This master’s thesis optimizes the modelling of the environment of industrial mobile robots.
In the industrial context, robots have to navigate or interact. They plan actions using information
from a world model. This model usually only contains data that is directly used or
that can be generated by the robot.
An object-oriented approach for a universal world model is presented. Objects are saved in
a graph to picture hierarchical and topological relations between objects. The object graph
saves storage and computing power. An object can contain geometric and semantic data. In
addition, a future trajectory can be added, marking a step beyond other works.
Several robots can use the same world model with an update strategy which includes incomplete
synchronization. This leads to reduced access latency and network traffic.
Part of the concept has been implemented in C++. The software includes a server and clients
for reading from a database, for visualization and for integration with the Robot Operating
System (ROS). A test case with a real mobile robot and a dynamic obstacle has showed the
successful storing and communication of future trajectories.
«
This master’s thesis optimizes the modelling of the environment of industrial mobile robots.
In the industrial context, robots have to navigate or interact. They plan actions using information
from a world model. This model usually only contains data that is directly used or
that can be generated by the robot.
An object-oriented approach for a universal world model is presented. Objects are saved in
a graph to picture hierarchical and topological relations between objects. The object graph...
»