|
Figure 2.9: A typical CIR architecture
The second aspect is sensor-based programming and operation in industrial robots. De-
spite the fact that sensor-based robotics has been a staple of robotics research in the last
20 years (Christensen et al. 2009, P. 4), it has yet to gain wide-spread acceptance in
production facilities. Reasons ranging from doubtful sensor readings and the robustness of
control algorithms to the lack of RT interfaces on commercial robots are among the most
commonly encountered (Christensen et al. 2009, P. 14)(Georgia Institute of Technology
et al
. 2009, P. 17). Not only are they lacking, but their implementation is also troublesome
considering the closed nature of much of today’s industrial manipulators. They lack
standardized real-time interfaces and open access to internal data in the controllers. In a
technical sense the latter reasons are enough to hinder sensor-based architectures from
becoming more popular. However a holistic view of the issue discloses an overarching
reason, namely that tight integration of sensors in robot architectures in practice is vir-
tually nonexistent. For instance the utilization of sensors in di
fferent capacities during
programming and during operation is rarely investigated. Furthermore, simulating the
sensors o
ff-line is still limited to research platforms such as the Player/Stage/Gazebo
framework (Gerkey et al. 2003) (Collett et al. 2005) or Microsoft Robotics Developer
Studio (Microsoft 2011). The latter discussion also applies for CIR which aren’t much
36
2.6 Discussion
radically di
fferent than industrial robots (see Figure 2.9). As already summarized in Figure
2.8 such architectures additionally possess a communication channel through which all the
robot controllers can synchronize their movement according to the master
/slave principle
referred to in section 2.5.2. Where a dedicated controller
/robot controller or simply one
robot controller generates a clock-cycle to be followed by all the others.
2.6.3 Programming paradigms
Although robot programming has been in the forefront of research and development
for robot manufacturers, it has yet to experience radical changes regarding the applied
programming paradigms. This could be readily traced back to the relation between pro-
gramming, architectures and additionally the human machine interface (HMI). In order to
understand the extent of this work the intrinsic and sometimes ambiguous relation between
robot programming and architectures will be elaborated upon. As already discussed in the
latter section the types of components in a robotic work-cell and how they are arranged
and interconnected directly influences not only the operation capability but also the pro-
gramming capability. MacDonald et al. (2003) emphasized this acute connection by
considering the robot infrastructure and its interconnections as a major component of a
robot programming system. They stated that one of the conceptual components necessary
for a programming system is:
"the underlying infrastructure including designs for architectures that support
and execute robot behavior descriptions, especially in distributed environ-
ments"
The rigidity of robot architectures and their significant influence on programming systems
led to the emergence of two distinct phases that occur during robot operation:
1. First phase: Robot programming is regarded as the process during which a human
operator
/programmer define instructions for the robot in order to execute a certain
task.
2. Second phase: Robot execution is regarded as the process during which those
instructions are executed at runtime without human intervention.
The classical programming paradigm handles the two phases distinctively in a manner so
that the phases do not overlap with each other, i.e. the robot program (instruction list) is
invariable throughout the runtime until further human intervention. Any disturbance arising
during task execution can not be accounted for. Similar to a new task, any disturbance has
to be defined and incorporated in the final instruction list. This classical definition assumes
that the robot cell is incapable of gathering information about its surrounding environment
and thus incapable of taking decisions regarding any dynamic change in runtime. This is,
in turn, bound to the capability of the work-cell to integrate and utilize sensory feedback
to enhance and adapt its behavior which is directly linked to the work-cell’s architecture.
Moreover, other than a few e
fforts from manufacturers such as integrating a space mouse or
37
2 Literature Review
a joystick, no simple HMI have found their way in commercial CIR systems. Furthermore,
graphical user interfaces utilized for programming both single industrial robots and CIR
systems are usually cluttered with functionalities that render them complicated and hence
require extra training.
38
3 Motivation and Objective
Do'stlaringiz bilan baham: |
