Motion Generation

Motion Generation is concerned with the problem of finding a way to move the robot from an initial to a final position. During this motion the robot has to avoid other objects in its environment. At the same time, certain requirements are imposed to its motion by the specifications of the task, which the robot is assigned to fulfill. Think of a robot carrying a tray with drinks; the robot has to know somehow that the tray should remain horizontal!

Figure 1: Planning in a 2D configuration space. The white space is corresponds to free configuration while the grayed one to obstructed configurations. Motion planners search this space to compute a solution.

© Robotics

Solutions are computed in two stages. Firstly, an algorithm searches for a sequence of intermediate configurations of the robot that solve the task. Afterwards, controllers ensure that the robot moves through these intermediate points. The problem is that the number of possible configurations is too big, making a "dummy" search very slow. And what happens if something changes in the environment or if the model our search was based on, is not accurate enough? A new search is needed, throwing away all the previous computation. This prevents current motion generation algorithms from solving real-world tasks.

Uman_line

We want to find methods for motion generation that are more robust to changes in the environment. To achieve this, we shift the boundary between planning and control. Giving more responsibility to control allows us to leverage the inherent capability of controllers; dealing with dynamic situations. The notion of feedback, which is instrumental in control theory, comes into play. Feedback Motion Planning is a motion generation method that takes into consideration information acquired while the robot performs a task. So, instead of calculating explicitly paths in the configuration space of the robot and let the responsibility of execution to the controllers, we want to calculate sequences of controllers that carry out the task. This tackles the problem of uncertainty, while ensuring a more reactive behaviour towards changes in the environment.

Contact: Arne Sieverling

References:
[2] S. M. LaValle. Planning Algorithms. Cambridge University Press, Cambridge,
U.K., 2006. Available at planning.cs.uiuc.edu.
[3] Motion Planning, 2009. en.wikipedia.org/wiki/Motion_planning.

Funding

© AvH

Alexander von Humboldt professorship - awarded by the Alexander von Humboldt foundation and funded through the Ministry of Education and Research, BMBF,
July 2009 - June 2014  

© EC

© First MM

Flexible Skill Acquisitionen and Intuitive Robot Tasking for Mobile Manipulation in the Real World (First MM) - funded by European Commision, in the program Cognitive Systems and Robotics,
award number FP7-ICT-248258,
February 2010 - July 2013

Publications

Order by: Author  Year  Journal