Human Trajectory Prediction Model and Its Coupling With a Walking Pattern Generator of a Humanoid Robot

In order to smoothly perform interactions between a humanoid robot and human, knowledge about the human locomotion can be efficiently used. Indeed, in human-robot collaboration, prediction model of behaviour allows act proactively. this letter, an optimal control based predicting Center Mass (CoM) trajectory during gait is presented. A Walking Pattern Generator (WPG) on non-linear predictive is, then, introduced generate CoM footsteps along predicted trajectory. The combination new WPG aims allow proactively walk with instead passively follow him. These models have been tested simulation Gazebo TALOS using measured trajectories. To foot trajectories computed by WPG, real-time whole-body controller This Quadratic Program which solves inverse dynamics at torque level.