Touchless interactive teaching of soft robots through flexible bimodal sensory interfaces.

Area of Science:

• Robotics
• Human-Robot Interaction
• Materials Science

Background:

• Soft robots offer unique advantages in dexterity and safety but often require complex programming for skilled tasks.
• Current human-robot teaching methods can be cumbersome and time-consuming, limiting practical applications.

Purpose of the Study:

• To develop a multimodal flexible sensory interface for intuitive, non-programmable teaching of soft robots.
• To enable soft robots to learn and perform skilled locomotion and specific tasks through human guidance.

Main Methods:

• Development of a flexible bimodal smart skin (FBSS) integrating triboelectric nanogenerator and liquid metal sensing for simultaneous tactile and touchless input.
• Implementation of a distance control method for teaching robot movements via bare hand-eye coordination.
• Evaluation of the system's effectiveness in teaching complex 3D motions and task-specific skills to a soft continuum manipulator.

Main Results:

• The FBSS successfully distinguished between tactile and touchless stimuli, allowing real-time robot self-reaction.
• Participants could teach complex motions to the soft robot within minutes using the proposed 'shifting sensors and teaching' method.
• The trained soft robot accurately replicated taught motions at varying speeds and successfully completed tasks like maze navigation and object grasping.

Conclusions:

• The developed flexible multimodal sensory interface provides a user-friendly and efficient method for teaching soft robots.
• This approach significantly simplifies human-robot interaction, potentially expanding the use of soft robots in diverse applications.
• The non-programmable teaching paradigm facilitates broader adoption and utilization of soft robotic systems.