At RTCA’s August webinar, Rebecca Morrison and FAA’s Chief Scientist and Technical Advisor for Artificial Intelligence (AI)/Machine Learning (ML), Dr. Trung T. Pham discussed the FAA’s recently published, Roadmap for Artificial Intelligence Safety Assurance. This webinar on AI/ML algorithms discussed the role of software for aviation systems, the role of existing standards like DO-178C, and the future of technical guidance on AI/ML, follows the release of the FAA’s AI Roadmap and RTCA’s recent AI/ML workshop that drew more than 100 participants.
The guiding principles of FAA’s Roadmap for Safety of AI in Aviation includes effectively using as much existing regulation as possible, taking an incremental approach, and extensive FAA/Industry collaboration, said Dr. Pham. As aviation seeks to incorporate AI/ML in aviation development, government and industry must conduct research and development for guidance and use computational data analytics to support verification. An important point to remember is that any AI/ML algorithm is the creation of engineers, and developers will always be held responsible for the products they produce.
Dr Pham reminded listeners that FAA regulates safety, and the FAA’s roadmap is focused on integrating AI into safety critical applications for aviation. Dr. Pham mentioned the good record of safety that the industry enjoys which has been supported by existing regulations and standards such as DO-178C on how to develop software for aviation. The aviation industry uses this existing framework on the path forward with AI.
The challenges in implementing AI include the need for clearly defined terminology, integration and traceability into the existing approval process on the path to certification, and with coordination to reduce the risk of duplicating efforts.
Dr. Pham announced the FAA will be releasing its Technical Concept Paper in the near future. These papers will address software development guidance on ML, ML in the context of model development, software implementation of computational algorithms, and specific examples of use cases relative to the existing standards, like DO-178C, for software development. He is looking forward to seeing the research on the use of AI for safety, and said the FAA will work with the technical community toward sensible and feasible guidance/regulation of AI technology.
To view this webinar, click here. RTCA’s next AI/ML workshop will take place on November 13. Members can visit AerOpus and request to join the AI/ML group for more details here.
To view the FAA’s Roadmap for Artificial Intelligence Safety Assurance, click here.
