Publikation
Experience-Based Adaptation of Locomotion Behaviors for Kinematically Complex Robots in Unstructured Terrain
Alexander Dettmann; Anna Born; Sebastian Bartsch; Frank Kirchner
In: In Proceedings of the 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2015). IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS-2015), September 28 - October 1, Hamburg, Germany, Pages 4504-4511, IEEE, 2015.
Zusammenfassung
Kinematically complex robots such as legged
robots provide a large degree of mobility and flexibility, but
demand a sophisticated motion control, which has more tunable
parameters than a general planning and decision layer should
take into consideration. A lot of parameterizations exist which
produce locomotion behaviors that fulfill the desired action
but with varying performance, e.g., stability or efficiency. In
addition, the performance of a locomotion behavior at any given
time is highly depending on the current environmental context.
Consequently, a complex mapping is required that closes
the gap between robot-independent actions and robot-specific
control parameters considering the environmental context and
a given prioritization of performance indices.
In the proposed approach, the robot learns from experiences
made during its interaction with the environment. A knowledge
base is created which links locomotion behaviors with
performance features for visited contexts. This behavior library
is utilized by a case-based reasoner to select motion control
parameters for a desired action within the current context.
The paper provides an overview of the control approach, the
algorithms used to determine the current context and the
robot’s performance, as well as a description of the reasoner
which selects appropriate locomotion behaviors. In experiments,
different behavior libraries were automatically built when operators
had to control a walking robot manually through obstacle
courses. Afterwards, the collected experiences and a trajectory
follower were used to traverse an obstacle course autonomously.
The provided experimental evaluation shows the performance
dependency of the autonomous control with respect to different
sizes and qualities of utilized behavior libraries and compares
it to manual control.