18.07.2014 Views

PHD Thesis - Institute for Computer Graphics and Vision - Graz ...

PHD Thesis - Institute for Computer Graphics and Vision - Graz ...

PHD Thesis - Institute for Computer Graphics and Vision - Graz ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

1.6. Contribution of this thesis 7<br />

environments. Global localization applies when a robot operates in a known environment, that<br />

is the map has already been built. Global localization is then the computation of the pose of<br />

the robot, consisting of position <strong>and</strong> heading, from the actual sensor reading, i.e. camera image<br />

without using the previous pose in<strong>for</strong>mation. Global localization is necessary quite frequently<br />

when encountering the following situations:<br />

Switching the robot on: After switching on a mobile robot the position is not known. However,<br />

the robot may already possess a complete environment map e.g. from a previous run.<br />

But be<strong>for</strong>e it can start useful operations (like navigation <strong>and</strong> path planning) its position<br />

has to be determined first. Global localization allows to compute the pose of the mobile<br />

robot from the actual camera image.<br />

Kidnapped robot problem: The kidnapped robot problem has been stated by Engelson <strong>and</strong><br />

McDermott [27]. In the kidnapped robot problem a well-localized robot is teleported (or<br />

simply moved with switched-off sensors) to some other location. The problem with this<br />

scenario is, that the robot still believes to be at the location from where it has been<br />

kidnapped. Path planning <strong>and</strong> navigation based on such an assumption will not work.<br />

The environment predicted from the map <strong>and</strong> its assumed position will not match. The<br />

kidnapped robot problem is basically a test <strong>for</strong> the ability of a robot to recover from a<br />

catastrophic localization failure. Global localization is the solution to the kidnapped robot<br />

problem. With the ability of global localization the robot can immediately determine that<br />

it has been moved from outside.<br />

Recover from failure: Recovery from a failure is also possible with global localization. Consider<br />

the case when a mobile robot moves into an area without l<strong>and</strong>marks. Abruptly the<br />

current sensor readings do not contain any l<strong>and</strong>marks. For vision based systems this can<br />

occur easily when the robot is moving into an untextured area, e.g. a part of an room<br />

containing only white walls. In such a case the robot would loose track. The robot would<br />

then move around r<strong>and</strong>omly to get back into an area where l<strong>and</strong>marks can be detected.<br />

However, when he finally enters an area where l<strong>and</strong>marks re-occur his global position got<br />

lost, global localization is necessary. It would be possible to rely on the robots wheel odometry<br />

in the l<strong>and</strong>mark-less area. However, keeping track with odometry only will introduce<br />

too big deviations in pose estimation <strong>and</strong> global localization would still be necessary.<br />

Loop closing: Loop closing is the ability of a mobile robot to recognize previously visited<br />

areas. It is important in the stage of map building. During map building a mobile robot<br />

traverses the environment <strong>and</strong> adds new structure <strong>and</strong> l<strong>and</strong>marks to the world map. If<br />

the robot enters an already mapped area <strong>and</strong> does not recognize this, the same features<br />

will be added twice, usually not on the same position because of small drift errors. Global<br />

localization can be used to notice that the current location has already been visited, that<br />

the loop has been closed. This gives the in<strong>for</strong>mation, that the l<strong>and</strong>marks could already be<br />

in the map <strong>and</strong> an update would be appropriate <strong>and</strong> not a simple insertion.<br />

Homing: The last example <strong>for</strong> a situation where global localization would be required is homing.<br />

Homing is the task that a mobile robot has to go back to some start position. The<br />

start position may <strong>for</strong> instance mark an automatic charging device <strong>and</strong> the robot will go<br />

there to recharge its batteries. Global localization can tell the robot that its target position<br />

is reached.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!