NC重磅，浙江大学机械工程学院Grasp Lab赋予软体夹爪“类人感知”：大变形下的全覆盖多模态触觉突破

Core Insights - The article discusses the innovative development of FlexiRay, a soft robotic gripper that integrates optical systems, flexible structures, and AI perception to replicate human-like tactile modalities, achieving over 87% effective perception coverage even under significant deformation [3][5][14]. Group 1: Structural and Optical Innovations - FlexiRay's design overcomes traditional limitations by utilizing a segmented multi-mirror array to enhance visual perception during large deformations, transforming visual occlusion into a design advantage [5]. - The system achieves continuous monitoring of contact areas through a combination of direct and reflected light paths, maintaining an effective perception coverage rate of 87.2% even in extreme deformations [5][7]. Group 2: Multi-Modal Sensory Fusion - FlexiRay employs a unique hardware design combined with deep learning algorithms to replicate five key tactile sensations: force, contact position, texture, temperature, and proprioception, using only a single built-in camera [7]. - It achieves a normal force estimation accuracy of 0.17 N and a contact positioning accuracy of 0.96 mm, with a spatial deformation localization error of just 0.24 mm, demonstrating a "hand-eye" integration [7]. Group 3: Interaction and Operation Planning - The gripper's design allows for safe interaction and robust performance in unstructured environments, enabling it to handle fragile objects like chips and eggs without damage [8]. - In complex tasks, such as tea brewing demonstrations, FlexiRay provides comprehensive feedback on force, position, deformation, and temperature, showcasing its potential in remote medical applications and hazardous situations [10]. Group 4: Core Value and Future Outlook - FlexiRay represents a paradigm shift in soft robotic perception systems, maintaining safety and adaptability while enhancing high-precision, multi-modal sensing capabilities, indicating strong commercial potential across various applications [14]. - The ongoing research aims to explore the integration of variable stiffness structures and advanced sensing materials to further enhance robotic tactile intelligence [14].