Scalable Techniques from Nonparametric Statistics for Real Time Robot Learning

By Stefan Schaal, Christopher G. Atkeson, Sethu Vijayakumar. 2002

278 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Control and Systems Engineering

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Control and Systems Engineering

Citations in the corpus

Reinforcement learning in robotics: A survey (2013, background): “However, this form of a state dimensionality reduction severely limits the dynamic capabilities of the robot according to our experience (Schaal et al., 2002; Peters et al., 2010b).”
Deep Lagrangian Networks: Using Physics as Model Prior for Deep Learning (2019, methodology): “E.g., authors used Linear Regression (Schaal et al., 2002; Haruno et al., 2001), Gaussian Mixture Regression (Calinon et al., 2010; Khansari-Zadeh & Billard, 2011), Gaussian Process Regression (Kocijan et al., 2004; Nguyen-Tuong et al., 2009; Nguyen-Tuong & Peters, 2010), Support Vector Regression (”
Adapting human motion for the control of a humanoid robot (2002, methodology): “Earlier work with this robot has included a dynamics model for use in feedforward control [12].”
Reinforcement learning by reward-weighted regression for operational space control (2007, methodology): “A method of learning such a forward model is Locally Weighted Projection Regression (LWPR) ( Schaal et al., 2002 ).”
Model Learning with Local Gaussian Process Regression (2009, methodology): “The regression parameter θ̂k can be computed incrementally and online using the partial least squares method [1, 15].”
Creating a Reinforcement Learning Controller for Functional Electrical Stimulation of a Human Arm. (2008, methodology): “LWR performed best using a neighborhood containing the 20 nearest points, using a weighting scale parameter h (Schaal et al., 2002) on the order of 10 –5 .”
Learning to select and generalize striking movements in robot table tennis (2012, background): “While this problem has been recognized by the robotics community (Schaal et al., 2002), it is far from being solved.”
Towards Motor Skill Learning for Robotics (2007, methodology): “This approach was inspired by LWPR [2], which employs linear models.”
Learning Forward Models for the Operational Space Control of Redundant Robots (2010, methodology): “As described in Schaal et al (2002), lots of models have followed, such as lwpls which include partial least square to reduce input dimensionality or rfwr (Schaal and Atkeson, 1997) which transform the algorithm into an incremental regression method, avoiding to store data.”
Local Gaussian Process Regression for Real Time Online Model Learning (2008, methodology): “For complex robots such as humanoids or light-weight arms, it is often hard to model the system sufﬁciently well and, thus, modern regression methods offer a viable alternative [7,8].”
Robot Programming by Demonstration (2010, methodology): “The mapping between robot action and geometric features is learned by means of locally weighted regression (LWR) [2,22,14] and artificial neural networks (ANN) [4].”
On-line regression algorithms for learning mechanical models of robots: A survey (2011, methodology): “[90], it can be seen as combining the properties of lwpls and rfwr.”
Learning to Control in Operational Space (2008, methodology): “It allows the highly data-efficient computation of the regression problem solutions with the supervised learning method LWPR (Schaal et al. 2002 Vijayakumar et al. 2002, 2005).”