Artificial Intelligence based Soft Robotics: Materials and intelligent control systems

Abstract

The combination of Artificial Intelligence (AI) and soft robotics has established a new dimension of robotic systems design, functionality and autonomy. In contrast to the rigid-bodied robots, the soft robots use compliant and deformable materials that resembles the adaptability and strength of the biological organisms. The combination of smart algorithms and soft adaptive materials has made it possible to have robots that engage with humans and unstructured environments in a safe and efficient way. With the implementation of deep learning, reinforcement learning and bioinspired control strategies, AI-enhanced soft robotics presents the ability to learn in the real-time, adapt to the environment, and reconfigure itself. Electroactive polymers, liquid crystal elastomers, and shape-memory alloys are examples of intelligent materials that perform the roles of actuators and sensors, and provide continuously operating feed-back to control systems run by artificial intelligence. These adaptive systems allow the ability to manipulate, navigate and self-heal with very accuracy which has never been possible with the traditional robotic architectures. Moreover, neural networks can be used to model nonlinear behaviors that are complex in nature and occur in soft materials, whereas reinforcement learning can be used to make adaptive decisions in dynamic environments. This paper discusses the interaction of AI algorithms and intelligent materials with particular emphasis on their co-design, which leads to greater mechanical intelligence and autonomy. It also discusses the issues of deformable structure modeling, energy saving, and stability of control, which are key to the development of soft robotics in the health care, exploration, and manufacturing fields. The paper brings out the transformative nature where the AI may be embraced to guide the creation of soft robots in perceiving, learning, and evolving and thus bridging the gaps between the artificial and the biological intelligence.