Effective reinforcement learning for mobile robots

By William D. Smart, Leslie Pack Kaelbling. 2002

455 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Networks and Communications

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 Computer Networks and Communications

Citations in the corpus

Improving Robot Controller Transparency Through Autonomous Policy Explanation (2017, methodology): “demonstration or simulation have emerged as a reasonable alternative to manual encoding [37, 1].”
Linear Off-Policy Actor-Critic (2012, neutral): “…methods are able to, for example, learn about an optimal policy while executing an exploratory policy (Sutton & Barto, 1998), learn from demonstration (Smart & Kaelbling, 2002), and learn multiple tasks in parallel from a single sensori-motor interaction with an environment (Sutton et al., 2011).”
Bootstrapping intrinsically motivated learning with human demonstration (2011, background): “nd what it has observed with a human. Many prior works have given a human trainer control of the reinforcement learning reward [8], [9], provide advice [10], or tele-operate the agent during training [11]. However, the more dependent on the human the system, the more challenging learning from intera”
Teachable robots: Understanding human teaching behavior to build more effective robot learners (2008, methodology): “Prior works have shown this by allowing a trainer to influence action selection with domain-specific advice [8,21] or by directly controlling the agent’s actions during training [28].”
Fast Reinforcement Learning for Vision-guided Mobile Robots (2005, methodology): “Smart and Kaelbling addressed the practical issues of getting Q-learning to work on a real mobile robot [6].”
Socially guided intrinsic motivation for robot learning of motor skills (2013, background): “Prior works have also given a human trainer control of the reinforcement learning reward (Blumberg et al, 2002; Kaplan et al, 2002), provide advice (Clouse and Utgoff, 1992), or teleoperate the agent during training (Smart and Kaelbling, 2002).”
Risk-Sensitive Reinforcement Learning Applied to Control under Constraints (2005, methodology): “Nevertheless, RL algorithms were successfully applied to many diﬃcult problems with continuous state spaces and nonlinear dynamics (see e.g. Sutton & Barto, 1998; Crites & Barto, 1998; Smart & Kaelbling, 2002; Stephan, Debes, Gross, Wintrich, & Wintrich, 2001).”
Double Q-learning (2010, methodology): “Q-learning has been used to find solutions on many problems [7] [8] [9] and was an inspiration to similar algorithms, such as Delayed Q-learning [10], Phased Q-learning [11] and Fitted Q-iteration [12], to name some.”
A survey of robot learning from demonstration (2009, unknown): “Demonstration may be used to highlight interesting areas of the state space in domains with sparse rewards, for example using teleoperation to show the robot the reward states and thus eliminating long periods of initial exploration that acquire no reward feedback [75].”
Quantum Reinforcement Learning (2008, methodology): “Many specific improvements are also implemented to modify related RL methods in practice [7], [9], [10], [20], [21], [22].”
Learning grounded finite-state representations from unstructured demonstrations (2015, background): “, 2012), (inverse) reinforcement learning (Smart and Kaelbling, 2002; Abbeel and Ng, 2004; Ziebart et al., 2008; Konidaris et al., 2012), and behavior-based approaches (Nicolescu and Matarić, 2003).”
Exploration from Demonstration for Interactive Reinforcement Learning (2016, background): “Several papers have shown how the policy information in human demonstrations can be used for inverse optimal control [28, 2], for teaching [8], to seed an agent’s exploration [5, 38], and in some cases be used entirely in place of exploration [21, 33].”