Abstract
In this paper we discuss the application of reinforcement learning
algorithms to the problem of autonomous robot navigation. We show that
the autonomous navigation using the standard delayed reinforcement
learning algorithms is an ill posed problem and we present a more
efficient algorithm for which the convergence speed is greatly improved.
The proposed algorithm (Deep-Sarsa) is based on a combination
between the Depth-First Search (a graph searching algorithm) and
Sarsa (a delayed reinforcement learning algorithm).