000 06631nam a2200721 i 4500
001 6828193
003 IEEE
005 20200413152914.0
006 m eo d
007 cr cn |||m|||a
008 140520s2014 caua foab 000 0 eng d
020 _a9781627052009
_qebook
020 _z9781627051996
_qpaperback
024 7 _a10.2200/S00568ED1V01Y201402AIM028
_2doi
035 _a(CaBNVSL)swl00403381
035 _a(OCoLC)880357617
040 _aCaBNVSL
_beng
_erda
_cCaBNVSL
_dCaBNVSL
050 4 _aQ325.5
_b.C447 2014
082 0 4 _a006.31
_223
090 _a
_bMoCl
_e201402AIM028
100 1 _aChernova, Sonia.,
_eauthor.
245 1 0 _aRobot learning from human teachers /
_cSonia Chernova, Andrea L. Thomaz.
264 1 _aSan Rafael, California (1537 Fourth Street, San Rafael, CA 94901 USA) :
_bMorgan & Claypool,
_c2014.
300 _a1 PDF (xi, 109 pages) :
_billustrations.
336 _atext
_2rdacontent
337 _aelectronic
_2isbdmedia
338 _aonline resource
_2rdacarrier
490 1 _aSynthesis lectures on artificial intelligence and machine learning,
_x1939-4616 ;
_v# 28
538 _aMode of access: World Wide Web.
538 _aSystem requirements: Adobe Acrobat Reader.
500 _aPart of: Synthesis digital library of engineering and computer science.
500 _aSeries from website.
504 _aIncludes bibliographical references (pages 83-107).
505 0 _a1. Introduction -- 1.1 Machine learning for end-users -- 1.2 The learning from demonstration pipeline -- 1.3 A note on terminology --
505 8 _a2. Human social learning -- 2.1 Learning is a part of all activity -- 2.2 Teachers scaffold the learning process -- 2.2.1 Attention direction -- 2.2.2 Dynamic scaffolding -- 2.2.3 Externalizing and modeling metacognition -- 2.3 Role of communication in social learning -- 2.3.1 Expression provides feedback to guide a teacher -- 2.3.2 Asking questions -- 2.4 Implications for the design of robot learners --
505 8 _a3. Modes of interaction with a teacher -- 3.1 The correspondence problem -- 3.2 Learning by doing -- 3.3 Learning from observation -- 3.4 Learning from critique -- 3.5 Design implications --
505 8 _a4. Learning low-level motion trajectories -- 4.1 State spaces for motion learning -- 4.2 Modeling an action with dynamic movement primitives -- 4.3 Modeling action with probabilistic models -- 4.4 Techniques for handling suboptimal demonstrations --
505 8 _a5. Learning high-level tasks -- 5.1 State spaces for high-level learning -- 5.2 Learning a mapping function -- 5.3 Learning a task plan -- 5.4 Learning task objectives -- 5.5 Learning task features -- 5.6 Learning frame of reference -- 5.7 Learning object affordances -- 5.8 Techniques for handling suboptimal demonstrations -- 5.9 Discussion and open challenges --
505 8 _a6. Refining a learned task -- 6.1 Batch vs. incremental learning -- 6.2 Reinforcement learning based methods -- 6.3 Corrective refinement from the teacher -- 6.4 Active learning -- 6.4.1 Label queries -- 6.4.2 Demonstration queries -- 6.4.3 Feature queries -- 6.5 Summary --
505 8 _a7. Designing and evaluating an LfD study -- 7.1 Experimental design -- 7.2 Evaluating the algorithmic performance -- 7.3 Evaluating the interaction -- 7.3.1 Subjective measures -- 7.3.2 Objective measures -- 7.4 Experimental controls -- 7.5 Experimental protocol -- 7.6 Data analysis -- 7.6.1 Choosing the right statistical tool -- 7.6.2 Drawing conclusions -- 7.7 Additional resources --
505 8 _a8. Future challenges and opportunities -- 8.1 Real users, real tasks -- 8.2 HRI considerations -- 8.3 Advancing learning through benchmarking and integration -- 8.4 Opportunities -- 8.5 Additional resources --
505 8 _aBibliography -- Authors' biographies.
506 1 _aAbstract freely available; full-text restricted to subscribers or individual document purchasers.
510 0 _aCompendex
510 0 _aINSPEC
510 0 _aGoogle scholar
510 0 _aGoogle book search
520 3 _aLearning from Demonstration (LfD) explores techniques for learning a task policy from examples provided by a human teacher. The field of LfD has grown into an extensive body of literature over the past 30 years, with a wide variety of approaches for encoding human demonstrations and modeling skills and tasks. Additionally, we have recently seen a focus on gathering data from nonexpert human teachers (i.e., domain experts but not robotics experts). In this book, we provide an introduction to the field with a focus on the unique technical challenges associated with designing robots that learn from naive human teachers. We begin, in the introduction, with a unification of the various terminology seen in the literature as well as an outline of the design choices one has in designing an LfD system. Chapter 2 gives a brief survey of the psychology literature that provides insights from human social learning that are relevant to designing robotic social learners. Chapter 3 walks through an LfD interaction, surveying the design choices one makes and state of the art approaches in prior work. First, is the choice of input, how the human teacher interacts with the robot to provide demonstrations. Next, is the choice of modeling technique. Currently, there is a dichotomy in the field between approaches that model low-level motor skills and those that model high-level tasks composed of primitive actions. We devote a chapter to each of these. Chapter 7 is devoted to interactive and active learning approaches that allow the robot to refine an existing task model. And finally, Chapter 8 provides best practices for evaluation of LfD systems, with a focus on how to approach experiments with human subjects in this domain.
530 _aAlso available in print.
588 _aTitle from PDF title page (viewed on May 20, 2014).
650 0 _aMachine learning.
650 0 _aRobots
_xControl systems.
650 0 _aHuman-robot interaction.
653 _aLearning from Demonstration
653 _aimitation learning
653 _aHuman-robot Interaction
700 1 _aThomaz, Andrea Lockerd.,
_eauthor.
776 0 8 _iPrint version:
_z9781627051996
830 0 _aSynthesis digital library of engineering and computer science.
830 0 _aSynthesis lectures on artificial intelligence and machine learning ;
_v# 28.
_x1939-4616
856 4 2 _3Abstract with links to resource
_uhttp://ieeexplore.ieee.org/servlet/opac?bknumber=6828193
856 4 0 _3Abstract with links to full text
_uhttp://dx.doi.org/10.2200/S00568ED1V01Y201402AIM028
999 _c562068
_d562068