Learning Augmented Joint-Space Task-Oriented Dynamical Systems: A Linear Parameter Varying and Synergetic Control Approach

Shavit, Yonadav; Figueroa, Nadia; Salehian, Seyed Sina Mirrazavi; Billard, Aude

doi:10.1109/lra.2018.2833497

Cited by 16 publications

(8 citation statements)

References 30 publications

(38 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Human safety is one of the biggest concerns in HRC. Previous works [11,12] focused on humans' physical safety of HRC. Many methods are proposed for collision-free motion planning, such as probabilistic roadmap (PRM) and rapidly exploring random trees (RRT).…”

Section: Related Workmentioning

confidence: 99%

“…After the experiment, count the mean value of the Sp value of all participants for all speeds and minimum distance of the head, chest, and abdomen; then, take the mean value of Sp value as the law of psychological stress of participants for cobot approaching motion and conduct surface fitting to obtain the law formula of psychological stress considering cobot motion. The law formulas for head, chest, and abdomen are Formulas ( 9)- (11), respectively. The parameters in Equations ( 9)-( 11) are shown in Table A2 in Appendix A.…”

Section: Psychological Stress Field Modelmentioning

confidence: 99%

“…The parameters in Equations (9)-(11).Sh = z oh + a h • d + b h • v + c h • d 2 + d h • v 2 + e h • d • v…”

unclassified

See 2 more Smart Citations

Cobot Motion Planning Algorithm for Ensuring Human Safety Based on Behavioral Dynamics

Liu

Wang

et al. 2022

Sensors

View full text Add to dashboard Cite

Recently, the safety of workers has gained increasing attention due to the applications of collaborative robots (cobot). However, there is no quantitative research on the impact of cobot behavior on humans’ psychological reactions, and these results are not applied to the cobot motion planning algorithms. Based on the concept of the gravity field, this paper proposes a model of the psychological safety field (PSF), designs a comprehensive experiment on different speeds and minimum distances when approaching the head, chest, and abdomen, and obtains the ordinary surface equation of psychological stress about speed and minimum distance by using data fitting. By combining social rules and PSF models, we improve the robot motion planning algorithm based on behavioral dynamics. The validation experiment results show that our proposed improved robot motion planning algorithm can effectively reduce psychological stress. Eighty-seven point one percent (87.1%) of the experimental participants think that robot motion planned by improved robot motion planning algorithms is more “friendly”, can effectively reduce psychological stress, and is more suitable for human–robot interaction scenarios.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Psychological Stress Field Modelmentioning

confidence: 99%

See 1 more Smart Citation

Cobot Motion Planning Algorithm for Ensuring Human Safety Based on Behavioral Dynamics

Liu

Wang

et al. 2022

Sensors

View full text Add to dashboard Cite

show abstract

“…Both types of demonstration learning typically require the expert demonstrator to be in the same context as the learner, limiting their scalability to real-world applications. Moreover, applications that train the learner in the same context as the demonstrator rely on kinesthetic Maeda et al [2017], Shavit et al [2018] or teleoperation Abbeel et al [2010], Ng et al [2004], Aler et al [2005] demonstration techniques. However, these require expert skills to perform and still limit the observation setting to the one in the demonstration.…”

Section: Related Workmentioning

confidence: 99%

Contrastive Learning from Demonstrations

Correia¹,

Alexandre²

2022

Preprint

View full text Add to dashboard Cite

This paper presents a framework for learning visual representations from unlabeled video demonstrations captured from multiple viewpoints. We show that these representations are applicable for imitating several robotic tasks, including pick and place. We optimize a recently proposed selfsupervised learning algorithm by applying contrastive learning to enhance task-relevant information while suppressing irrelevant information in the feature embeddings. We validate the proposed method on the publicly available Multi-View Pouring and a custom Pick and Place data sets and compare it with the TCN triplet baseline. We evaluate the learned representations using three metrics: viewpoint alignment, stage classification and reinforcement learning, and in all cases the results improve when compared to state-of-the-art approaches, with the added benefit of reduced number of training iterations.

show abstract

“…As this case involves passive observation imitation learning, it suffers from the correspondence issue to transform the demonstration from the teacher's joint space to the robot's joint space. In Shavit et al ( 2018 ), a dynamical system (DS) is proposed to learn from kinesthetic demonstrations. The DS is then capable of computing the desired motion to be executed in joint space to reach a target in task-space.…”

Section: Introductionmentioning

confidence: 99%

A Bio-Inspired Mechanism for Learning Robot Motion From Mirrored Human Demonstrations

et al. 2022

View full text Add to dashboard Cite

Different learning modes and mechanisms allow faster and better acquisition of skills as widely studied in humans and many animals. Specific neurons, called mirror neurons, are activated in the same way whether an action is performed or simply observed. This suggests that observing others performing movements allows to reinforce our motor abilities. This implies the presence of a biological mechanism that allows creating models of others' movements and linking them to the self-model for achieving mirroring. Inspired by such ability, we propose to build a map of movements executed by a teaching agent and mirror the agent's state to the robot's configuration space. Hence, in this study, a neural network is proposed to integrate a motor cortex-like differential map transforming motor plans from task-space to joint-space motor commands and a static map correlating joint-spaces of the robot and a teaching agent. The differential map is developed based on spiking neural networks while the static map is built as a self-organizing map. The developed neural network allows the robot to mirror the actions performed by a human teaching agent to its own joint-space and the reaching skill is refined by the complementary examples provided. Hence, experiments are conducted to quantify the improvement achieved thanks to the proposed learning approach and control scheme.

show abstract

Learning Augmented Joint-Space Task-Oriented Dynamical Systems: A Linear Parameter Varying and Synergetic Control Approach

Cited by 16 publications

References 30 publications

Cobot Motion Planning Algorithm for Ensuring Human Safety Based on Behavioral Dynamics

Cobot Motion Planning Algorithm for Ensuring Human Safety Based on Behavioral Dynamics

Contrastive Learning from Demonstrations

A Bio-Inspired Mechanism for Learning Robot Motion From Mirrored Human Demonstrations

Contact Info

Product

Resources

About