IEEE Consumer Electronics Magazine - April 2017 - 90

For distant pointing, the proper design of mapping of interaction space can help improve the usability, including pointing
accuracy [1]. There are two main aspects involved in the interaction space of human-computer interaction (HCI): control space
and visual or display space [8]. Control space is the available
physical space for users to operate, which is constrained by a
user's physical limit and the interaction system setup. The visual
space is the visual representation of the environment perceived by
a user, which is constrained by the field of view of the display.
Mapping of the interaction space in HCI involves the spatial relation between the visual space and user's control space and how
the virtual tool responds to the movement of the input device.
In freehand pointing, interaction can occur anywhere within
the field angle of the camera instead of the area near the input
device, so the interaction space is relatively open, and the relative position between the user and the visual user interface (UI)
often changes during the interaction, which makes it difficult
to design proper mapping for the interaction space. In addition,
the features of interaction based on gesture, such as motor
demand and command input, are greatly different from those
of device-based interaction, which requires further analysis
when designing the mapping of freehand pointing. To provide
suggestions for designing natural mapping of interaction space
in freehand pointing, we will discuss some issues about the
mapping of freehand pointing.

Human pHysical limit in ergonomics
and biomecHanics
Nielsen et al. [9] point out that a user's physical limit is the
significant feature of gesture interaction as a user controls the terminal through using his or her own body in gesture interaction. A
human's available control space is determined by his or her physical limit in ergonomics and biomechanics. If one uses only the
arm to the exclusion of the trunk and legs, then the hand can cover
a maximum movement distance of about 0.5 m [10]. Imagining
the situation where a user is seated in a couch controlling the TV
set 3 m away, using freehand pointing, it is very inconvenient for a
user to move to another location for acquiring a target just because
the target is out of the range of user's control space in the former
location. Proper mapping should enable a user to cover the whole
visual space within his or her control space.
As freehand pointing usually involves upper-limb movement in a three-dimensional (3-D) control space, the fatigue
problem is induced by large amplitude movement in midair
[11]. On the one hand, there is no proper support for the
weight of the arm. The weight of the total arm and the hand is
about 5-6% of body weight [12]. This becomes a load over
time and may rapidly increase a user's fatigue. On the other
hand, 3-D arm movements also involve more shoulder movements, usually scapular and humeral movements [13]. As a
result, 3-D movement requires higher force than twodimensional (2-D) movements.
For the purpose of comfort, postures and motions that are
not natural for a user should be avoided as much as possible.
Nielsen et al. [9] propose several principles in ergonomics when
designing gesture vocabulary, including avoiding outer posi90 IEEE ConsumEr ElECtronICs magazInE

APRIL 2017

tions, relaxing muscles, and avoiding staying in static position,
among others [14]-[17]. According to the study of Casiez et al.
[18], a human's pointing performance is best when the movement distance is within 30 cm, indicating that a control space
with a range of about 30 cm is suitable for pointing interaction.
All of these principles can provide important guidance to set
proper control space for gesture interaction.

Human control precision
of freeHand pointing
Human control precision refers to the smallest target size that a
person can acquire, using the device, with an ordinary amount
of effort [19]. Human control precision reflects the capability
of a user to control the object in the information world using a
certain interaction method. In a sense, human control precision
determines the information capacity of a user's control space.
As mentioned previously, a user's available control space is
about 0.5 m if one uses only the arm to the exclusion of the trunk
and legs [10]. Considering that a user's control precision is 1 mm,
then a user's information capacity is 17.9 b ^2 log 2 ^500/1hh in a
2-D control space and 26.9 b ^3 log 2 ^500/1hh in a 3-D control
space. If the control space is linearly mapped to the visual space,
the information capacity of the visual space should be no more
than those values for the purpose of enabling that user to operate
each object in the visual space with ordinary effort.
One application of control precision is that designers can set
proper sizes of the small interactive elements of the visual UI
and make good use of the capacity of the display. For example,
the interactive elements in the UI of a smartphone are usually
about the size of a human fingertip for the purpose that a user
can operate the element easily and directly using his or her finger. Besides its utilization in visual UI designing, human control
precision is also very useful for the control/dislay (CD) gain
techniques. A CD gain-oriented technique is a software mechanism to improve pointing precision. A user can achieve an arbitrarily small target using CD gain low enough, at the cost of
large movement. With higher control precision, a user can
acquire a smaller target using the same CD gain. As a result,
proper CD gain values can be set according to a user's control
precision to avoid unnecessary large hand movement.
It is known that low precision is one of the main drawbacks of
freehand pointing [7]. One important factor is that hand movement in 3-D space is less stable than in 2-D space, as there is no
proper support for the hand. Another factor is that the device resolution of freehand pointing techniques is usually not ideal. Taking
Kinect as an example, hand movement shorter than 1 cm cannot
be detected by the depth camera, while the smallest human workable movement is about 1 mm [10]. Moreover, freehand pointing
does not provide tactile feedback because it lacks buttons. When
a mode-switching gesture is introduced, the pointing precision is
affected. Knowing the precision limit of a user using freehand
pointing is important for the application of such interaction methods. However, there is no clear answer for what the human control precision is in freehand pointing.
Berard et al. [19], [20] propose a method to measure human
control precision using a mouse and two other kinds of 3-D input

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - April 2017