Human Machine Interaction and Communication in Cooperative Actions - Bioinspired Approaches for Human-Centric Technologies - page 246

widely discussed in Chaps. 5 and 6 for the Coman and ICub platforms. In particular,
the importance of adjustable compliance to deal with unpredictable tasks has been
the focus of several recent behavioral and robotic studies. Already in early exper-
iments, it has been shown (Shadmehr and Mussa-Ivaldi 1994 ) that humans learn
and adapt internal dynamical models of their own arm in interaction with the
environment. Such internal models appear to be crucial in predicting how muscle
activations produce hand movements and therefore may play an essential predictive
role in movement planning. However, Burdet and co-workers ( 2001 ) have shown
that when prediction is not a viable strategy, humans rely on arm compliance
regulation (by means of muscle co-activation) to cope with the unpredictability
that naturally arises from feedback delays when performing arm-reaching move-
ments in unstable environments. Basic research and robotics technology appear
ready to extend such insights from single limb movements to whole-body interac-
tion and the validation of these models appears feasible.
As mentioned in the introduction of this chapter, joint action requires shared
representations and the transmission of information. This also holds true in the case
of joint action involving physical interaction. While scenarios involving physical
interaction do not exclude explicit planning and the uses of verbal instruction and
miscellaneous visual cues for coordination, physical interaction provides an addi-
tional communication channel between the partners. Unfortunately, little is known
about how contact and interaction forces are used to transmit information about the
goal of the action or the intention and state of each partner (review in Reed 2012 ).
Most of the research on physical interaction has focused on haptic dyads, i.e.,
scenarios where two persons interact physically to achieve some goal. Haptic dyads
are very common in human activities, like physical collaboration in handling bulky
objects (Reed and Peshkin 2008 ; Van der Wel et al. 2011 ) or in performing arts, like
dancing (Gentry and Murray-Smith 2003 ). Interestingly, it has been found that
dyads produce much more overlapping forces than individuals, especially for tasks
with higher coordination requirements, thus suggesting that dyads use larger forces
in the joint action to generate a haptic information channel.
A limited number of studies have analyzed human-robot haptic dyads in coop-
erative tasks. For example, Corteville et al. ( 2007 ) investigated a human-inspired
robot assistant for fast point-to-point movements: the robot scaled the offered level
of assistance in order to give the operator the opportunity to gradually learn how to
interact with the system. The results of the study showed a bidirectional, synergic
influence: while the robot was programmed to adapt to the human motion, the
operator also adapted to the offered assistance, inducing a highly natural type of
interaction. In a shared virtual object manipulation task, performance-related
energy exchange in haptic dyadic interaction has been analyzed, and the results
indicate that the interacting partners benefit from role distribution which can be
associated with different energy flows (Feth et al. 2009 ). On the other hand, in
physical collaborative tasks, it has been found that it may be beneficial to switch
continuously between two distinct extreme behaviors (leader and follower), thus
creating an implicit bilateral coupling within the dyad (Evrard and Kheddar 2009 ).
In a similar vein, it has been found (Oguz et al. 2010 ) that in order to facilitate the