Collect useful sentences in Papers (7)
These solutions are however unfeasible or
unreliable in many real-world situations, for example when
the robots are operating in unstructured indoor environments or under the surface of the water. A much preferable
approach is to preserve the system autonomy by letting
the robots perform the localization on their own.
In this
case, robot-to-robot measurements are collected and broadcast to the group to be used in a cooperative localization
scheme.
Clearly, this approach requires that each robot is
equipped with a sensor that provides some form of relative measurement (e.g. relative distance or bearing) as well
as a communication module for transmitting and receiving
information.
All of these
works show that the agents’ ability of sensing each other
can improve the localization of the entire system.
To the best of the authors’ knowledge, in all previous
works on network, cooperative, or mutual localization it is
assumed that relative measurements include the identity of
the robots, i.e. that the sensor ‘knows’ to which robot each
measurement refers to. This ubiquitous assumption requires
in practice the identification of a distinctive feature for each
robot, an ability that is not embedded in simple sensors
(e.g. cameras, range finders, RGB-D scanners) and typically calls for sophisticated classification algorithms based
on some form of robot tagging.
the process may be prone
to failures and false positives whenever the conditions of
the surrounding environment are not ideal (e.g. visual tagging may be dramatically affected by slight changes in the
lighting conditions).
Second, it is difficult to discriminate
robots in large groups using features such as shape or color,
i.e. most kinds of tagging do not scale well with the number
of robots.
In this paper we develop to a mature stage the latter approach.
Reference:
Franchi, A., Oriolo, G., & Stegagno, P. (2013). Mutual localization in multi-robot systems using anonymous relative measurements. The International Journal of Robotics Research, 32(11), 1302-1322.
unreliable in many real-world situations, for example when
the robots are operating in unstructured indoor environments or under the surface of the water. A much preferable
approach is to preserve the system autonomy by letting
the robots perform the localization on their own.
In this
case, robot-to-robot measurements are collected and broadcast to the group to be used in a cooperative localization
scheme.
Clearly, this approach requires that each robot is
equipped with a sensor that provides some form of relative measurement (e.g. relative distance or bearing) as well
as a communication module for transmitting and receiving
information.
All of these
works show that the agents’ ability of sensing each other
can improve the localization of the entire system.
To the best of the authors’ knowledge, in all previous
works on network, cooperative, or mutual localization it is
assumed that relative measurements include the identity of
the robots, i.e. that the sensor ‘knows’ to which robot each
measurement refers to. This ubiquitous assumption requires
in practice the identification of a distinctive feature for each
robot, an ability that is not embedded in simple sensors
(e.g. cameras, range finders, RGB-D scanners) and typically calls for sophisticated classification algorithms based
on some form of robot tagging.
the process may be prone
to failures and false positives whenever the conditions of
the surrounding environment are not ideal (e.g. visual tagging may be dramatically affected by slight changes in the
lighting conditions).
Second, it is difficult to discriminate
robots in large groups using features such as shape or color,
i.e. most kinds of tagging do not scale well with the number
of robots.
In this paper we develop to a mature stage the latter approach.
Reference:
Franchi, A., Oriolo, G., & Stegagno, P. (2013). Mutual localization in multi-robot systems using anonymous relative measurements. The International Journal of Robotics Research, 32(11), 1302-1322.
Comments
Post a Comment