Incremental kinesthetic teaching of end-effector and null-space motion primitives

Saveriano, Matteo; An, Sang-ik; Lee, Dongheui

doi:10.1109/icra.2015.7139694

Cited by 41 publications

(36 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In order to achieve intuitive teaching of natural motions, we proposed a method for incremental learning by using physical interaction [1] [3] [4]. In order to ensure synchronization of complex whole body motions on a humanoid robot, our imitation learning procedure starts with observation learning (i.e.…”

Section: Proposed Approachmentioning

confidence: 99%

“…We extended the kinesthetic teaching based on disturbance observers using force torque sensors in two aspects: to a legged robot [5], and to teaching multiple tasks using task transition [4]. In the legged full body humanoid case [5], the challenge is how to ensure balancing even there are external forces exerted by a human.…”

Section: Proposed Approachmentioning

confidence: 99%

“…In the second case [4], the focus is how to teach and iteratively refine multiple tasks (end-effector and null pace tasks) only by kinesthetic teaching. Although a whole body motion of a high DOF system is difficult to teach only by the kinesthetic teaching, if we utilize the interaction force measurements well, the robot can be taught to make use of all its degrees-of-freedom (DoF) without any external visual perception (eg., camera for observing human motions).…”

Section: Proposed Approachmentioning

confidence: 99%

See 2 more Smart Citations

Incremental robot skill learning by human motion retargetting and physical human guidance

Lee

2015

2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence (URAI)

Self Cite

View full text Add to dashboard Cite

-Research on skill acquisition and generalization to a different scenario has grown steadily in importance and became now a main topic of robotics research. Imitation learning, one of the main streams for robot learning, provides an efficient way to learn new skills through human guidance, which can reduce time and cost to program the robot. This extended abstract presents our research on incremental skill learning through physical human robot interactions. We introduce our method to teach a robot how to learn synchronized and coordinated whole body motions. Our controller provides a human user comfortable assistance for physical guidance beyond the gravity compensation. External force torque estimation allows further possibilities. One is teaching motion primitives of a legged humanoid robot by taking human intervention into consideration for a balancing problem. Another is teaching multiple tasks like end-effector motions and null space motions. The proposed algorithms are verified on multiple robotic systems including full size humanoid robots.Keywords -Incremental skill learning and refinement, physical human robot interaction.

show abstract

Section: Proposed Approachmentioning

confidence: 99%

Section: Proposed Approachmentioning

confidence: 99%

Section: Proposed Approachmentioning

confidence: 99%

See 1 more Smart Citation

Incremental robot skill learning by human motion retargetting and physical human guidance

Lee

2015

2015 12th International Conference on Ubiquitous Robots and Ambient Intelligence (URAI)

Self Cite

View full text Add to dashboard Cite

show abstract

“…An important category of applications currently attracting a lot of attention concerns the problems requiring priority constraints [44,28,19,50,36]. With an appropriate definition of the frames and with an initial set of candidate task hierarchies, such constraints can be learned and encoded within a TP-GMM.…”

Section: Extension To Minimal Intervention Controlmentioning

confidence: 99%

Robot Learning with Task-Parameterized Generative Models

Calinon

2017

Springer Proceedings in Advanced Robotics

View full text Add to dashboard Cite

Task-parameterized models provide a representation of movement/behavior that can adapt to a set of task parameters describing the current situation encountered by the robot, such as location of objects or landmarks in its workspace. This paper gives an overview of the taskparameterized Gaussian mixture model (TP-GMM) presented in previous publications, and introduces a number of extensions and ongoing challenges required to move the approach toward unconstrained environments. In particular, it discusses its generalization capability and the handling of movements with a high number of degrees of freedom. It then shows that the method is not restricted to movements in task space, but that it can also be exploited to handle constraints in joint space, including priority constraints.

show abstract

“…1) Data collection: Kinesthetic teaching is often used for data collection in PbD [6], [13]. In kinesthetic teaching, a teacher physically holds the robot's end-effector for generating the required motion.…”

Section: A Deep-dmpmentioning

confidence: 99%

Learning deep movement primitives using convolutional neural networks

Pervez

Mao

Lee

2017

2017 IEEE-RAS 17th International Conference on Humanoid Robotics (Humanoids)

Self Cite

View full text Add to dashboard Cite

Abstract-Dynamic Movement Primitives (DMPs) are widely used for encoding motion data. Task parameterized DMP (TP-DMP) can adapt a learned skill to different situations. Mostly a customized vision system is used to extract task specific variables. This limits the use of such systems to real world scenarios. This paper proposes a method for combining the DMP with a Convolutional Neural Network (CNN). Our approach preserves the generalization properties associated with a DMP, while the CNN learns the task specific features from the camera images. This eliminates the need to extract the task parameters, by directly utilizing the camera image during the motion reproduction. The performance of the developed approach is demonstrated through a trash cleaning task, executed with a real robot. We also show that by using the data augmentation, the learned sweeping skill can be generalized for arbitrary objects. The experiments show the robustness of our approach for several different settings.

show abstract

Incremental kinesthetic teaching of end-effector and null-space motion primitives

Cited by 41 publications

References 17 publications

Incremental robot skill learning by human motion retargetting and physical human guidance

Incremental robot skill learning by human motion retargetting and physical human guidance

Robot Learning with Task-Parameterized Generative Models

Learning deep movement primitives using convolutional neural networks

Contact Info

Product

Resources

About