Author(s): Stephane Ross, Geoffrey J. Gordon, J. Andrew Bagnell
Venue: 14th International Conference on Artificial Intelligence and Statistics
Year Published: 2011
Keywords: neural networks, learning from demonstration, dynamical systems
Expert Opinion: Imitation learning is a very appealing approach to learning robot skills. This paper shows that the straightforward technique of 'behavioral cloning' - simply copying the expert demonstrations, is actually not a good idea in sequential tasks. The reason is due to an effect of accumulating errors - once the learning agent strays away from states seen in the demonstration, it's learned policy is no longer accurate, causing it to stray even further away from the demonstration. There beauty of the paper is in capturing this idea mathematically, using no regret theoretical framework, and suggesting a simple algorithmic solution to the problem. The method, dubbed Dataset Aggregation (DAgger), asks for additional expert actions *on states visited by the policy*. The idea of controlling the distribution shift between the expert and the learner has since been fundamental to robotic imitation learning, and has manifested in various other methods.

Author(s): Brian D. Ziebart, Andrew Maas, J.Andrew Bagnell, and Anind K. Dey
Venue: AAAI Conference on Artificial Intelligence
Year Published: 2008
Keywords: probabilistic models, learning from demonstration, reinforcement learning
Expert Opinion: This work is one of the first to connect probabilistic inference with robot policy learning. Maximum Entropy Inverse Reinforcement Learning poses the classical Inverse Reinforcement Learning problem, well-studied for several years before this work, as maximizing the likelihood of observing a state distributing given a noisily optimal agent w.r.t an unknown reward function. The inference method, model, and general principles not only inspired future IRL works (such as RelEnt-IRL, GP-IRL, and Guided Cost Learning), they also have been applied in Human Robot Interaction and general policy search algorithms.

Author(s): Alexandros Paraschos, Christian Daniel, Jan Peters, and Gerhard Neumann
Venue: Neural Information Processing Systems Conference (NeurIPS)
Year Published: 2013
Keywords: manipulation, probabilistic models, gaussians, planning, learning from demonstration
Expert Opinion: This and the following papers using ProMPs, because they provided a very nice formulation for representing probabilistic movement primitives. ProMPs have many advantages and I found them better than classical DMPs in many robotics applications, from gestures to whole-body manipulations.

Author(s): Dean A. Pomerleau
Venue: MITP
Year Published: 1989
Keywords: mobile robots, learning from demonstration, neural networks
Expert Opinion: This work was pioneering with respect to machine learning in robotics broadly, learning from demonstration specifically, and also autonomous driving. It applied a neural net to learn steering angles from examples of human driving (even online!), way back in 1989. By today's deep learning standards the net was tiny (5 hidden units) and the sensor input extremely limited (30x32 image pixels), but it worked... and at a time when robots rarely operated outside of the lab or factory, and machine learning was rarely deployed on real hardware. It is the first* example of using demonstration-based learning for high-stakes control, that required (comparatively) fast sampling (25Hz) and operated a large van at regular road speeds (20mph). The vehicle was a part of NavLab, which was the precursor to CMU's DARPA Grand and Urban Challenges entires in the early 2000's, and those challenges in turn played a big role in accelerating today's driverless car boom. * To my knowledge! ...also, there actually are two papers (and my words above mix the two): [first publication] D. Pomerleau. ALVINN: An Autonomous Land Vehicle in a Neural Network. In Advances in Neural Information Processing Systems, 1989. [real world driving results] D. Pomerleau. Efficient Training of Artificial Neural Networks for Autonomous Navigation. Neural Compuation, 3(1), 88-97, 1991.

Author(s): Marc Peter Deisenroth, Carl Edward Rasmussen
Venue: International Conference of Machine Learning
Year Published: 2011
Keywords: state estimation, reinforcement learning, probabilistic models, gaussians, dynamical systems, visual perception, policy gradients
Expert Opinion: One of the first papers to really take uncertainty seriously in the RL + robotics space. Probably the first paper to convince me that model-based RL is worthwhile to think about, even in hard-to-model robotics domains.

