Skill learning and task outcome prediction for manipulation

By Peter Pastor, Mrinal Kalakrishnan, Sachin Chitta, Evangelos A. Theodorou, Stefan Schaal. 2011

235 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Biomedical Engineering, Control and Systems Engineering, Reinforcement Learning, Robotics & Control

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 Biomedical Engineering

In Control and Systems Engineering

In Reinforcement Learning

In Robotics & Control

Citations in the corpus

Data-efficient Deep Reinforcement Learning for Dexterous Manipulation (2017, background): “For the latter, kinesthetic teaching is widely used [26, 13, 25, 39].”
Reinforcement learning in robotics: A survey (2013, background): “, 2011), and playing pool and manipulating a box (Pastor et al., 2011).”
Learning contact-rich manipulation skills with guided policy search (2015, background): “While contact-rich manipulation tasks such as the ones considered in this work are less common, several prior methods have addressed opening doors and picking up objects [13], as well as using force sensors during manipulation [14].”
Learning robot in-hand manipulation with tactile features (2015, methodology): “These methods are driven by feedback from tactile sensing [12, 35], Cartesian- and joint-space coordinates [33, 34], or both [32].”
A Framework to Describe, Analyze and Generate Interactive Motor Behaviors (2012, background): “Finally, robot teach pendants where the human teacher directly moves the robot that records the motion to reproduce, or imitation learning [62] where the robot is moved according to data of human movement recorded in some other way, also correspond to an assistance scheme, because the robot is passi”
A Review of Robot Learning for Manipulation: Challenges, Representations, and Algorithms (2019, methodology): “The robot may also learn the nominal sensory signals as a regression problem [289, 189, 366].”
Movement primitives via optimization (2015, background): “(15): TD(z) = s + z(g− s) + fD(z) (14) T̂ (z) = ŝ + z(ĝ− ŝ) + fD(z) (15) The linear function from s to g is adapted to the new endpoints, becoming ŝ + z(ĝ− ŝ) (black trajectories in Fig.1(b)), and the deviation fD remains fixed (purple deviations in Fig.1(b)).”
Learning grounded finite-state representations from unstructured demonstrations (2015, background): “Much LfD research has focused on the case in which the robot learns a monolithic policy from a demonstration of a simple task with a well-defined beginning and end [Abbeel and Ng, 2004; Ijspeert et al., 2003; Pastor et al., 2011a].”
Acquiring visual servoing reaching and grasping skills using neural reinforcement learning (2013, methodology): “The time requirements make fully reactive control difficult, even while more recent works are nearly capable of planning in real-time [6] and may re-plan their target when failure seems likely [7].”
A Review of Tactile Information: Perception and Action Through Touch (2020, unknown): “robots then compare the predictive output of this model to the actual sensor values during the action execution [155], [160].”
Imitation Learning (2017, methodology): “This led to a series of work that utilizes the dynamic system in Ijspeert et al. [2002b] to learn from demonstrations and subsequently use reinforcement learning for self-improvement [Kober and Peters 2010, 2014; Buchli et al. 2011; Pastor et al. 2011].”
Learning objective functions for manipulation (2013, background): “Reinforcement learning has found wide application in the domain of robotic manipulation [2], [3].”
Recent Advances in Robot Learning from Demonstration (2020, background): “In an effort to leverage the benefits of LfD’s data efficiency, researchers have demonstrated that LfD can be combined with reinforcement learning in order to improve sample efficiency (189, 190, 191, 192, 193, 194, 195, 196).”
Learning force control policies for compliant manipulation (2011, background): “Acquisition of manipulation skills using reinforcement learning has been previously demonstrated [1], [2], [3], [4], [5].”
Catch & Carry (2019, methodology): “rect approach involves fitting the demonstrations to a parametric form and using RL to modulate the parameters of the fitted model [Guenter et al. 2007; Peters and Schaal 2008; Kober and Peters 2009; Pastor et al. 2011]. Slightly more indirect approaches consist of using the demonstrations to learn ”
Stable reinforcement learning with autoencoders for tactile and visual data (2016, background): “taskspecific controllers [5], anomaly detectors [13], trajectories in sensor space [11], or by adding attractors [14].”