Whole‐Body Multicontact Haptic Human–Humanoid Interaction Based on Leader–Follower Switching: A Robot Dance of the “Box Step”

Kobayashi, Taisuke; Dean-León, Emmanuel; Guadarrama-Olvera, Julio Rogelio; Bergner, Florian; Cheng, Gordon

doi:10.1002/aisy.202100038

Cited by 19 publications

(7 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As the workforce shrinks and robotic technology develops, the tasks required of robots becomes more complex: e.g. physical human-robot interaction [1], [2]; work on disaster sites [3], [4]; and manipulation of various objects [5], [6]. Many of these complex tasks are difficult to design in advance, and reinforcement learning (RL) [7] has received a lot of attention as an alternative to classic model-based control.…”

Section: Introductionmentioning

confidence: 99%

L2C2: Locally Lipschitz Continuous Constraint towards Stable and Smooth Reinforcement Learning

Kobayashi¹

2022

Preprint

Self Cite

View full text Add to dashboard Cite

This paper proposes a new regularization technique for reinforcement learning (RL) towards making policy and value functions smooth and stable. RL is known for the instability of the learning process and the sensitivity of the acquired policy to noise. Several methods have been proposed to resolve these problems, and in summary, the smoothness of policy and value functions learned mainly in RL contributes to these problems. However, if these functions are extremely smooth, their expressiveness would be lost, resulting in not obtaining the global optimal solution. This paper therefore considers RL under local Lipschitz continuity constraint, socalled L2C2. By designing the spatio-temporal locally compact space for L2C2 from the state transition at each time step, the moderate smoothness can be achieved without loss of expressiveness. Numerical noisy simulations verified that the proposed L2C2 outperforms the task performance while smoothing out the robot action generated from the learned policy.

show abstract

Section: Introductionmentioning

confidence: 99%

L2C2: Locally Lipschitz Continuous Constraint towards Stable and Smooth Reinforcement Learning

Kobayashi¹

2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Children could interact with robots, such as dancing with robots and improve their concentration abilities. Kobayashi et al ( 2022 ) proposed an exploratory study based on impromptu dance to explore the interaction between human dancers and mobile robots. A basic improvisation action algorithm was developed after many iterations with professional dancers.…”

Section: Introductionmentioning

confidence: 99%

Influence of Human–Computer Interaction-Based Intelligent Dancing Robot and Psychological Construct on Choreography

Yang

2022

Front. Neurorobot.

View full text Add to dashboard Cite

To study the influence of Artificial Intelligence (AI) on dancing robots in choreography, this paper introduces the biped-humanoid robot-imagined choreography model alongside the Psychological Space Construction (Psychological Construct) and Human–Computer Interaction (HCI). The proposed model is based on deep learning and imitating human thinking and is capable of imagining new dance elements. Finally, simulation experiments are designed to verify the model's effectiveness. Dance professionals are invited to evaluate the robot-imagined dance posture. The results show that the proposed model can vividly imitate human dancers and imagine and create new dance movements. The average basic feature retention and innovation scores of 30 new dance elements imagined on the L1 (head) are 7.29 and 7.64, respectively. By comparison, similar scores on 30 new elements in L2 (upper-body) are 7.73 and 7.40, respectively. Therefore, the proposed intelligent robot-imagined choreography model can help the dancing robot choreograph more finely and improve the choreography efficiency. The research results have significant practical value for dance teaching.

show abstract

“…As robotic and machine learning technologies remarkably develop, the tasks required of intelligent robots becomes more complex: e.g. physical human-robot interaction [1], [2]; work on disaster sites [3], [4]; and manipulation of various objects [5], [6]. In most cases, these complex tasks have no accurate analytical model.…”

Section: Introductionmentioning

confidence: 99%

Consolidated Adaptive T-soft Update for Deep Reinforcement Learning

Kobayashi¹

2022

Preprint

Self Cite

View full text Add to dashboard Cite

Demand for deep reinforcement learning (DRL) is gradually increased to enable robots to perform complex tasks, while DRL is known to be unstable. As a technique to stabilize its learning, a target network that slowly and asymptotically matches a main network is widely employed to generate stable pseudo-supervised signals. Recently, T-soft update has been proposed as a noise-robust update rule for the target network and has contributed to improving the DRL performance. However, the noise robustness of T-soft update is specified by a hyperparameter, which should be tuned for each task, and is deteriorated by a simplified implementation. This study develops adaptive T-soft (AT-soft) update by utilizing the update rule in AdaTerm, which has been developed recently. In addition, the concern that the target network does not asymptotically match the main network is mitigated by a new consolidation for bringing the main network back to the target network. This so-called consolidated AT-soft (CAT-soft) update is verified through numerical simulations.

show abstract

Whole‐Body Multicontact Haptic Human–Humanoid Interaction Based on Leader–Follower Switching: A Robot Dance of the “Box Step”

Abstract: The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/aisy.202100038.

Cited by 19 publications

References 27 publications

L2C2: Locally Lipschitz Continuous Constraint towards Stable and Smooth Reinforcement Learning

L2C2: Locally Lipschitz Continuous Constraint towards Stable and Smooth Reinforcement Learning

Influence of Human–Computer Interaction-Based Intelligent Dancing Robot and Psychological Construct on Choreography

Consolidated Adaptive T-soft Update for Deep Reinforcement Learning

Contact Info

Product

Resources

About