Author(s): Richard S. Sutton and Andrew G. Barto
Venue: Book
Year Published: 2018
Keywords: mobile robots, reinforcement learning, unsupervised learning, optimal control, genetic algorithms
Expert Opinion: Great introductory text book to the underpinnings of of a lot of the modern approaches in ML/RL for robotics.

Author(s): George Konidaris, Leslie Pack Kaelbling, Tomas Lozano-Perez
Venue: Journal of Artificial Intelligence Research
Year Published: 2018
Keywords: probabilistic models, planning
Expert Opinion: There exists a representational gap between the continuous sensorimotor world of a robot and the discrete symbols used by advanced AI planning methods. Many existing studies typically assume the existence of precoded planning symbols, and investigate how to learn the relations between these pre-coded symbols and continuous world of the robot. Few others argue that symbols should be formed in relation to the experience of agents through their sensorimotor experience. This paper presents a structured approach, which is built on Markov-decision process formalism, to discover symbolic abstract representations from low-level high-dimensional continuous sensorimotor experience. The learned symbols and rules can automatically and effectively expressed in PDDL, a canonical high-level planning domain language, enabling high-level planning with traditional off-the-shelf AI planners.

Author(s): Pierre-Yves Oudeyer, Frederic Kaplan, and Verena V. Hafner
Venue: IEEE Transactions on Evolutionary Computation (Volume 11, Issue 2)
Year Published: 2007
Keywords: reinforcement learning, evolution, neural networks
Expert Opinion: This work contributes to the general question of obtaining life-long learning robotic systems. Large body of the existing robot learning literature mostly focus on methods that enable the robots to learn particular pre-defined skills and achieve particular tasks. Life-long learning, on the other hand, requires the robots to learn skills and adapt to situations that were not (and cannot be) foreseen. Inspired from human development, intrinsic motivation is an important drive that guides the robots towards regions that can be most effectively and efficiently learned with the capabilities developed so far; exploiting metrics such as novelty, curiosity, diversity, etc. This paper, in particular, is a seminal study that exploits maximization of learning progress in a real robot that explores its continuous sensorimotor space. It nicely shows that the robot exhibits stage-like development, learning easy tasks first, and focusing to more complex problems later; progressively developing more advanced skills.

Author(s): Marc Peter Deisenroth, Gerhard Neumann, Jan Peters
Venue: Book
Year Published: 2013
Keywords: survey, reinforcement learning
Expert Opinion: A great unifying view on policy search

Author(s): Auke Jan Ijspeert, Jun Nakanishi, Heiko Hoffmann, Peter Pastor, Stefan Schaal
Venue: Neural Computation (Volume 25, Issue 2)
Year Published: 2013
Keywords: planning, learning from demonstration, dynamical systems, nonlinear systems
Expert Opinion: It provides a detailed explanation of derivation and usage of dynamic movement primitives, a learning by demonstration method that encodes motion with a set of weighted kernel function, which offers great means of further learning, modification etc., and has been extensively applied in robotics.

Author(s): Nate Kohl, Peter Stone
Venue: IEEE International Conference on Robotics and Automation (ICRA)
Year Published: 2004
Keywords: reinforcement learning, policy gradients, locomotion, legged robots
Expert Opinion: The work is practical in that it allowed the authors to improve the walking speed of Aibos, something essential to creating top-flight robocup players. The reason I adore this work and frequently cite it in my talks on machine learning is the fantastic way it allowed the robots to learn autonomously. In particular, for the Aibo robots to succeed in robocup, they need to be able to localize on the field based on their perception of provided markers. The authors enabled the robots to measure their own walking speed leveraging this capability. By marching a team of robots back and forth across the width of the pitch, experimenting with and evaluating different gaits each time, the robots were able to find movement patterns that surpassed hand-designed ones. It's a beautiful example of exploiting measurable quantities to drive learning---a key enabling technology for robot learning.

Author(s): Sergey Levine, Chelsea Finn, Trevor Darrell, Pieter Abbeel
Venue: Journal of Machine Learning Research
Year Published: 2016
Keywords: manipulation, probabilistic models, planning, locomotion, learning from demonstration, reinforcement learning, neural networks, visual perception
Expert Opinion: This is an excellent example of reinforcement learning applied to closed-loop visual control for challenging robotics tasks and a good example of the application of deep-learning to real-world robotics.

