Creative and Adaptive Cooperation between Diverse Autonomous Robots

Our aim is to significantly advance the understanding of how diverse autonomous robots may cooperate in intelligent ways in ad hoc peer encounters. Imagine the following scenario: A delivery robot with a heavy package, e.g., a tool rented online, comes to a construction site previously unknown to it. It has a goal: deliver the tool to a specific location. There are also a number of other robots with different capabilities executing their own tasks, such as cleaning the place, who know their cleaning areas very well. As their cleaning may leave time to other tasks as well, they can help the delivery robot. Such creative cooperation of autonomous robots requires the robots to a) become aware of each other skills, objectives, and knowledge; b) be able to define their joint problem: delivering the package; and to c) together form and execute good enough plans for solving the joint problem.

The project develops novel software architecture design methods for cooperative robotics drawing from artificial intelligence (AI), emphasizing a subfield of AI called computational creativity, ultimately producing emergent collective behavior which is novel and beneficial for all participants. All these three elements must be brought together to produce cross-cutting solutions to the challenges which autonomous, embodied systems pose when deployed in changing environments.

The nucleus of this research proposal is to bring rigorous software engineering approaches for defining a solid basis for the implementation of diverse autonomous robots cooperating creatively. That is, we study what kind of peer models are useful for creative cooperation and how they can be incorporated to the software architecture. To gain multifaceted insight into the peer models and architectural solutions, we will use three advancing stages of development and evaluation (an abstract “block world” simulator, a 3D simulator and physical robots operating in real-world scenarios), identifying challenges and opportunities on each stage. The novel insight is utilized as a seamless toolchain for refining creatively cooperative robots from simulations to the real world.