UDRI is a national leader in scientific and engineering research, serving government, industry and nonprofit customers. UDRI contributes to the fulfillment of the University’s commitments to education, research and public service through the involvement of students, faculty and external partners. Join our team today as an Autonomy Researcher! Verification of autonomy is known to be a hard problem. Autonomous systems include complex decision-making mechanisms that are likely to be sensitive to changing mission conditions, resulting in a “state space explosion” of possible system behaviors that traditional test-based approaches cannot adequately cover. To address this problem, “formal methods” – i.e., mathematically-based tools and techniques for design, specification, and verification – can supplement testing by providing evidence of system correctness through proof and analysis. Formal methods, in conjunction with existing best practices for system and software development, can also be used to make autonomy software more robust and reliable by checking for general properties such as absence of runtime errors, memory leaks, access to undefined memory locations, etc. Using OpenUxAS, an autonomy framework developed by AFRL, one of our goals is to make it easier for our collaborators, e.g. academic or industrial partners who are using it as a platform to build or test autonomous or human-automation systems, to use OpenUxAS in their own work. However, our larger goal is to use OpenUxAS as a case study and demonstration of best practices for software development, documentation, and verification using formal methods alongside more traditional approaches. To date, we have been working to formalize requirements and verify individual services in UxAS (Aiello, et al. 2019), which we plan to continue. We are also working to restructure UxAS to better enable traditional approaches like unit testing. We are also working to develop better documentation for UxAS, including user guides and developer guides, and we are developing additional tutorial examples for UxAS and associated simulation framework OpenAMASE (AFRL/RQ 2020). We are also working with collaborators who are developing verified algorithms for OpenUxAS, e.g. path planning algorithms. Since OpenUxAS is implemented as a service-oriented architecture, we also plan to develop methods to compositionally verify that collections of services satisfy higher-level system properties. We are also interested in developing a multi-vehicle air collision avoidance algorithm using NASA DAIDALUS (NASA 2020) as a foundation, incorporating it into OpenUxAS, and demonstrating it in OpenAMASE.
Stand Out From the Crowd
Upload your resume and get instant feedback on how well it matches this job.
Job Type
Full-time
Career Level
Mid Level
Number of Employees
1,001-5,000 employees