CAREER: Robust and Lightweight Formal Methods for Mobile Robot System Development

移动机器人有可能扩大人们的能力，减少目前人们在广泛的人类工作中手工完成的枯燥、肮脏或危险的工作，特别是在医疗保健、国防、建筑和农业方面。然而，构建移动机器人是具有挑战性和昂贵的，因为机器人系统开发人员努力以经济高效的方式构建、测试和维护可靠的软件。我们对机器人系统开发过程的了解存在差距，这阻碍了构建支持该过程的软件工具。以目前的工具和技术，特别难建立一个原型机器人，可以升级为安全、可靠和可靠的移动机器人，而不必从头开始。该项目旨在创建轻量级技术和软件工具，帮助机器人系统开发人员制造更安全、更有能力的机器人，同时保持经济上的可行性。这项工作利用视频游戏开发工具的进步，使测试移动机器人应该做什么和不应该做什么变得更容易。制造移动机器人需要几个领域的专业知识，包括电气、机械和软件工程，而协调这些领域的信息可能是具有挑战性的。除了构建特殊的软件工具，该项目还将研究和教学结合在一起，以便最近在工程学方面的进展，称为基于模型的工程，可以使学生准备成为未来跨学科劳动力的一部分，能够构建、操作和维护更可靠的移动机器人。该项目利用行为树的模型检查和物理单元的抽象类型推理来自动建议系统测试，并帮助确保不存在某些类别的软件缺陷。自动化程序分析和测试是现代软件持续集成和部署(CI/CD)的基本部分，但机器人开发人员努力实现系统测试工作的自动化、跟踪和评估。目前的技术不能满足软件/硬件系统的异质协同进化的需求，特别是当系统在现场进行验证时。该项目旨在开发从机器人系统开发人员已经使用的现有构件中推断机器人系统测试用例的技术。此外，该项目寻求创建用于推断和比较通常一起使用的物理单元组的技术，以更好地理解程序并发现如何在机器人系统实现中使用物理单元的新模式。这个项目的预期结果包括创建一个软件工具，帮助开发人员检测和跟踪他们的软件中共同出现的物理单元，创建一个工具，帮助机器人系统开发人员测量他们的系统执行了多少行为树，以及为移动机器人系统的现场测试开发测试用例优先策略。这项研究将基于模型的测试覆盖范围的概念从软件工程应用到机器人自主，该研究领域的一个目标是降低将正式方法应用于机器人软件系统的障碍。这一奖项反映了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。

Mobile robots have the potential to amplify people's capabilities and reduce the amount of dull, dirty, or dangerous work that people currently do by hand across a broad set of human endeavors, especially in health care, national defense, construction, and agriculture. However, building a mobile robot is challenging and expensive because robot system developers struggle to cost-effectively build, test, and maintain reliable software. There is a gap in our knowledge of the robotic system development process that hinders building software tooling to support that process. With current tools and techniques, it is especially hard to build a prototype robot that can be upgraded into a safe, reliable, and dependable mobile robot without having to start from scratch. This project aims to create lightweight techniques and software tools that help robot system developers build safer and more capable robots while remaining economically feasible. This work leverages advances in video game development tools to make it easier to test what a mobile robot should and should not do. Building mobile robots requires expertise in several areas, including electrical, mechanical, and software engineering, and coordinating information across these areas can be challenging. In addition to building special software tools, this project puts research and teaching together so that recent advances in engineering, called model-based engineering, can prepare students to be part of an interdisciplinary future workforce that can build, operate, and maintain more reliable mobile robots.This project utilizes model checking of behavior trees and abstract type inference of physical units to automatically suggest system tests and to help ensure the absence of certain classes of software defects. Automated program analysis and testing are fundamental parts of modern software's continuous integration and deployment (CI/CD), but robot developers struggle to automate, track, and assess system testing efforts. Current techniques fall short of the demands of heterogeneous co-evolution of software/hardware systems, especially when systems are validated in the field. The project aims to develop techniques for inferring robotic system test cases from existing artifacts that robot system developers are already using. Additionally, this project seeks to create techniques for inferring and comparing groups of physical units that are commonly used together to better understand programs and discover novel patterns in how physical units are used in robotic system implementations. The expected outcomes of this project include creating a software tool that helps developers detect and track the physical units that co-occur in their software, creating a tool that helps robot system developers measure how much of a behavior tree their system has exercised, and developing test case prioritization strategies for field testing of mobile robotic systems. This research applies notions of model-based test coverage from software engineering to robot autonomy, and a goal of this research area is to lower the barriers to applying formal methods to robotic software systems.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.