Generation of Human-Like Movement from Symbolized Information

Area of Science:

• Robotics
• Control Systems
• Artificial Intelligence

Background:

• Current robotic systems lack human-like behavior generation from symbolized information.
• Distinguishing human motion from machine motion is crucial for advanced robotics.
• Understanding mechanisms for automatic motor commands is key to achieving complex behaviors.

Purpose of the Study:

• To clarify mechanisms generating automatic motor commands for symbolized behavior.
• To develop a controller enabling human-like robotic behavior.
• To assess the potential for diverse robotic actions through novel control strategies.

Main Methods:

• Designed a controller incorporating tacit learning for system-environment interactions.
• Implemented a mechanical resonance mode (MRM) transfer method for control signals.
• Validated through simulations and experiments on balance control and bipedal walking.

Main Results:

• Standing balance control was achieved with an inverted pendulum by adjusting MRM-space signals.
• Humanoid robots demonstrated diverse bipedal walking patterns via MRM-space adjustments.
• Tacit learning effectively managed system-environment interactions for adaptive control.

Conclusions:

• Transferring control signals to an appropriate space (MRM-space) simplifies environmental signal complexity.
• This approach is key to achieving diverse and human-like robotic behaviors.
• The developed controller shows promise for advancing robotic adaptability and functionality.