Author(s): Pieter Abbeel, Andrew Y. Ng
Venue: International Conference on Machine Learning
Year Published: 2004
Keywords: reinforcement learning, learning from demonstration
Expert Opinion: Provided a convincing demonstration of the usefulness of inverse reinforcement learning

Author(s): Marcin Andrychowicz, Filip Wolski, Alex Ray, Jonas Schneider, Rachel Fong, Peter Welinder, Bob McGrew, Josh Tobin, Pieter Abbeel, Wojciech Zaremba
Venue: Neural Information Processing Systems Conference (NeurIPS)
Year Published: 2018
Keywords: manipulation, humanoid robotics, reinforcement learning, neural networks
Expert Opinion: HER addresses the issue of sample inefficiency in DRL, especially for those problems with sparse and binary reward functions. It has become one of the most effective algorithms for learning problems with multiple goals which have the potential to solve many challenging manipulation tasks. The idea of "EVERY experience is a good experience for SOME task" is a powerful insight that succinctly reflects how we teach our children to be lifelong learners. We should teach our robots the same way.

Author(s): Ashutosh Saxena, Justin Driemeyer, Andrew Y. Ng
Venue: International Journal of Robotics Research
Year Published: 2008
Keywords: neural networks, dynamical systems, visual perception, learning from demonstration, manipulation, planning
Expert Opinion: One of the first papers using general visual features for grasping

Author(s): Sebastian Thrun, Wolfram Burgard, Dieter Fox
Venue: Book
Year Published: 2005
Keywords: probabilistic models
Expert Opinion: It laid out basis for robotics in uncertain real world.

Author(s): Peter Pastor, Heiko Hoffmann, Tamim Asfour, and Stefan Schaal
Venue: IEEE International Conference on Robotics and Automation (ICRA)
Year Published: 2009
Keywords: planning, learning from demonstration
Expert Opinion: Not the first DMP paper, but the most understandable and with fixes to some annoying problems with the original formulation. Incredibly simple idea, but that's the nice thing about it -- it is a great starting point for talking about what generalization means in policy learning and how a restricted policy representation with the right inductive bias can allow you to learn something meaningful from a single trajectory, as well as learn quickly from practice.

Author(s): Lerrel Pinto, Abhinav Gupta
Venue: IEEE International Conference on Robotics and Automation (ICRA)
Year Published: 2015
Keywords: manipulation, reinforcement learning, neural networks
Expert Opinion: This paper demonstrated that it's possible to have a robot interact in a self-supervised way with the environment in order to learn useful tasks, like grasping. By running a robot for a long period of time, it's possible to collect enough data to train policies using simple algorithms. This lead the way for a lot of follow up work from Google and others, and is likely an area where we'll see a lot of interest in the future.

Author(s): Auke Jan Ijspeert, Jun Nakanishi, Stefan Schaal
Venue: IEEE International Conference on Robotics and Automation (ICRA)
Year Published: 2002
Keywords: probabilistic models, nonlinear systems, dynamical systems, learning from demonstration, humanoid robotics
Expert Opinion: In this work, a robust and scaleable movement primitive learning approach is proposed. The key insight is the embedding of motion trajectories in a 2nd order dynamical system. Goal attractors enable the generalization to different targets and simplify the learning of the model parameters from rewards. Complex motion can be learned through least squares regression from demonstrations.

Author(s): Pieter Abbeel, Adam Coates and Andrew Y. Ng
Venue: International Journal of Robotics Research
Year Published: 2010
Keywords: learning from demonstration, optimal control, dynamical systems
Expert Opinion: This paper presents a beautiful and compelling demonstration of the strength of learning dynamical models and using optimal control to learn complex tasks on intrinsically unstable systems even if the learned models rather crude and the optimal controllers are based on linearization, both strong approximations of reality. Furthermore, it addresses the problem of learning from demonstrations and improving from such demonstrations to beat human performance. To the best of my knowledge, on of the first paper demonstrating the use of learning by demonstration, model learning and optimal control together to achieve acrobatic tasks.