Prediction of Intention during Interaction with iCub with Probabilistic Movement Primitives

Abstract

This article describes our open-source software for predicting the intention of a user physically interacting with the humanoid robot iCub. Our goal is to allow the robot to infer the intention of the human partner during collaboration, by predicting the future intended trajectory: this capability is critical to design anticipatory behaviors that are crucial in human-robot collaborative scenarios, such as in co-manipulation, cooperative assembly, or transportation. We propose an approach to endow the iCub with basic capabilities of intention recognition, based on Probabilistic Movement Primitives (ProMPs), a versatile method for representing, generalizing, and reproducing complex motor skills. The robot learns a set of motion primitives from several demonstrations, provided by the human via physical interaction. During training, we model the collaborative scenario using human demonstrations. During the reproduction of the collaborative task, we use the acquired knowledge to recognize the intention of the human partner. Using a few early observations of the state of the robot, we can not only infer the intention of the partner but also complete the movement, even if the user breaks the physical interaction with the robot. We evaluate our approach in simulation and on the real iCub. In simulation, the iCub is driven by the user using the Geomagic Touch haptic device. In the real robot experiment, we directly interact with the iCub by grabbing and manually guiding the robot's arm. We realize two experiments on the real robot: one with simple reaching trajectories, and one inspired by collaborative object sorting. The software implementing our approach is open source and available on the GitHub platform. In addition, we provide tutorials and videos.

Multi-modal Intention Prediction with Probabilistic Movement Primitives

“…The human physically interacts with the robot's arm to start an action and/or uses his directional gaze to guide the robot. We build on our previous work [7], where elementary actions are represented by Probabilistic Movement Primitives that enable prediction of goals from early observations. During physical guidance, the robot uses the haptic information to recognize the current action, then it is able to accurately predict the goal, the future intended trajectory and its duration.…”

“…This section presents the ProMP method used to learn the motion primitives and to predict the trajectory of the ProMP given one modality. See [7] for further information.…”

“…The paragraphs below provide a brief review of research literature on intention and gaze prediction. For the state of the art on movement primitives and inference during pHRI, we refer to [7].…”

“…5. From [7], we retain two methods to compute the time modulation: "maximum likelihood" (ML) and "model ", where the latter consists on estimating the trajectory duration according to the global partner's head orientation variation: "δn o = A(no) − A(1)".…”

“…In our previous work [7], we addressed the problem of predicting the future intended trajectory during a physical human-robot interaction when the human partner moves the robot's arm to start a movement. We proposed to use Probabilistic Movement Primitives (ProMPs [21]) to learn the movement primitives from a set of demonstrations and to compute the intended trajectory given early observations of the action, guided by the human partner.…”