A fully stretchable octopus‐inspired soft robotic tentacle empowered by multimodal tactile sensory toward embodied intelligence

Octopus has been regarded as an example of embodied intelligence, due to its entirely soft body and numerous ganglia. Inspired by the biology structure and actuating strategy of octopus tentacles, we present a fully stretchable soft robotic tentacle equipped with bionic suction cups. In addition to suction and grasping abilities, the tentacle also possesses recognition capability, by integrating multimodal carbon nanotube-based sensors with the suction cups, and employing multi-granularity scanning deep cascade forest algorithm. Through different tasks involving various objects, the sensory system demonstrates superior recognition accuracy, excellence in few-shot learning, and great promise for human-robot interaction applications.