Walking Human Trajectory Models and Their Application to Humanoid Robot Locomotion

Abstract

In order to fluidly perform complex tasks in collaboration with a human being, such as table handling, a humanoid robot has to recognize and adapt to human movements. To achieve such goals, a realistic model of the human locomotion that is computable on a robot is needed. In this paper, we focus on making a humanoid robot follow a human-like locomotion path. We mainly present two models of human walking which lead to compute an average trajectory of the body center of mass from which a twist in the 2D plane can be deduced. Then the velocities generated by both models are used by a walking pattern generator to drive a real TALOS robot [1]. To determine which of these models is the most realistic for a humanoid robot, we measure human walking paths with motion capture and compare them to the computed trajectories.

Inverse optimal control to model human trajectories during locomotion

“…Thus, this model provides a close approximation of the average human CoM trajectory and pelvis orientation during locomotion. Moreover, its computational time allows real-time application on a humanoid robot which was not the case in Maroger et al (2020b). However, it is still too slow to fluidly assist a human due to the present limits of humanoid robot.…”

“…Thus, holonomic locomotion models allow more degrees of freedom in the human locomotion, like the one described in Mombaur et al (2010), and should better describe human behaviour. Moreover, in Maroger et al (2020b) and Maroger et al (2020a), we already showed that an OC model adapted from Mombaur et al (2010) better fits human trajectories than a clothoid based model according to a metrics which evaluates the distance between generated trajectories and measurements on human subjects. Both models aimed to generate smooth CoM trajectories.…”

“…This dynamics does not take into account the medio-lateral oscillations due to the steps. It only allows to generate smooth CoM trajectories (Maroger et al 2020b).…”

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

“…C. Prediction process simulation Thus, the described prediction process needs a human trajectory as an input. In this paper, we used trajectories of 10 healthy subjects walking from 40 starting positions to one goal position in front of a table collected as part of a study of human walking trajectories [16] [17].…”

“…Thus, models of human walking trajectories using parametric curves called clothoids [12] or OC models [13] [14] [15] have been designed. In [16], the authors assessed those models according to a metrics evaluating the linear and angular distances between human trajectories and generated trajectories. The most human-like model was the one introduced in [15] which approximates the human by a holonomic system instead of a non-holonomic one.…”