Full Paper Reinforcement Learning for Imitating Constrained Reaching Movements

By Florent Guenter, Micha Hersch, Sylvain Calinon, Aude Gemma Billard

How researchers see this paper

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

Citations in the corpus

Reinforcement learning in robotics: A survey (2013, methodology): “This method has resulted in applications including T-ball batting (Peters and Schaal, 2008c,b), reaching tasks (Guenter et al., 2007; Bitzer et al., 2010), ball in a cup (Kober and Peters, 2009), flipping a light switch (Buchli et al., 2011), playing pool and manipulating a box (Pastor et al.,…”
Learning contact-rich manipulation skills with guided policy search (2015, methodology): “Previous policy search methods in robotics generally use one or more of the following techniques to achieve good performance and reasonable training times: the use of example demonstrations [15], [9], [16], the use of a prior controller that allows the policy to only set high level targets, such as ”
Learning dexterous manipulation for a soft robotic hand from human demonstrations (2016, methodology): “This has been applied to reaching motions [25] and dynamic manipulation tasks.”
Learning motor primitives for robotics (2009, background): “such as T-Ball batting [6] and constrained movement [9].”
Natural Actor-Critic (2003, background): “The current range of additional applications includes optimization of constrained reaching movements of humanoid robots [12], traffic-light system optimization [23], multi-agent system optimization [11,28], conditional random fields [27] and gait optimization in robot locomotion [26,17].”
Latent spaces for dynamic movement primitives (2009, methodology): “DMPs have also been used as compact representations of movements for reinforcement learning [20], [9].”
Robot motor skill coordination with EM-based Reinforcement Learning (2010, methodology): “Guenter et al explored in [7] the use of Gaussian Mixture Model (GMM) and Gaussian Mixture Regression (GMR) to respectively encode compactly a skill and reproduce a generalized version of it.”
Imitation and Reinforcement Learning (2010, background): “Imitation Learning by Weighted Regression Many movements can be learned [2], [11] or at least initialized [4], [7] using imitation learning.”
Task-Specific Generalization of Discrete and Periodic Dynamic Movement Primitives (2010, background): “speed of convergence, which is important for online applications of reinforcement learning and is an active topic of research [17], [18].”
Towards transferring skills to flexible surgical robots with programming by demonstration and reinforcement learning (2016, unknown): “The importance sampler [28-30] used in this work can be defined by: 極岫守賃 − 守珍岻R岫酵賃岻玉磁岫滋入岻極R岫酵賃岻玉磁岫滋入岻 = ∑ 岫守沈津鳥勅掴岫賃岻 − 守珍岻R岫酵沈津鳥勅掴岫賃岻岻庭賃退怠 ∑ R岫酵沈津鳥勅掴岫賃岻岻庭賃退怠 (24)”
Kernelized movement primitives (2017, unknown): “, 2005) was used to optimize the centers of GMM components (Guenter et al., 2007).”
Dynamical System Modulation for Robot Learning via Kinesthetic Demonstrations (2008, methodology): “Whereas [9] t ries to reproduce asingle joint angle trajectory , and [18] learns a ask specific velocity profile , here we learn a whole dynamical systemcapturing thecorrelations across multiple variables for a given task.”
Imitation Learning (2017, methodology): “In Guenter et al. [2007], GMR is used to train a robot on an object grasping task.”
Learning to select and generalize striking movements in robot table tennis (2012, methodology): “Here, learning from demonstrations can provide a good starting point for motor skill learning as it allows the efficient acquisition of single movement primitives (Guenter et al., 2007; Peters and Schaal, 2008b; Kober et al., 2008; Peters and Schaal, 2008a; Bitzer et al., 2010).”