Stabilization Strategies for Unstable Dynamics

Abstract

BackgroundWhen humans are faced with an unstable task, two different stabilization mechanisms are possible: a high-stiffness strategy, based on the inherent elastic properties of muscles/tools/manipulated objects, or a low-stiffness strategy, based on an explicit positional feedback mechanism. Specific constraints related to the dynamics of the task and/or the neuromuscular system often force people to adopt one of these two strategies.Methodology/FindingsThis experiment was designed such that subjects could achieve stability using either strategy, with a marked difference in terms of effort and control requirements between the two strategies. The task was to balance a virtual mass in an unstable environment via two elastic linkages that connected the mass to each hand. The dynamics of the mass under the influence of the unstable force field and the forces applied through the linkages were simulated using a bimanual, planar robot. The two linkages were non-linear, with a stiffness that increased with the amount of stretch. The mass could be stabilized by stretching the linkages to achieve a stiffness that was greater than the instability coefficient of the unstable field (high-stiffness), or by balancing the mass with sequences of small force impulses (low-stiffness). The results showed that 62% of the subjects quickly adopted the high-stiffness strategy, with stiffness ellipses that were aligned along the direction of instability. The remaining subjects applied the low-stiffness strategy, with no clear preference for the orientation of the stiffness ellipse.ConclusionsThe choice of a strategy was based on the bimanual coordination of the hands: high-stiffness subjects achieved stability quickly by separating the hands to stretch the linkages, while the low-stiffness subjects kept the hands close together and took longer to achieve stability but with lower effort. We suggest that the existence of multiple solutions leads to different types of skilled behavior in unstable environments.

Stabilization strategies for unstable dynamics

“…As shown by Saha and Morasso (2012) the shape and size of the stiffness ellipse depend on the relative position of the two hands and the virtual mass. The major axis of the ellipse is approximately aligned with the line linking the two hands and the ratio between the major and the minor axis grows with an increasing separation of the two spring terminals.…”

“…A novel approach (Saha and Morasso, 2012;Zenzeri et al, 2014) was conceived with the aim to put the participants in an experimental situation that was as neutral as possible with the respect to the choice of the strategy: in this approach, individuals are trained to use a virtual tool (VUBT: Virtual Underactuated Bimanual Tool), with a tool-tip under the action of a saddle-like force field, and the studies investigated two main questions, namely which strategy do naïve people prefer and how difficult is it to switch from one strategy and the other for expert users?…”

“…It is not an easy task and a suitable training time is required, but all the naïve subjects who participated to the experiments succeeded to do it (Saha and Morasso, 2012). With the device parameter values reported above, participants could either use the stiffness strategy, by stretching apart the two virtual springs an amount sufficient to overcome the critical value of the medio-lateral stiffness; or the intermittent feedback strategy, because the falling time constant is long enough to allow the delayed feedback to be effective.…”

“…In the first set of experiments with naïve people (Saha and Morasso, 2012) it was found that there was not a strong preference for a strategy or the other, in spite of the rather great difference in terms of energetic costs (the average effort in the stiffness strategy is more than twice the effort in the intermittent feedback strategy): about 60% of participants quickly adopted the high-stiffness strategy, with stiffness ellipses that were aligned along the direction of instability, and the remaining participants applied the intermittent feedback strategy, operating with an under-critical stiffness value and no clear preference for the orientation.…”

Human-human physical interaction in the joint control of an underactuated virtual object

“…Secondly, this task forces the subject to finely tune his motor output in response to force perturbations to contrast the instability. We have previously demonstrated that, despite the task being very challenging for a naïve user, it supports the learning of two specific different patterns of bimanual coordination that solve the balancing problem [11], [12]: a high stiffness feedforward strategy and a low stiffness positional feedback strategy. In this work, we analyze how these two patterns accommodate in the case of a dyadic configuration and test whether it allows for performance advantages compared to a bimanual scheme.…”