Coordination between digit forces and positions: interactions between anticipatory and feedback control

Abstract

Humans adjust digit forces to compensate for trial-to-trial variability in digit placement during object manipulation, but the underlying control mechanisms remain to be determined. We hypothesized that such digit position/force coordination was achieved by both visually guided feed-forward planning and haptic-based feedback control. The question arises about the time course of the interaction between these two mechanisms. This was tested with a task in which subjects generated torque (± 70 N·mm) on a virtual object to control a cursor moving to target positions to catch a falling ball, using a virtual reality environment and haptic devices. The width of the virtual object was varied between large (L) and small (S). These object widths result in significantly different horizontal digit relative positions and require different digit forces to exert the same task torque. After training, subjects were tested with random sequences of L and S widths with or without visual information about object width. We found that visual cues allowed subjects to plan manipulation forces before contact. In contrast, when visual cues were not available to predict digit positions, subjects implemented a "default" digit force plan that was corrected after digit contact to eventually accomplish the task. The time course of digit forces revealed that force development was delayed in the absence of visual cues. Specifically, the appropriate digit force adjustments were made 250-300 ms after initial object contact. This result supports our hypothesis and further reveals that haptic feedback alone is sufficient to implement digit force-position coordination.

Digit Position and Forces Covary during Anticipatory Control of Whole-Hand Manipulation

“…The authors argued that modulating digit forces in response to digit position minimized M com variability. Digit position and force modulation has been replicated in other two-digit manipulation studies (Zhang et al, 2010 ; Fu and Santello, 2014 ; Marneweck et al, 2015 ). Thus, constraining digit position, like most previous work, prevents fundamental features of dexterous grasp control: (1) modulating digit position to object properties and task demands; and (2) modulating digit forces to compensate for trial-to-trial digit position variability.…”

Stability and predictability in human control of complex objects

“…To date, most studies on object manipulation have focused either on grip forces needed for transporting solid objects or multi-digit grasping of a static object. [6][7][8][9] It is frequently assumed that humans use inverse dynamics and acquire an internal model of their limbs, the object, and the environment. [10][11][12][13][14] For example, Gawthrop and colleagues 15,16 simulated controlling an inverted pendulum as a model for human balancing and showed that a non-predictive controller was unable to explain how humans performed the task.…”

On neuromechanical approaches for the study of biological and robotic grasp and manipulation

“…Briefly, it has been shown that humans are able to modulate manipulative forces in an anticipatory fashion, i.e., between contact and onset of manipulation, according to where the object is grasped [136,137]. This phenomenon, which has been confirmed by several studies [139,138,235,140], ensures attainment of the manipulation goal despite trial-to-trial variability in finger placement that may naturally occur while using the same or different number of fingers ( [136] and [137], respectively). Finger force-to-position modulation is a phenomenon that is very useful for inferring its underlying neural control mechanisms.…”

“…One example is the ability of the hand to swiftly change control strategies when transitioning from finger motion to force application [135]. Another example is humans' ability to modulate finger force distribution shortly after contact and prior to onset of manipulation to account for trial-to-trial variability in finger placement [136,137,138,139,140]. Such problems are extremely challenging to replicate in a robotic system.…”