Data-Driven Biped Control Device and Method

Data-Driven Biped Control Technology Using Real-Time Pose Modulation and Reference Motion Synchronization

This technology is a data-driven biped control device and method that maintains bipedal balance by modulating reference pose data and editing trajectories based on real-time feedback of current pose information.

Previously, bipeds faced issues with losing balance and falling due to environmental changes or external physical forces, as well as measurement errors that occurred when tracking motion data.

This technology proposes a method that modulates target poses in real-time via a balance maintenance module and adds or deletes frames from reference motions via a synchronization module. By correcting discrepancies between the current pose and reference data in real-time, it enables stable walking even under external force. It can be applied to humanoid robots, walking robots, and robot motion production, accelerating the commercialization of bipedal robots by achieving stable walking that resists falling even when subjected to external forces.

‍

‍

Key Features:

• Configuration utilizing current pose information corresponding to the first time point of the bipedal feedback
• Balance maintenance module that generates target pose information by modulating reference pose information corresponding to a second time point
• Configuration that edits at least a portion of multiple reference pose streams using current pose information
• Synchronization module that synchronizes reference pose information with current pose information by editing a portion of the reference pose stream