Autonomous Horizons: The Way Forward

  • Published

Autonomous Horizons: The Way Forward by Dr. Greg L. Zacharias. Air University Press, 2019, 326 pp. 

As automation sweeps across every sector of industry, defense officials must constantly update the road map for dominance in a digital future. In particular, the proliferation of autonomous systems (AS) within defense requires new ways of thinking to fully leverage new capabilities. Autonomous Horizons: The Way Forward provides the needed reference text to map the future of AS. The text is a widely sourced reference guide with hundreds of authoritative citations for further research. The book’s multidisciplinary approach provides new thinking for both novice and advanced practitioners because it covers the numerous disciplines involved in AS’s design and employment. It also provides numerous lessons learned from previously deployed AS.

Autonomous Horizons details the leading edge of AS, associated technologies, and solutions for human-system integration. Also, the text sequences numerous past paradigms on the subject to delineate the evolution among the varying schools of thought within research communities. The authors recognize the importance of common definitions and reach across government, industry, and academia to present a unified lexicon. Common definitions are critical to accommodating the convergence of six key professional communities: robotics, cybernetics, cognitive psychology, neuroscience, hard artificial intelligence (AI), and soft AI. The book correctly highlights the most important issues to remember when designing for autonomy: how do the AS interact with humans, and which human capabilities does the system augment?

At a minimum, the six identified professional disciplines are critical stakeholders within AS development. Engineering, computer sciences, and neurosciences all contribute to the necessary body of knowledge. Because the development of AS involves such a broad array of practitioners with extremely different backgrounds, the book provides a calibration mechanism for diverse teams looking to unify development and deployment strategies. To drive efficiencies and collapse development timelines, these groups must agree on common definitions, ethical priorities, and developmental frameworks. Autonomous Horizons effectively identifies theoretical advances, practical advances, and opportunities for collaboration between the diverse disciples. The collaborative approach enables the development of the needed capabilities without unintended or unethical design flaws.

Autonomous Horizons defines three key dimensions for autonomous system design: proficiency, trust, and flexibility. These three aspects earned a critical designation because they all present human-systems integration issues that must be resolved to field effective AS. In particular, the text identifies how artificial intelligence and machine learning can be technically and organizationally implemented across systems of systems.

If implemented to their full potential, AS could shift military operations and acquisition strategies from a platform-centric model to an information-centric model. At present, the platform-centric model of military systems divides and subdivides resources by mission: sensor platforms, attack platforms, support platforms, and others. Further, national assets such as space and cyber capabilities present parallel resource bases that require significant coordination across agencies. Autonomous Horizons envisions an architecture where information flow integrates across all platforms. In the same way that the worldwide web operates agnostically across a diverse array of hardware, an information-centric military will yield enhanced capabilities and improved proficiency. To deliver on these large promises, AS developers will focus on identified challenge problems, developmental processes, and organizational structures.

In addition to identifying critical human systems integration elements, Autonomous Horizons highlights the importance of designing “flexible autonomy” that enables task, peer, and cognitive flexibility. Inherent flexibility enables a system to rapidly and transparently reorient its relationship with human team members between subordinate, peer-to-peer, and supervisory roles. For example, an unmanned aerial vehicle (UAV) autonomously flies to a designated target. Once in range, the UAV pilot assumes control of weapons employment tasks while the UAV maintains basic flight control functions. In addition to moving vertically within organizational hierarchies, system architecture and organizational structure must easily shift horizontally between one-to-one, one-to-many, and many-to-one relationships to best integrate human and autonomous system resources. Expanding the previous example, a human signals intelligence analyst designates threat parameters to a fleet of orbiting UAVs (one-to-many). The squadron of UAVs autonomously search for the designated signals and promote matches to a human targeteer (many-to-one). The human targeteer approves the proposed target and passes the target to a UAV pilot for weapons engagement (one-to-one). A third dimension for scale is achievable by integrating sensor networks across manned and unmanned platforms in all domains.

Reviewer’s Recommendations

Autonomous Horizons envisions a total system redesign of warfare platforms and employment but fails to explicitly link this redesign to combat efficiency. Core arguments for fundamental changes in military acquisition programs must articulate their value in relation to their ability to accomplish items on the Joint Task List (JTL). To overcome the inertia of legacy, platform-oriented programs, product evangelists must explain how AS can fulfill JTL requirements faster, better, cheaper, or more safely to justify the switching cost.

Also, early in the text Autonomous Horizons predicts that implementing AS will “reduce manning requirements” while increasing system performance. Historically, highly technical acquisition programs have a mixed record of delivering on such promises, so these advertisements will likely be met with skepticism. Instead, autonomy advocates should seek to unitize return on investment through discrete, comparable metrics such as reconnaissance flight hour per pilot or cost of delivered weapons payload. Ideally, the unitized variable of comparison would be the most valuable or scarce commodity needed to achieve the JTL effect.

As a final criticism, the architecture strategy proposed by Autonomous Horizons uses analogies from popular platform business models like Amazon and Uber to illustrate deployment strategies for AS. These technology companies provide valuable lessons for computing intensive organizations like the Defense Intelligence Systems Agency and cyber-oriented commands that benefit from scale and commercially available hardware, but enterprise similarities drop significantly upon departure from the computing realm. Amazon and Uber’s genius involved leveraging existing infrastructure and spare capacity like the postal service and personal automobiles. However, the economy lacks few commercial options or suitable infrastructure for inherently military capabilities such as long-range strike in a contested environment. As previously mentioned, architecture redesign must begin and end with defined JTL requirements to ensure the delivery of needed military capabilities.

LCDR James M. Landreth, USN

"The views expressed are those of the author(s) and do not reflect the official policy or position of the US government or the Department of Defense."