Existing research on multi-agent autonomous systems are unable to solve an important class of real problems. At the root of this problem is the assumption of pervasive, predictable, reliable, and free communications.

• • pervasive: agents can communicate at any time. (includes robustness against failure)

  • predictable: agents can communicate at predictable times.

  • reliable: agents’ communications are delivered without error.

  • free: agents’ communications have no cost to the agent.

These assumptions are the basis of the state of the art in many existing approaches to distributed planning and plan execution. Even if the environment is uncertain, agents can take advantage of the assumption of pervasive or predictable communications to exchange information about what they have learned and what they plan to do next. Reliable and free communications ensure coordination does not come at the expense of goal achievement.

Such approaches may not be viable if a variety of challenges are present in the mission environment: if agents do not know if communication is possible, if changes in the world require unplanned communications, if those communications take time, energy, resources, if an agents’ ability to communicate changes, and in general, if communication comes at the cost of resources needed to achieve other goals. Such challenges are central to autonomy in many aerospace applications; examples include remote poorly characterized environments such as planetary (including Lunar) surfaces, or around Icy Moons such as Europa or Enceladus. Closer to home, examples include constellations of Low Earth Orbit satellites with unreliable communications and low-power computing, swarms of high-altitude Uncrewed Air Vehicles (UAVs), lower altitude UAVs in remote environments with poor communication infrastructure, or densely packed urban environments with a high degree of uncertainty. Finally, it has long been recognized that any form of communications is difficult underwater, and there is a severe trade off between range and achievable data rates, primarily due to the transmission medium and its variability with time and location. These problems are especially daunting for multi-agent systems designed to coordinate; uncertainty in the ability to communicate profoundly under-cuts the utility of such systems, and the combination of remote environments, poorly characterized environments, and fragility of many aerospace systems makes this an important problem to address. When agents are responsible for communication relay duties, as well as the ability to perform other tasks as part of a multi-agent system, fault management also takes on a new importance.

This symposium aims to identify the challenges and bridge the gaps between theoretical frameworks for multi-agent autonomous system and the challenges imposed by deployment in these environments, in which the usual assumptions do not apply. Our goal is to help identify research avenues that can move the AI community beyond theoretical results for simple domains. We actively solicit research that pushes the boundaries of current multi-agent systems theory, as well as applications that require technology that may not exist today. Our focus is on aerospace applications that require multi-agent systems to operate in resource-constrained and unstructured environments, and we invite challenge problems and modeling and simulation frameworks to enable the research community to test their ideas on such problems. Many of these aerospace systems have a significant human element, especially when we consider either scientific UAV multi-agent systems or UAVs operating in communication-poor urban environments, and new methods for managing human-multi-machine teams are also solicited.