Touching with torque enables human-level robotic dexterity.

Area of Science:

• Robotics and Artificial Intelligence
• Sensor Technology
• Human-Robot Interaction

Background:

• Forceful manipulation in robotics is hindered by the lack of environmental interaction cues like collisions, balance, and resistance.
• Existing robotic systems struggle with tasks requiring fine motor control and real-time adaptation in unstructured environments.

Purpose of the Study:

• To introduce a novel torque-angle-pressure (TAP) tactile sensor for enhanced robotic manipulation.
• To enable robots to perceive and react to environmental cues for improved performance in complex tasks.
• To advance the capabilities of robots in achieving human-like forceful manipulation and collaboration.

Main Methods:

• Developed a torque-angle-pressure (TAP) tactile sensor utilizing magnetic flux density gradients.
• Achieved bidirectional, ultrasensitive, and high-linearity sensing over a wide range via a single readout channel.
• Integrated the TAP sensor into robotic systems for vision-free object placement, balance beam challenges, and adaptive slicing.

Main Results:

• The TAP sensor demonstrated high linearity (R² = 0.99) and sensitivity (~0.1°, ~0.4 Nm) over a broad range (±241.6 Nm).
• TAP-equipped robots surpassed human performance in vision-free stable object placement and balance beam stacking (2.4s, 81.5% success rate).
• Robots with TAP sensors exhibited adaptive daikon slicing with real-time adjustments, a capability rare in current systems.

Conclusions:

• The TAP tactile sensor significantly advances robotic tactile sensing capabilities.
• This technology enables forceful manipulation in unstructured environments, paving the way for more capable robots.
• The TAP sensor represents a key advancement toward effective and seamless human-robot collaboration.