NAO Human Motion Imitation

Sun, 01 Jan 2017 00:00:00 +0000

Overview

This project developed an ankle-based balance strategy that enables a NAO humanoid robot to imitate human motion while preserving stability. The work connects whole-body imitation with balance control so that a humanoid can reproduce a demonstrator’s movement without losing its support constraints.

Technical Approach

The controller uses an inverted-pendulum model based on the robot’s computed center of mass, calculates support polygons for double- and single-support phases, and treats the ground projection of the center of mass as a balance criterion. Ankle-joint corrections are produced with a PID controller whose initial coefficients were estimated with the Ziegler-Nichols method and then tuned for imitation behavior.

Contribution

Simulation and practical tests on a NAO H-25 V4 showed that the approach improves balance during soft-real-time whole-body and quasi-static imitation. The project demonstrates a control-oriented route to more stable, expressive humanoid imitation for human-robot interaction research.

Motion Imitation | Pourya Shahverdi

NAO Human Motion Imitation

Overview

Technical Approach

Contribution

Project Links