenges: there is no Global Positioning System (GPS); communication, when available, has extremely limited bandwidth; visibility conditions, even in the best-case scenarios, are limited due to particulates in the water that obstruct the view. The investigator will advance the state of the art in four areas: information from different robot sensors will be used to calculate the position of the robot as it moves through the underwater domain; then, the investigator and his graduate students will use all available information to produce a representation of the environment the robot can use to navigate; next, planning will be implemented to guide the robot through the environment taking into account the shorter distance and the areas with viewing interest; and finally, the team will investigate new strategies for exploring unknown environments efficiently. The investigator will use his research results from the underwater realm to raise interest for students and the general populace towards science, technology, engineering, and mathematics. The goal of this project is to enable autonomous operations of robotic systems in underwater environments. In order to achieve this goal, the robot needs situational awareness. Additionally, the robot needs to develop motion strategies that would facilitate the efficient and accurate estimation of its pose and the location of points of interest in the environment, while taking into account uncertainty buildup and the effect of external forces such as wind or current. The underwater domain renders satellite-based GPS ineffective. Communications, when available, have extremely limited bandwidth; and visibility conditions are limited due to hazing and blurring, lighting variations over time, and color loss. The investigator will advance the state of the art in four areas: information from different sensors will be used to calculate the pose of the robot as it moves through the underwater domain; all available information will be utilized to produce a dense representation of the environment; next, a decision process will be implemented to guide the robot through the environment taking into account efficiency (shorter distance) and the areas with viewing interest; finally, new strategies for exploring and covering unknown environments efficiently will be investigated. More specifically, robustness measures and divergence predictors will be developed for the state estimation in order to provide early warnings of erroneous estimates. Measuring the quality of the different sensors will result in the judicious use of the subset of sensors that provide accurate information. The mapping challenge will be addressed by augmenting the feature-based map with features generated from the lighting variations, such as shadows and caustic patterns. Coverage patterns will be employed in open areas with limited obstacles, while a frontier-based strategy will guide the underwater vehicle to unexplored areas. Returning to mapped areas in a systematic manner will maintain the localization uncertainty below a user defined level. The results will be published in conferences and journals of robotics.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.