Embodied Perception Interaction, and Cognition for Wearable Robotics: A Survey

Wearable robotics, encompassing exoskeletons and other assistive devices, have emerged as promising tools for constructing and augmenting human capabilities in various domains, including healthcare, rehabilitation, and assistive technologies. This review explores the domains of perception, interaction, and cognition in wearable robotics, with a focus on advancing embodied intelligence in these systems. Perception involves the recognition of human motion intentions through sensory signal acquisition, multimodal fusion, and intent recognition. Interaction encompasses various modes of human-robot interaction, ranging from physical interaction control relying on interaction force to human-in-the-loop control methods reliant on human physiological or biomechanical feedback signals. Cognition extends beyond motor function reconstruction, focusing on the cognitive reconstruction of humans, facilitated by sensory feedback and human-machine interface technologies. Several challenges hinder their seamless integration into wearable robotics, including difficulties in signal processing and integration, robot control, bidirectional information transfer, and interface design. Future research efforts should prioritize addressing these challenges to advance embodied intelligence in wearable robotics, leveraging artificial intelligence algorithms to empower exoskeleton technology, designing ergonomic interfaces for enhanced wearer comfort and safety, developing advanced signal processing techniques for accurate interpretation of human biological signals, and integrating multimodal sensing for comprehensive interaction.