Learning Robotic Assembly Tasks with Lower Dimensional Systems by Leveraging Physical Softness and Environmental Constraints

Hamaya, Masashi; Lee, Robert; Tanaka, Kazutoshi; Drigalski, Felix von; Nakashima, Chisato; Shibata, Yoshiya; Ijiri, Yoshihisa

doi:10.1109/icra40945.2020.9197327

Cited by 21 publications

(6 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Martín-Martín et al (2019) similarly proposed to learn the controller selection and parameterization during a peg-in-hole task. Hamaya et al (2020) applied a model-based RL via GP on a peg-in-hole task for an industrial position-controlled robot by attaching a compliant wrist to the robot end-effector, which compensates for perception inaccuracy. Mitsioni et al (2021) instead proposed to learn the environment dynamics from an NN in order to apply a model predictive control, if the current state is classified as safe via a GP classifier.…”

Section: Robot Skill Learning On Reduced Parameter Spacesmentioning

confidence: 99%

Bayesian optimization with unknown constraints in graphical skill models for compliant manipulation tasks using an industrial robot

Gabler

Wollherr

2022

Front. Robot. AI

View full text Add to dashboard Cite

This article focuses on learning manipulation skills from episodic reinforcement learning (RL) in unknown environments using industrial robot platforms. These platforms usually do not provide the required compliant control modalities to cope with unknown environments, e.g., force-sensitive contact tooling. This requires designing a suitable controller, while also providing the ability of adapting the controller parameters from collected evidence online. Thus, this work extends existing work on meta-learning for graphical skill-formalisms. First, we outline how a hybrid force–velocity controller can be applied to an industrial robot in order to design a graphical skill-formalism. This skill-formalism incorporates available task knowledge and allows for online episodic RL. In contrast to the existing work, we further propose to extend this skill-formalism by estimating the success probability of the task to be learned by means of factor graphs. This method allows assigning samples to individual factors, i.e., Gaussian processes (GPs) more efficiently and thus allows improving the learning performance, especially at early stages, where successful samples are usually only drawn in a sparse manner. Finally, we propose suitable constraint GP models and acquisition functions to obtain new samples in order to optimize the information gain, while also accounting for the success probability of the task. We outline a specific application example on the task of inserting the tip of a screwdriver into a screwhead with an industrial robot and evaluate our proposed extension against the state-of-the-art methods. The collected data outline that our method allows artificial agents to obtain feasible samples faster than existing approaches, while achieving a smaller regret value. This highlights the potential of our proposed work for future robotic applications.

show abstract

Section: Robot Skill Learning On Reduced Parameter Spacesmentioning

confidence: 99%

Bayesian optimization with unknown constraints in graphical skill models for compliant manipulation tasks using an industrial robot

Gabler

Wollherr

2022

Front. Robot. AI

View full text Add to dashboard Cite

show abstract

“…Although this choice compromises the reduced training time, we adopt sim2real to address this issue and argue that the use of MPs as meta-actions makes the sim to real transfer more efficient. In the same vein, Hamaya et al [5] divide the peg-in-hole task into five steps with different action spaces and state spaces. This decomposition greatly speeds up training through dimensional reduction of action and state spaces.…”

Section: Related Workmentioning

confidence: 99%

Learning Sequences of Manipulation Primitives for Robotic Assembly

Vuong¹,

Pham²,

Pham³

2020

Preprint

View full text Add to dashboard Cite

This paper explores the idea that skillful assembly is best represented as dynamic sequences of Manipulation Primitives, and that such sequences can be automatically discovered by Reinforcement Learning. Manipulation Primitives, such as "Move down until contact", "Slide along x while maintaining contact with the surface", have enough complexity to keep the search tree shallow, yet are generic enough to generalize across a wide range of assembly tasks. Policies are learned in simulation, and then transferred onto the physical platform. Direct sim2real transfer (without retraining in real) achieves excellent success rates on challenging assembly tasks, such as round peg insertion with 100 micron clearance or square peg insertion with large hole position/orientation estimation errors.

show abstract

“…In introducing the compliance device, Hamaya et al [11] equipped the robot with a compliant module on its wrist and presented a novel control framework for exploring assembly strategies based on the softness and environmental constraints. Xing et al [12] utilized a compliant mechanism with multiple degree of freedoms to realize compliant insertion in precision assembly and proposed an effective assembly strategy accordingly.…”

Section: Introductionmentioning

confidence: 99%

Compliant peg-in-hole assembly for nonconvex axisymmetric components based on attractive region in environment

et al. 2023

View full text Add to dashboard Cite

With the development of intelligent manufacturing, more and more nonstandard parts are used in high-precision assembly. The robotic assembly method based on attractive region in environment (ARIE) has been proven to have good performance in the high-precision assembly under the limitation of robot system accuracy or sensing accuracy. However, for the assembly of nonstandard parts, especially nonconvex parts, the existing ARIE-based strategy lacks a targeted design. In the assembly process, the nonconvex structure may cause blocking problems, which will lead to assembly failure when using the strategy. In order to solve this problem, this paper proposes a new assembly method by using the geometric features of constraint region based on the concept of ARIE. Specifically, first, when using the ARIE-based classic strategy, the reasons for the possible blocking problem in the assembly of a class of nonconvex axisymmetric parts are analyzed in detail. Second, a multi-step sliding strategy is proposed based on the theory of ARIE to solve the possible blocking problem in the assembly process. Third, impedance control is used to enable the peg to achieve the desired compliant motion in the proposed strategy. The improvement in the success rate of the proposed method is verified by the comparison experiment of small clearance peg-in-hole assembly, where the structure of the peg is nonconvex and axisymmetric.

show abstract

Learning Robotic Assembly Tasks with Lower Dimensional Systems by Leveraging Physical Softness and Environmental Constraints

Cited by 21 publications

References 24 publications

Bayesian optimization with unknown constraints in graphical skill models for compliant manipulation tasks using an industrial robot

Bayesian optimization with unknown constraints in graphical skill models for compliant manipulation tasks using an industrial robot

Learning Sequences of Manipulation Primitives for Robotic Assembly

Compliant peg-in-hole assembly for nonconvex axisymmetric components based on attractive region in environment

Contact Info

Product

Resources

About