Adaptive Robots in Space: Role of Machine Learning and Soft Robotics

Muhammad Sunny

National University of Singapore

We, as a human race, created robots to do some of the dangerous jobs we are at a risk doing. While we managed to achieve some of these goals, we are nowhere close to achieving systems with abilities to actually perform tasks with human/animal capabilities in diverse surroundings. On another note, biological systems have evolved for billions of years to become what they are today, capable of performing extremely sophisticated tasks to not only survive but thrive in a range of environments. These systems are a construct of a body with different properties (related to material and geometry) and a decision-making entity (may be construed as a brain sometimes, but not necessarily). It is a belief of many researchers around the globe that learning from the properties of these bodies or mimicking primitives of their behaviors may unleash capabilities currently unachievable with the traditional robots. The multi-disciplinary field, researching on related beliefs, is soft robotics. It is, however, acknowledged here that soft robotics is a subclass of robotics we know today. In this talk, I will highlight some of the salient attributes of this class of robotics, my research in this domain, and a role of machine learning in learning from and mimicking biological systems. Finally, I would like to present my perspective/ideas on how I would like to see soft robotics in space applications.