Self: Symbolic & Connectionist AI for Embodied Cognition     


Grady Booch photo

Self: Symbolic & Connectionist AI for Embodied Cognition - overview

Self is a platform for embodied cognition, serving to orchestrate sensors that perceive the world, actuators that manipulate or influence the world, actors that react as well as bring agency to the world, and models that give the instantaneous and historical context of the world, of others in the world, and of the system itself. Embodied cognition takes a system that reasons and that learns then literally places it in the physical world, making it manifest in the form factor of a robot, an avatar, a space, or an object/device. Self is primarily focused on the use cases associated with augmented intelligence, that is, systems that expand individual and collaborative human senses and abilities, systems that work in cooperation and conversation with humans but that are also capable of degrees of self-agency and self-understanding. Self may be seen as the middleware or operating system that coordinates individual cognitive components to bring higher-level reasoning and learning to a system as a whole, made present in one or more of particular embodied form factors.