A tutorial on task-parameterized movement learning and retrieval

By Sylvain Calinon. 2015

477 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, 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

1. Sergey Levine
2. Pieter Abbeel
3. Jan Peters
4. Chelsea Finn
5. J. Bagnell
6. Gerhard Neumann
7. Byron Boots
8. Jens Kober

In Computer Vision and Pattern Recognition

1. Jan Peters
2. Byron Boots
3. Sergey Levine
4. J. Bagnell
5. Gerhard Neumann
6. Sonia Chernova
7. Pieter Abbeel
8. Jens Kober

In Control and Systems Engineering

1. Sergey Levine
2. Jan Peters
3. J. Bagnell
4. Pieter Abbeel
5. Gerhard Neumann
6. Jens Kober
7. Chelsea Finn
8. Sonia Chernova

Citations in the corpus

• Benchmark for Skill Learning from Demonstration: Impact of User Experience, Task Complexity, and Start Configuration on Performance (2019, methodology): “Based on the choice of motion generation strategy, the majority of skill learning approaches can be further broadly categorized into four groups: time-invariant (stable) dynamical systems [2], [3], [13], time-dependent dynamical systems [7], [9], [10], time-dependent statistical methods [8], [14], [”
• An Algorithmic Perspective on Imitation Learning (2018, methodology): “The approach based on TP-GMM has been recently employed in several studies [Calinon, 2016, Rozo et al., 2016].”
• Generalized Cylinders for Learning, Reproduction, Generalization, and Refinement of Robot Skills (2017, methodology): “For GMM/GMR we tuned the number of Gaussian components (5 and 10, in figures (c) and (d), respectively), and for DMPs, we tuned the number of attractors (5 and 10, in figures (e) and (f), respectively).”
• Trajectory-Based Skill Learning Using Generalized Cylinders (2018, methodology): “To improve its generalization capabilities, a task parameterized extension of GMM/GMR was developed that assigns reference frames to task-related objects and landmarks (Calinon, 2016).”
• Towards Robust Skill Generalization: Unifying Learning from Demonstration and Motion Planning (2017, background): “Task-parameterized GMM/GMRs [8] generalize better by assigning reference frames to relevant objects and landmarks.”
• Learning Interactively to Resolve Ambiguity in Reference Frame Selection (2020, methodology): “In [17] and [18], a Task Parametrized GMM is used for describing the set of demonstrations respect each of the reference frames.”
• Kernelized movement primitives (2017, unknown): “In order to address this problem, we extend KMP by using local coordinate systems and affine transformations as in Calinon (2016); Huang et al. (2018), which allows KMP to exhibit better extrapolation capabilities (Section 3.3).”