Intuitive skill-level programming of industrial handling tasks on a mobile manipulator

Pedersen, Mikkel Rath; Herzog, Dennis; Krüger, Volker

doi:10.1109/iros.2014.6943203

Cited by 19 publications

(22 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The exact scenario and set of available skills differ slightly between the two experiments. In this paper we present additional results compared to our previous work [37], as well as qualitative feedback from all participants.…”

Section: Methodsmentioning

confidence: 96%

“…However, our previous experiments show that for the skill approach, the world model need only contain information about the valid parameters for the skills running on the robot system, e.g. objects or locations [37,38].…”

Section: Skill Modelmentioning

confidence: 99%

“…For testing this approach, boxes and locations are marked by QR codes, where the data contained in the code specifies the unique object or location. The 3D poses of the QR codes are detected using the Kinect camera on the robot -for details we refer to [37].…”

Section: Skill Implementationmentioning

confidence: 99%

“…For these experiments we use the direction of the forearm as the pointing direction. For additional details of the gesture recognition we refer to our previous work [37].…”

Section: Human Tracking and Gesture Recognitionmentioning

confidence: 99%

“…This paper builds upon our previous work, described in [37], where we in this paper have increased the number of experiments, and included qualitative feedback from all participants. Since it is not feasible to explicitly program reoccurring skill sequences for each box (the robot can only carry one box, and the number of boxes at the station may vary), we also investigate a method for including loops in the presented programming approach.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Gesture-Based Extraction of Robot Skill Parameters for Intuitive Robot Programming

Pedersen

Krüger

2015

J Intell Robot Syst

Self Cite

View full text Add to dashboard Cite

Despite a lot of research in the field, only very little experience exists with Teaching by Demonstration (TbD) in actual industrial use cases. In the factory of the future, it is necessary to rapidly reprogram flexible mobile manipulators to perform new tasks, when the need arises, for which a working system capable of TbD would be ideal. Contrary to current TbD approaches, that generally aim to recognize both action and where it is applied, we propose a division of labor, where the operator manually specifies the action the robot should perform, while gestures are used for specifying the relevant action parameter (e.g. on which object to apply the action). Using this two-step method has the advantages that there is no uncertainty of which action the robot will perform, it takes into account that the environment changes, so objects do not need to be at predefined locations, and the parameter specification is possible even for inexperienced users. Experiments with 24 people in 3 different environments verify that it is indeed intuitive, even for a robotics novice, to program a mobile manipulator using this method. M. R. Pedersen ( ) · V. Krüger Robotics, Vision and Machine Intelligence (RVMI) Lab,

show abstract

Section: Methodsmentioning

confidence: 96%

Section: Skill Modelmentioning

confidence: 99%

Section: Skill Implementationmentioning

confidence: 99%

“…For these experiments we use the direction of the forearm as the pointing direction. For additional details of the gesture recognition we refer to our previous work [37].…”

Section: Human Tracking and Gesture Recognitionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations