Attaining autonomous flight is an important task
in aerial robotics. Often flight trajectories are not only subject to
unknown system dynamics, but also to specific task constraints.
We are interested in producing a trajectory for an aerial robot
with a suspended load that delivers the load to a destination
in a swing-free fashion. This paper presents a motion planning
framework for generating trajectories with minimal residual
oscillations (swing-free) for rotorcraft carrying a suspended
load. We rely on a finite-sampling, batch reinforcement learning
algorithm to train the system for a particular load. We find
the criteria that allow the trained agent to be transferred to a
variety of models, state and action spaces and produce a number
of different trajectories. Through a combination of simulations
and experiments, we demonstrate that the inferred policy is
robust to noise and to the unmodeled dynamics of the system.
The contributions of this work are 1) applying reinforcement
learning to solve the problem of finding a swing-free trajectory
for a rotorcraft, 2) designing a problem-specific feature vector
for value function approximation, 3) giving sufficient conditions
that need to be met to allow successful learning transfer to
different models, state and action spaces, and 4) verification
of the resulting trajectories in simulation and to autonomously
control quadrotors.