Imitation Learning Based on Visuo-Somatic Mapping

Abstract

Abstract. Imitation learning is a powerful approach to humanoid behavior generation, however, the most existing methods assume the availability of the information on the internal state of a demonstrator such as joint angles, while humans usually cannot directly access to imitate the observed behavior. This paper presents a method of imitation learning based on visuosomatic mapping from observing the demonstrator's posture to reminding the self posture via mapping from the self motion observation to the self posture for both motion understanding and generation. First, various kinds of posture data of the observer are mapped onto posture space by self organizing mapping (hereafter, SOM), and the trajectories in the posture space are mapped onto a motion segment space by SOM again for data reduction. Second, optical flows caused by the demonstrator's motions or the self motions are mapped onto a flow segment space where parameterized flow data are connected with the corresponding motion segments in the motion segment space. The connection with the self motion is straightforward, and is easily acquired by Hebbian Learning. Then, the connection with the demonstrator's motion is automatic based on the learned connection. Finally, the visuo-somatic mapping is completed when the posture space (the observer: self) and image space (the demonstrator: other) are connected, which means observing the demonstrator's posture associcates the self posture. Experimental results with human motion data are shown and the discussion is given with future issues.

Correspondence Mapping Induced State and Action Metrics for Robotic Imitation

“…How such correspondences can be built is one of the hardest problems of imitation learning and comprises an important aspect of the what to imitate problem. For related work on visuo-somatic mapping, see [7] and [13].…”

“…For example, consider correspondences between a human body as model to a dolphinlike imitator 7 : a dolphin, using its mouth corresponding to either human hand (grasping an object) or using its tail to toss a ball back to a human that used both arms, comprise realworld examples of many-to-one mappings. These two examples also illustrate that the correspondence needs not be static-the human hands are mapped to different dolphin body parts in 7 Different mappings do appear to be employed by real-life dolphins in imitating humans [17]. each case-but can be adapted depending on the context and the tasks involved.…”

Abstraction Levels for Robotic Imitation: Overview and Computational Approaches

“…Examples of other methods posing imitation within this visuo-somatic perspective include [7,53]. An interesting approach is proposed in [18] where facial expression are learned by interacting with a person and creating a resonance of expressions.…”

Intrinsically Motivated Exploration for Developmental and Active Sensorimotor Learning

“…Each region is responsible for monitoring the evolution of the error rate in the anticipation of consequences of the robot's actions, if the associated contexts are covered by this region. 9 equally spaced between the extrema values of , thus we evaluate . | | splits in total, and the one with minimal , is finally chosen.…”