Learning objective functions for manipulation

By Mrinal Kalakrishnan, Peter Pastor, Ludovic Righetti, Stefan Schaal. 2013

156 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Control and Systems Engineering, 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 Control and Systems Engineering

In Robotics & Control

Citations in the corpus

Data-efficient Deep Reinforcement Learning for Dexterous Manipulation (2017, methodology): “Finally, the use of demonstration data has played an important role in robot learning, both as a means to obtain suitable cost functions [2, 14, 4, 8] but also to bootstrap and thus speed up learning.”
Learning from Physical Human Corrections, One Feature at a Time (2018, methodology): “A natural solution for restricting the number of learned features might be to switch the regularization term in (5) to the L1 norm [13, 15], which encourages sparsity of the weight update.”
Guided Cost Learning: Deep Inverse Optimal Control via Policy Optimization (2016, methodology): “Prior sample-based IOC methods use a linear representation of the cost function, which simpliﬁes the corresponding cost learning problem (Boularias et al., 2011; Kalakrishnan et al., 2013).”
Third-Person Imitation Learning (2017, background): “…weighted combination of features (Abbeel & Ng, 2004; Ratliff et al., 2006; Ramachandran & Amir, 2007; Ziebart et al., 2008; Boularias et al., 2011; Kalakrishnan et al., 2013; Doerr et al., 2015), or could also involve feature learning (Ratliff et al., 2007; Levine et al., 2011; Wulfmeier et al.,…”
AVID: Learning Multi-Stage Tasks via Pixel-Level Translation of Human Videos (2019, methodology): “Prior methods for robotic imitation learning typically have assumed access to observations and actions that are given directly on the robot, rather than learning from human videos [7]–[10].”
An Introduction to Deep Reinforcement Learning (2018, unknown): “Note that recent works bypass the requirement of the knowledge of the system dynamics (Boularias et al., 2011; Kalakrishnan et al., 2013; Finn et al., 2016b) by using techniques based on the principle of maximum causal entropy (Ziebart, 2010).”
Deep Mean Field Games for Learning Optimal Behavior Policy of Large Populations (2017, methodology): “This is related to a path integral IRL formulation, where the likelihood of measured optimal trajectories is evaluated only using trajectories generated from their local neighborhood, rather than uniformly over the whole trajectory space (Kalakrishnan et al. , 2013).”
Learning Deep Mean Field Games for Modeling Large Population Behavior (2017, methodology): “This is related to a path integral IRL formulation, where the likelihood of measured optimal trajectories is evaluated only using trajectories generated from their local neighborhood, rather than uniformly over the whole trajectory space (Kalakrishnan et al., 2013).”
Correcting Robot Plans with Natural Language Feedback (2022, methodology): “This is related to Inverse Reinforcement Learning (IRL) [1] that learns to recover reward functions from demonstrations, and has been successfully applied to infer objectives for manipulation motion planners [20].”
A Survey of Imitation Learning: Algorithms, Recent Developments, and Challenges (2023, unknown): “Notable contributions include the works of Aghasadeghi and Bretl [61] and Kalakrishnan et al. [62], who introduced path integral methods to enhance the scalability and versatility of MaxEntIRL.”
A Survey of Inverse Reinforcement Learning: Challenges, Methods and Progress (2018, methodology): “A prominent group of methods in this regard are path integral [59] based approaches [60, 44].”
Few-Shot Goal Inference for Visuomotor Learning and Planning (2018, background): “Other works have sought to learn objectives by training a classiﬁer [12, 13, 14] or reward function [10, 24, 11], including in the framework of inverse RL [25, 7, 8, 26, 27, 9, 28, 29, 30].”
Imitation from Observation: Learning to Imitate Behaviors from Raw Video via Context Translation (2017, methodology): “policy by running standard reinforcement learning [18], [19], though some more recent IRL methods alternate between steps of forward and inverse RL [20], [21], [22], [8].”
Socially compliant mobile robot navigation via inverse reinforcement learning (2016, methodology): “To this end, Boularias et al. (2011) apply importance sampling to compute the gradient of the model parameters while Vernaza and Bagnell (2012) constrained the features to have a certain low-dimensional structure.”
Recent Advances in Robot Learning from Demonstration (2020, methodology): “To circumvent this issue, LfD has been successfully applied to trajectory optimization-based methods by assuming that the expert minimizes a hidden cost function when demonstrating a skill (96, 110, 111, 112).”
Learning manipulation skills from a single demonstration (2018, methodology): “Kalakrishnan et al. (2013) use inverse optimal control on manipulation skills.”
Adversarial Imitation via Variational Inverse Reinforcement Learning (2018, methodology): “…entropy (MaxEnt) IRL is a widely used approach towards IRL, and has been extended to use non-linear function approximators such as neural networks in scenarios with unknown dynamics by leveraging sampling-based techniques (Boularias et al., 2011; Finn et al., 2016b; Kalakrishnan et al., 2013).”