Engineering Autonomy Mr. Robert Gold Director, Engineering Enterprise Office of the Deputy Assistant Secretary of Defense for Systems Engineering Society of Automotive Engineers (SAE) Aerospace Standards Summit McLean, VA | October 2, 2018

Outline Defense Research and Engineering (R&E) Strategy Key Research and Development Areas Background Engineering Challenges Summary

Defense Research & Engineering Strategy Mitigate current and anticipated threat capabilities Enable new or extended capabilities affordably in existing military systems Create technology surprise through science and engineering Focus on Technical Excellence Deliver Technologically Superior Capabilities Grow and Sustain our S&T and Engineering Capability

Key Research & Development Investment Areas Autonomy & Robotics Electronic Warfare / Cyber Microelectronics Hypersonics Directed Energy Manufacturing Artificial Intelligence / Man-Machine Interface Future of Computing Novel Engineered Materials Precision Sensing: Time, Space, Gravity, Electromagnetism Emerging Biosciences Understanding Human and Social Behavior

Background DoD emphasis on the increased use of autonomous systems DASD(SE), in collaboration with Services, assessed current autonomy efforts and associated engineering challenges The purpose was to ascertain the ramifications of autonomous systems on DoD engineering practice

Engineering Challenges Increase Level of Experimentation Understand autonomy trade-space for architecture/conceptual designs Engage Warfighter in experimentation to set expectations Engage Service Laboratories and Industry Partners to conduct mission-specific experiments Leverage computational biosciences, cognitive and social sciences, and robotics opportunities Standardize Taxonomy Develop autonomy-consistent terms, definitions, and phraseology (e.g., authorized/control entities, flexible/supervised autonomy, human on/outside the loop) Refine Requirements Development Apply tools to translate natural language, imagery, and video to representations and relations usable for reasoning Advance methods to encode interactions between operators and the system for requirements traceability The source of these voice-over recommendations is AFRL 711 HPW: -The bullet “Increase Level of Experimentation” – add the concept of building repositories of large labeled datasets to support future AI development. Computational biosciences is the pioneering field of study that leverages computational power of computer that advance human health and molecular understanding of life. It is the science of using biological data to develop algorithms or models to understand biological systems and relationships.

Engineering Challenges Understand/Manage Human-Machine Interaction Allocation of functions between human and machine Explore techniques for ensuring operators trust autonomous systems Advance sensor development and assessment methodologies Leverage wearable electronic sensor development and adapt to the cognitive/physical state of the warfighter Facilitate Trust and Social Interactions Develop software assurance tools to enhance ‘trust’ Define techniques for monitoring and bounding autonomous system behaviors Understand social dynamics of autonomous systems to effectively communicate and collaborate with humans The source of these voice-over recommendations is AFRL 711HPW: It is recommended that under the “Understand/Manage Human-Machine Interaction” bullet: -Voice-over on “Allocation of functions between human and machine”: Humans will likely train machines and feedback mechanisms must be designed without interfering with the human’s normal operation of the machine. -Voice-over on “Advance sensor development and assessment methodologies”:, it is recommended to note: “Assessment methodologies” include (1) machine learning and (2) advancing cognitive modeling. Examples of cognitive modeling include realistic avatars, adaptive training, and performance monitoring and prediction. (2) It is recommended that under the “Facilitate Trust and Social Interactions” bullet: -The software design will have to take into account the leveraging of human perceptual and cognitive processes. The human will need to see the display of information in and explainable and usable manner to effectively establish trust.

Engineering Challenges Enhance Analysis, Evaluation, and Certification Explore use of formal methods to analyze autonomous systems Enable rapid evolution of autonomous capabilities thru: Rapid deployment of software upgrades Perform system certifications concurrently with design Use of modular open systems architecture Synchronize Technology Development with Life Cycle Planning Rapid autonomous system development and technology transition will mandate effective coordination between engineering and product support activities.

Engineering Challenges Understand Consequences of Self-Learning Systems Evaluate consequences of autonomous system behavior being dictated by hardware, software, and system data. Artificial intelligence will allow new levels of autonomy Understand Impact to the Workforce Develop the Body of Knowledge for autonomous systems to support competency development Mission-specific work force education, training, and experience in human-autonomy teaming Establish Science, Technology, Engineering, and Mathematics relationships with academic institutions The source of this voice-over material is the AFRL 711 HPW: It is recommended that under the “Understand Consequences of Self-Learning Systems” bullet: -Self-learning systems will bring with them the ability to magnify the human-machine productivity. It will also bring with it the need to be cognitive of trust (including system safety) in its design, but also the requirement for system security to protect the system from cyber tampering. (2) Voice-over for “Understand Impact to the Work Force”: …it is important to state that DoD recognizes that the workforce must be educated and trained to support autonomous systems, artificial intelligence technologies, and the human-machine collaboration process.

Summary Fielding autonomy-enabled warfighting capability will require close collaboration with: Research, Engineering, and Test & Evaluation Acquisition and Operational Communities Our Industry Partners Collaboration needs to occur through planned demonstrations and prototyping, especially at Engineering Commands where these systems are currently designed. Autonomy technologies will impact the collective workforce, inclusive of the challenges unique to the engineering community.

Systems Engineering: Critical to Defense Acquisition Defense Innovation Marketplace http://www.defenseinnovationmarketplace.mil DASD, Systems Engineering http://www.acq.osd.mil/se

For Additional Information Mr. Robert Gold ODASD, Systems Engineering 703-695-3155 robert.a.gold4.civ@mail.mil