Optimality of human teachers for robot learners

Çakmak, Maya; Thomaz, Andrea L.

doi:10.1109/devlrn.2010.5578865

Cited by 24 publications

(25 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…how it scales to continuous domain. We will also consider how more complex instructions can be included in our formalism since the teaching models used spontaneously by people can be more complex than the simple meaning correspondences we assumed [7], [15]. Also the protocol could be enhanced to be more natural, the robot could ask questions [25] and accept asynchronous signals.…”

Section: Conclusionlimitations and Future Workmentioning

confidence: 99%

“…Several studies discuss the various behaviors naive teachers use when instructing robots [7], [15]. An important aspect is that the feedback is frequently ambiguous and deviates from the mathematical interpretation of a reward; or a sample trajectory deviates from an optimal policy.…”

Section: Introductionmentioning

confidence: 99%

“…For instance, in the work of [7] the teachers frequently gave a positive reward for exploratory actions even if the signal was used by the learner as a standard reward. Also, even if we can define an optimal teaching sequence, humans do not necessarily behave according to those strategies [15].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Robot learning simultaneously a task and how to interpret human instructions

Grizou

Lopes

Oudeyer

2013

2013 IEEE Third Joint International Conference on Development and Learning and Epigenetic Robotics (ICDL)

View full text Add to dashboard Cite

Abstract-This paper presents an algorithm to bootstrap shared understanding in a human-robot interaction scenario where the user teaches a robot a new task using teaching instructions yet unknown to it. In such cases, the robot needs to estimate simultaneously what the task is and the associated meaning of instructions received from the user. For this work, we consider a scenario where a human teacher uses initially unknown spoken words, whose associated unknown meaning is either a feedback (good/bad) or a guidance (go left, right, ...). We present computational results, within an inverse reinforcement learning framework, showing that a) it is possible to learn the meaning of unknown and noisy teaching instructions, as well as a new task at the same time, b) it is possible to reuse the acquired knowledge about instructions for learning new tasks, and c) even if the robot initially knows some of the instructions' meanings, the use of extra unknown teaching instructions improves learning efficiency.

show abstract

Section: Conclusionlimitations and Future Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Robot learning simultaneously a task and how to interpret human instructions

Grizou

Lopes

Oudeyer

2013

2013 IEEE Third Joint International Conference on Development and Learning and Epigenetic Robotics (ICDL)

View full text Add to dashboard Cite

show abstract

“…The results regarding the ways information is structured and presented to the robot are likely to have an impact on the success of socially guided learning (see Thomaz and Breazeal [43], Thomaz and Cakmak [44], Cakmak et al [5,6]). As Thomaz and Cakmak [44], for instance, demonstrate, naïve users structure information for robots intuitively in ways that improve learning from demonstration.…”

Section: Design Implicationsmentioning

confidence: 99%

The impact of the contingency of robot feedback on HRI

Fischer

Lohan

Saunders

et al. 2013

2013 International Conference on Collaboration Technologies and Systems (CTS)

View full text Add to dashboard Cite

In this paper, we investigate the impact the contingency of robot feedback may have on the quality of verbal human-robot interaction. In order to assess not only what the effects are but also what they are caused by, we carried out experiments in which na??ve participants instructed the humanoid robot iCub on a set of shapes and on a stacking task in two conditions, once with socially contingent, nonverbal feedback implemented in response to different gaze and demonstrating behaviors of the human tutor, and once with non-contingent, saliency-based feedback. The results of the analysis of participants' linguistic behaviors in the two conditions show that contingency has an impact on the complexity and the pre-structuring of the task for the robot, i.e. on the participants' tutoring behaviors. Contingency thus plays a considerable role for learning by demonstration. ?? 2013 IEEE

show abstract

“…Several studies discuss the different behaviors naive teach ers use when instructing robots [7], [18]. An important aspect is that, many times, the feedback is ambiguous and deviates from the mathematical interpretation of a reward or a sample from a policy.…”

Section: Introductionmentioning

confidence: 99%

Simultaneous acquisition of task and feedback models

Lopes

Cederbourg

Oudeyer

2011

2011 IEEE International Conference on Development and Learning (ICDL)

View full text Add to dashboard Cite

Ahstract-We present a system to learn task representations from ambiguous feedback. We consider an inverse reinforce ment learner that receives feedback from a teacher with an unknown and noisy protocol. The system needs to estimate simultaneously what the task is (i.e. how to find a compact representation to the task goal), and how the teacher is providing the feedback. We further explore the problem of ambiguous protocols by considering that the words used by the teacher have an unknown relation with the action and meaning expected by the robot. This allows the system to start with a set of known signs and learn the meaning of new ones. We present computational results that show that it is possible to learn the task under a noisy and ambiguous feedback. Using an active learning approach, the system is able to reduce the length of the training period.

show abstract

Optimality of human teachers for robot learners

Cited by 24 publications

References 7 publications

Robot learning simultaneously a task and how to interpret human instructions

Robot learning simultaneously a task and how to interpret human instructions

The impact of the contingency of robot feedback on HRI

Simultaneous acquisition of task and feedback models

Contact Info

Product

Resources

About