Robot learning from demonstration

By Aude Gemma Billard, R. Siegwart. 2004

826 citations recorded by the source corpus.

Subfields: Biomedical Engineering, 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 Biomedical Engineering

In Control and Systems Engineering

In Robotics & Control

Citations in the corpus

Observe and Look Further: Achieving Consistent Performance on Atari (2018, background): “Some early works in this area [1, 17] used expert demonstrations to ﬁnd a good initial policy before ﬁne-tuning it with RL.”
Accelerating Online Reinforcement Learning with Offline Datasets (2020, background): “This data can potentially come from demonstrations for the desired task [16, 2], suboptimal policies [6], demonstrations for related tasks [21], or even just random exploration in the environment.”
Implicit Behavioral Cloning (2021, background): “One route is by collecting on-policy data of the learned policy, 235 and potentially either labeling with rewards to perform on-policy reinforcement learning (RL) [52, 53, 54] 236 or labeling actions by an expert [2].”
A Minimalist Approach to Offline Reinforcement Learning (2021, methodology): “With the aim of accelerating reinforcement learning from examples (known as learning from demonstrations [Atkeson and Schaal, 1997]), BC has been used as a regularization for policy optimization with DDPG [Lillicrap et al., 2015, Nair et al., 2018, Goecks et al., 2020] and the natural policy…”
Overcoming Exploration in Reinforcement Learning with Demonstrations (2017, methodology): “Demonstrations have been used to accelerate learning on classical tasks such as cart-pole swing-up and balance [31].”
An Algorithmic Perspective on Imitation Learning (2018, background): “Atkeson [Schaal, 1997, Atkeson and Schaal, 1997]. Since then, learning robot motions has been a key domain of imitation learning. Recently, learning from human demonstrations has benefited from developments in deep neural networks. Recurrent neural networks such as long short-term memory (LSTM) netw”
Agnostic System Identification for Model-Based Reinforcement Learning (2012, positive): “DAgger generalizes the approach of Atkeson & Schaal (1997) and Abbeel & Ng (2005) so that we can use any no-regret algorithm to update the model, as well as any exploration distribution.”
Deep TAMER: Interactive Agent Shaping in High-Dimensional State Spaces (2017, neutral): “One area of research on learning from human interaction that has received a great deal of attention is focused on the specific problem of learning from demonstration (Schaal 1997; Argall et al. 2009; Hussein et al. 2017).”
A Divergence Minimization Perspective on Imitation Learning Methods (2019, methodology): “In many cases however, it is either too challenging to optimize a given reward (e.g. due to sparsity of signal), or it is simply impossible to design a reward function that captures the intricate details of desired outcomes [3, 4, 5, 6, 7, 8].”
SOLAR: Deep Structured Latent Representations for Model-Based Reinforcement Learning (2018, background): “Many of these prior methods learn global models of the system that are then used for planning, generating synthetic experience, or policy search [3, 33].”
Grounding Language in Play (2020, background): “Two of the most successful approaches are imitation learning (IL) (Atkeson and Schaal, 1997; Argall et al., 2009) and reinforcement learning (RL) (Kober et al., 2013; Levine et al., 2016; Kalashnikov et al., 2018; Haarnoja et al., 2018; Andrychowicz et al., 2020).”
On Learning, Representing, and Generalizing a Task in a Humanoid Robot (2007, background): “Recent work on PbD addresses these questions at different levels [4].”
Nonparametric Representation of Policies and Value Functions: A Trajectory-Based Approach (2002, methodology): “use this approach to compute a policy and value function [6].”
Gaussian process dynamic programming (2009, background): “Algorithms based on over-conﬁdent models can fail to yield good results due to model bias as reported by Atkeson and Santamar´ıa (1997) and Atkeson and Schaal (1997).”
Reinforcement Learning on Web Interfaces Using Workflow-Guided Exploration (2018, background): “A key difference between the web and traditional RL domains such as robotics (Atkeson & Schaal, 1997) or game-playing (Bellemare et al., 2013) is that the state space involves a mix of structured (e.g. HTML) and unstructured inputs (e.g. natural language and images).”
Integrating reinforcement learning with human demonstrations of varying ability (2011, background): “A naïvely learned policy will likely perform poorly in such areas [2].”
Imitating Latent Policies from Observation (2018, background): “This approach has an extensive breadth of applications, ranging from early successes in autonomous driving (Pomerleau, 1989), to applications in robotics (Schaal, 1997; Chernova & Thomaz, 2014) and software agents (e.”
"Why Did You Do That?" Explainable Intelligent Robots (2017, methodology): “Behavioural Cloning is similar to techniques such as learning from demonstration (Atkeson and Schaal 2016) and apprentice learning (Abbeel and Ng 2004).”
Social interactions in HRI: the robot view (2004, background): “…etc. Today, humanoid robots (or physics based simulations of them) can learn a specific physical skill by observing a human demonstration of it [12], can acquire a simple proto-language by engaging in imitative interactions with a human instructor [13], or can mimic a sequence of human…”