Abstract:Learning robot manipulation skills from human demonstration is one of the key technologies for the cooperative robot. However, the traditional learning methods are difficult to be rapidly applied to complex force interaction tasks such as button-pressing, assembly, polishing, etc, which mainly focus on acquiring the movement skill by encoding the kinematics features and lacking consideration of the interactive force, stiffness, and other compliant profiles. Meanwhile, due to a large amount of high-dimensional demonstrations and task constraints, the existing methods are faced with the problems of low learning efficiency and difficult online modulation. To address those problems, a novel framework of robot skill learning is presented in this paper, which not only can simultaneously consider the movement and compliant profiles but also the implementation of effective learning and online modulation. Specifically, a Gaussian Mixture Model-Gaussian Mixture Regression (GMM-GMR) is first proposed for learning underlying dynamics from demonstrated trajectories, such that the skill can be formulated with two parameters at each time step, i.e. mean and variance. And then, a GMMGMR-based vector-valued Gaussian Process is proposed to autonomously modulate the trajectory according to the unexpected environmental changes and perturbations, such as unseen start-, via- or end-points. Those points are introduced by an external-force aware variable impedance controller by human intervention during skill reproduction. Consequently, a nonlinear mapping relationship between the variance and robot stiffness is investigated for encapsulating the compliant profiles. Experimental verification is performed on a button-pressing task with a Franka-Emika panda robot, results indicate that the presented framework can learn humanlike manipulation skills and reduce the complexity of robot application in complex force interaction tasks.

SwingBot: Learning Physical Features from In-hand Tactile Exploration for Dynamic Swing-up Manipulation

Dexterous Manoeuvre Through Touch in a Cluttered Scene

Object-Centric Dexterous Manipulation from Human Motion Data

Dexterous In-Hand Manipulation of Slender Cylindrical Objects through Deep Reinforcement Learning with Tactile Sensing

Learning the Dynamics of Compliant Tool-Environment Interaction for Visuo-Tactile Contact Servoing

Slip Detection through Fusion of Visual-Tactile Data using Swin Transformer V2

Force, Impedance, and Trajectory Learning for Contact Tooling and Haptic Identification

Learning Fish-Like Swimming with a Cpg-Based Locomotion Controller

Learning Stable Robot Grasping with Transformer-based Tactile Control Policies

Learning active tactile perception through belief-space control

Feeling the Force: Integrating Force and Pose for Fluent Discovery Through Imitation Learning to Open Medicine Bottles.

Learning Diverse and Physically Feasible Dexterous Grasps with Generative Model and Bilevel Optimization

Learning Object Properties Using Robot Proprioception via Differentiable Robot-Object Interaction

Learning to use chopsticks in diverse gripping styles

Learning compliant grasping and manipulation by teleoperation with adaptive force control

Learned Slip-Detection-Severity Framework using Tactile Deformation Field Feedback for Robotic Manipulation

Effective Learning and Online Modulation for Robotic Variable Impedance Skills.

ASID: Active Exploration for System Identification in Robotic Manipulation

Visual-Tactile Based Manipulation Control for Complex Robot Manipulation Process Via Geometric and Contact State Modeling

Learning Extrinsic Dexterity with Parameterized Manipulation Primitives

Learning thin deformable object manipulation with a multi-sensory integrated soft hand