Kernelized movement primitives

@article{Huang2017KernelizedMP,
  title={Kernelized movement primitives},
  author={Yanlong Huang and Leonel Dario Rozo and Jo{\~a}o Silv{\'e}rio and Darwin Gordon Caldwell},
  journal={The International Journal of Robotics Research},
  year={2017},
  volume={38},
  pages={833 - 852}
}
Imitation learning has been studied widely as a convenient way to transfer human skills to robots. This learning approach is aimed at extracting relevant motion patterns from human demonstrations and subsequently applying these patterns to different situations. Despite the many advancements that have been achieved, solutions for coping with unpredicted situations (e.g., obstacles and external perturbations) and high-dimensional inputs are still largely absent. In this paper, we propose a novel… 

Uncertainty-Aware Imitation Learning using Kernelized Movement Primitives

KMP is leveraged to provide a new perspective on imitation learning by predicting variability, correlations and uncertainty using a single model and is showcased in a painting task.

Non-parametric Imitation Learning of Robot Motor Skills

A novel kernel treatment of imitation learning is proposed, which endows the robot with imitative and adaptive capabilities and is capable of learning human skills associated with high-dimensional inputs.

Generalized Orientation Learning in Robot Task Space

This paper proposes an approach that is capable of learning multiple orientation trajectories and adapting learned orientation skills to new situations (e.g., via-point and end-point), where both orientation and angular velocity are addressed.

Learning Skills from Demonstrations: A Trend from Motion Primitives to Experience Abstraction

Different motion and behaviour learning methods ranging from Movement Primitives to Experience Abstraction, applied to different robotic tasks are studied and scrutinized and then experimentally benchmarked by reconstructing a standard pick-n-place task.

Toward Orientation Learning and Adaptation in Cartesian Space

This article proposes an approach that is capable of learning multiple orientation trajectories and adapting learned orientation skills to new situations, where both orientation and angular velocity are considered, and introduces a kernelized treatment to alleviate explicit basis functions when learning orientations.

Adapting Object-Centric Probabilistic Movement Primitives with Residual Reinforcement Learning

This work proposes to combine ProMPs with the Residual Reinforcement Learning (RRL) framework, to account for both, corrections in position and orientation during task execution, and learns a residual on top of a nominal ProMP trajectory with Soft Actor-Critic and incorporates the variability in the demonstrations as a decision variable to reduce the search space for RRL.

A Linearly Constrained Nonparametric Framework for Imitation Learning

This paper proposes linearly constrained kernelized movement primitives (LC-KMP) to tackle the problem of imitation learning with linear constraints, and proposes to exploit the probabilistic properties of multiple demonstrations, and subsequently incorporate them into a linearly constraints optimization problem, which finally leads to a non-parametric solution.

A Unified Formulation of Geometry-aware Dynamic Movement Primitives

A novel and mathematically principled framework that uses concepts from Riemannian geometry to allow DMPs to properly embed geometric constrains is proposed that has been extensively evaluated both on simulated data and real robot experiments.

A Framework of Robot Skill Learning From Complex and Long-Horizon Tasks

This work proposes an entirely GP-based skill-learning framework that enables robots to achieve complex and long-horizon manipulation tasks in dynamic and unstructured environments and evaluated and compared the effectiveness and strengths of the framework with existing probabilistic methods on a Kinova robot that performed emergency button-pressing tasks.

Learning and Extrapolation of Robotic Skills using Task-Parameterized Equation Learner Networks

...

References

SHOWING 1-10 OF 49 REFERENCES

Task-level imitation learning using variance-based movement optimization

An imitation learning framework is presented, which allows the robot to learn the important elements of an observed movement task by application of probabilistic encoding with Gaussian Mixture Models and shows that the proposed system is suitable for transferring information from a human demonstrator to the robot.

Generalized Task-Parameterized Movement Primitives

This paper provides an alternative way to refine and adapt previously learned robot skills, which allows us to work on a low dimensional space and several examples are studied in simulated and real robotic systems, showing the applicability of this approach.

On Learning, Representing, and Generalizing a Task in a Humanoid Robot

We present a programming-by-demonstration framework for generically extracting the relevant features of a given task and for addressing the problem of generalizing the acquired knowledge to different

Probabilistic model-based imitation learning

This work proposes to learn a probabilistic model of the system, which is exploited for mental rehearsal of the current controller by making predictions about future trajectories, and learns a robot-specific controller that directly matches robot trajectories with observed ones.

Generalized Task-Parameterized Skill Learning

A novel learning perspective is provided which allows the robot to refine and adapt previously learned skills in a low dimensional space and is applicable in both simulated and real robotic systems.

Learning and generalization of motor skills by learning from demonstration

A general approach for learning robotic motor skills from human demonstration is provided and how this framework extends to the control of gripper orientation and finger position and the feasibility of this approach is demonstrated.

Reinforcement Learning to Adjust Robot Movements to New Situations

This paper describes how to learn such mappings from circumstances to meta-parameters using reinforcement learning, and uses a kernelized version of the reward-weighted regression to do so.

Policy search for motor primitives in robotics

A novel EM-inspired algorithm for policy learning that is particularly well-suited for dynamical system motor primitives is introduced and applied in the context of motor learning and can learn a complex Ball-in-a-Cup task on a real Barrett WAM™ robot arm.

Using probabilistic movement primitives in robotics

A stochastic feedback controller is derived that reproduces the encoded variability of the movement and the coupling of the degrees of freedom of the robot by using a probabilistic representation.

Demonstration based trajectory optimization for generalizable robot motions

A unified method to generalize robot motions to different workspaces, using a novel formulation of trajectory optimization that explicitly incorporates human demonstrations, and to locally adapt and reuse the optimized solution in the form of a distribution of trajectories is proposed.