Physically and Computationally Intelligent Behaviour in Robots

Robots have been entering more and more in our daily lives, with the aim to perform tasks in unstructured, dynamic and complex real-world environments. They need to be intelligent to achieve such goals autonomously, safely and robustly. Here, ‘intelligence’ can be functionally defined as their ability to perceive (sense, interpret), control (decide, plan, predict, regulate), act (move, change, affect, coordinate) and learn (adapt, evolve, acquire experience, infer) continuously and automatically. Intelligence of robots and biological organisms, such as animals and plants, is not only enabled by their computational intelligence (CI) in their brain (and spinal cord), but also by their physical intelligence (PI) encoded in their body. Therefore, it is essential to advance both PI and CI of robots to operate autonomously in real-world environments. Here, PI can be defined as encoding intelligence (perception, action and learning) physically in the robot body. Typically, PI is more specialized and relatively simple while CI is more general purpose and complex.

At centimeter and meter length scales, robots try to use CI dominantly while increasing PI in their body is crucial to minimize the load of, simplify or complement their CI. At millimeter length scales with limited on-board computation and powering capabilities, PI becomes as important as CI. At micrometer length scales, current autonomous microrobots do not have on-board computation and powering capabilities so that PI is their only option. Similar behavior in different length scales can be observed in animals while plants only have PI at all length scales. Also, in very special harsh environments, where extreme conditions can hinder the operation of electronic devices, mechanical PI would be the only option for robots at all length scales.

This conference will discuss the cutting-edge and new scientific and engineering methods and directions to generate intelligent behavior in robots using both physical and computational intelligence methods. The huge advances in artificial intelligence (AI) and machine learning fields will be discussed in view of their contribution to CI for robotics. Some PI and CI methods would be inspired by nature to understand and adapt the intelligent behavior of biological organisms to robotic systems. Also, intelligent behavior of robots and organisms will be investigated at different length scales and effect of length scale on their intelligent behavior will be discussed. We will discuss not only intelligent behavior of single robots and organisms, but also their teams and swarms through coupled physical and computational interactions to create collective, self-organized and other multi-agent intelligent behavior. This highly interdisciplinary research will be represented by speakers not only from the robotics field, but also from biology, physics, AI, machine learning, mechanics, materials science and other various fields.