Ontological design is a way of defining, or constructing, a world of possibilities. Robotics is often taught in a very rationalistic and didactic fashion, focusing on the kinematics of robot motion. But this approach rarely scales well, as fine theories rarely “survive contact with reality.” Ontological design is fundamentally inactive, where cognition arises through a dynamic interaction between an individual and their environment. With the ontological design of robots, the focus is on learning robotic design through practice and serendipity.
We explore a system for learning about robots controlled, not through conventional programming, but using simple neural circuits. The inspiration for this is Braitenberg’s `Vehicles’, which provide us with a cybernetic approach to modelling brain function. To make this a fully shared experience, the ‘coding’ of the robot is a tangible activity. Neural circuits are created using code cards spread out on the floor or on a table. An augmented reality (AR) application running on an ordinary smartphone captures the circuit and transfers it to the robot for execution.
We explore a sequence of robot designs that can be built using this language and consider ontological learning at each stage.
Keywords: robotics, ontological design, tangible programming