Publication
Start State Selection for Control Policy Learning from Optimal Trajectories
Christoph Zelch; Jan Peters; Oskar von Stryk
In: IEEE International Conference on Robotics and Automation, ICRA 2023, London, UK, May 29 - June 2, 2023. IEEE International Conference on Robotics and Automation (ICRA), Pages 3247-3253, IEEE, 2023.
Abstract
Combination of optimal control methods and ma-
chine learning approaches allows to profit from complementary
benefits of each field in control of robotic systems. Data from
optimal trajectories provides valuable information that can be
used to learn a near-optimal state-dependent control policy. To
obtain high-quality learning data, careful selection of optimal
trajectories, determined by a set of start states, is essential to
achieve a good learning performance.
In this paper, we extend previous work with new comple-
menting strategies to generate start points. These methods
complement the existing approach, as they introduce new
criteria to identify relevant regions in joint space that need
coverage by new trajectories. It is demonstrated that the
extensions significantly improve the overall performance of
the previous method in simulation on full nonlinear dynamics
model of the industrial Manutec robot arm.
Further, it is demonstrated that it suffices to learn a policy
that reaches the proximity of the goal state, from where a
PI controller can be used for stable control reaching the final
system state.