Systems, Man & Cybernetics - January 2016 - 17

six (force and torque), yet the additional vibrotactile
feedback requires higher output DoF for each HIP. A
well-defined input and output DoF interface should uniformly handle all of these situations while providing
backward compatibility to existing devices and SDKs.
3) Kinesthetic and tactile feedbacks on HIPs should be
implemented in a uniform way. This is the intrinsic way
humans perceive the world using fingers, and a logical
design should faithfully reflect this nature. Therefore, an
HIP should be composed of a kinesthetic component and
a tactile component. The kinesthetic component should
focus on macroscale interactions, such as transformations and force/torque-driven interactions. The tactile
component should receive information from its kinesthetic peer and focus on microscale vibrotactile feedback. Depending on the size and complexity of the tactile
array, the tactile component's DoF should also be flexible.
4) Intermanual, interfinger, and intersensory interactions
should be clearly defined. Existing SPHs do not have
these concepts. Only when they are combined as multimanual or multifinger MPHs does the concept of intermanual and interfinger interactions become vaguely
defined. The concept of intersensory interaction is even
more rare and has likely only been mentioned in the
design of holistic haptics. These interaction models can
be identified through the logical and hierarchical relationships among the equivalent components on human
hands, as shown in Figure 1.
In intermanual interactions, multiple manual-level
MPHs are loosely coupled, and their relative locations
are not constrained. Intermanual interactions are more
expensive to compute due to the fact that different
manual-level MPHs are more likely to be heterogeneous
and that each needs to be aware of the other's capabilities and may require conversion routines at the preprocessing stage to assist with the communications.
In interfinger interactions, multiple fingers (HIPs)
are strongly coupled and constrained by the physical
structures among themselves. Compared with intermanual interactions, interfinger interactions are less
expensive to compute due to their known logical and
physical restrictions. In addition, interfinger interactions have a higher probability of HIP homogeneity.
In intersensory interactions, each HIP possesses
both kinesthetic and tactile components. The two feedbacks are rendered separately but are also interdependent. An example of such interactions is to concurrently
identify the stiffness and texture of an object using a
single finger, while the texture friction (tactile) will be
correlated to the exerted pressure (kinesthetic).
5) Real-time interaction should be guaranteed on each HIP.
An SPH typically renders its kinesthetic frame at 1 KHz,
which is much faster than most graphical rendering
pipelines. According to [33], tactile devices also need to
be updated between 700 Hz and 1 KHz. Therefore, each
HIP should update its components at the desired speed

Abstract MPH Model That Supports Multimanual,
Multifinger, and Multisensory Interactions

Multimanual Abstraction
Manual-Level
Node

Manual-Level
Node
Intermanual
Communications

Multifinger Abstraction
Finger-Level
Node

Finger-Level
Node
Interfinger
Communications

Multisensory Abstraction on an HIP
Kinesthetic
Sensory
Component

Intersensory
Communications

Tactile Sensory
Component

Tactile Sensory Array
Tactile
Sensory
Node

Tactile
Sensory
Node

Intersensory
Communications

Figure 1. an overview of the proposed abstract mPH

model using a hierarchical representation. Details of
a higher layer are explained as its lower layer.

while the framework is still able to handle the overall
data throughput.
Framework Design Principles
In our analysis from a software design perspective, we created a framework that:
1) Is based on the abstract MPH model discussed in the "Principles of Multipoint Haptics" section, with a set of standardized function call templates fulfilling the principles.
2) Uniformly handles and supports various existing and
future MPHs without imposing mechanical and electronic hypotheses or restrictions to the design and
implementation of hardware, drivers, and SDKs.
Ja nu a r y 2016

IEEE SyStEmS, man, & CybErnEtICS magazInE

Table of Contents for the Digital Edition of Systems, Man & Cybernetics - January 2016