@TechReport{Pecka-TR-2014-17,
  IS = { zkontrolovano 02 Jan 2015 },
  UPDATE  = { 2014-10-27 },
author =      {Pecka, Martin and Svoboda, Tom{\'a}{\v s}},
title =       {Safe Exploration Techniques for Reinforcement Learning -- An Overview},
institution = {Center for Machine Perception, K13133 FEE
               Czech Technical University},
address =     {Prague, Czech Republic},
year =        {2014},
month =       {September},
type =        {Research Report},
number =      {CTU--CMP--2014--17},
issn =        {1213-2365},
pages =       {10},
figures =     {3},
authorship =  {50-50},
psurl       = {The publication is going to be issued in LNCS (Springer), so we'll add the link as soon as the publication is issued.},
project =     {SGS13/142/OHK3/2T/13, FP7-ICT-609763 TRADR},
annote =      {We overview diferent approaches to safety in
	       (semi)autonomous robotics. Particularly, we focus on how to
	       achieve safe behavior of a robot if it is requested to
	       perform exploration of unknown states. Presented methods
	       are studied from the viewpoint of reinforcement learning, a
	       partially-supervised machine learning method. Recently, it
	       has shown to be one of the most suitable learning methods
	       in robotics. However, to collect training data for this
	       algorithm, the robot is required to freely explore the
	       state space – which can lead to possibly dangerous
	       situations. The role of safe exploration is to provide a
	       framework allowing exploration while preserving safety. The
	       examined methods range from simple algorithms which utilize
	       precise physical models to sophisticated methods based on
	       previous experience or state prediction. Our overview also
	       addresses the issues of how to define safety in the
	       real-world applications. It is apparent that absolute
	       safety is unachievable in the continuous and random real
	       world, thus every exploration step may potentially threaten
	       the robot. In the conclusion we also suggest several ways
	       that are worth researching more thoroughly.},
keywords =    {Safe exploration, policy search, reinforcement learning.},
comment =     { Presented at Modeling and Simulation for autonomous systems Workshop - MESAS'14.},
}