Autonomous Intelligent Systems
© 2004 FrontLine Defence (Vol 1, No 1)

The future military is likely to include a collaborating, inter­operable mix of humans and technologically smart entities, called Autonomous Intelligent Systems.

Autonomous Intelligent Systems (AIS) are computer-automated systems that perform independent planning of an operation based on the high level instructions that they receive and the details of the local environmental situation provided by their sensors. They may function in the real world or in cyberspace.

AIS have a variety of commercial, industrial, medical, physical security as well as defence/military and national security uses and relate to many scientific and engineering disciplines and applications. They are based on enabling technologies such as: discrete event control; artificial intelligence; learning algorithms and methodologies; pattern recognition; image processing; and, human-machine interface technologies. AIS may also be coupled to technological advances in such other fields (nanotechnology, biotechnology, lasers, etc.) in order to derive novel or disruptive ­synergies.

AIS must be distinguished from unmanned (or uninhabited) ground, underwater or aerial “platforms”, on which they may be mounted. Unmanned “platforms” are not autonomous intelligent systems, if they are controlled remotely by humans or by other machines, or if they lack some of the qualities and aspects of intelligence or autonomy, described above.

The applications of AIS, as well as the enabling technologies contributing to them and derived from them, are of great interest in the defence and national security contexts, in the mid- to long- term (say, from 5 to 30 years hence). Within this time frame, with the accelerated pace of technological change, we can expect to see a progressively more complex set of AIS in operation: (a) independently-acting sensing; (b) pattern recognition and adaptive learning; (c) single, independently-operating robotic platform, incorporating the capabilities of (a) and (b); (d) many interacting, sharing, cooperating and collaborating robotic platforms; (e) human brain interacting directly and collaboratively with computer systems mounted on robotic platforms. While a number of these systems are in the very early stages of development and prototyping, all of these systems are expected to be in place by about 2025.

Controlled forms of “intelligent” devices include those under the direct (wireless or satellite-enabled) manipulation of an operator (often referred to as “man-in-the-loop”), as in robotic arms (such as the CANADARM) or in the lunar and planetary explorers. At the other extreme of the intelligence spectrum will be those systems which are truly “intelligent”, that is those that can operate as part of multi-component systems, that can communicate, interact, learn and collaborate with other “members” of the same system, or with other distinct but compatible systems, and that are interoperable with, as opposed to controlled by, human systems.

In the spirit of ‘letting the robots do the dangerous work”, AIS coupled to robotic systems can assist the human (through automation or where heavy lifting is involved) or replace the human (where dirty, dangerous or dull, repetitive tasks are involved). They may also complement the human (through multi-spectral sensing, information processing and task reporting).

Unmanned vehicles can be constructed far less expensively than manned vehicles, and in many cases are seen as cost-effective alternatives to manned vehicles, especially if they are recoverable and reusable. Some vehicles can be constructed so cheaply (and for limited and highly-specific missions) that they may be considered to be non-recoverable or expendable. For many missions, the unmanned vehicle will provide capabilities far superior to those of manned vehicles.

Unmanned vehicles (or drones) in their most basis AIS forms can be used for passive (non-weaponized) intelligence, surveillance and reconnaissance (ISR) missions, relaying data and images back to their home base or local control stations, via satellite or line-of-sight wireless transmission. In the more advanced forms, and especially in the defence context, the unmanned vehicle may take the form of both the passive (surveillance) and the active (weaponized) modes.

Micro-robotic “swarms” (or “robo-flies”) of low-flying UAVs could be used in the future military battlespace, for surveillance (or more broadly for ISR functions) in confined or “indoor” spaces, or for urban warfare, where high-flying UAVs or satellite-based sensors may not be effective.

In summary, major advances in AIS are anticipated, built on advances in numerous other current and emerging technological areas. When coupled to moving, unmanned vehicles or platforms, AIS are expected to become indispensable components of the defence and national security environments of the future.

Dr. Harold Stocker, Defence Scientist, Defence R&D Canada
Dr. Ingar Moen, Director, Science & Technology Policy, Defence R&D Canada
© FrontLine Defence 2004