I have developed "tennis elbow" from lugging this book around the past four weeks, but it is worth the pain, the effort, and the aspirin. It is also worth the (relatively speaking) bargain price. Including appendixes, this book contains 894 pages of text. The entire panorama of the neural sciences is surveyed and examined, and it is comprehensive in its scope, from genomes to social behaviors. The editors explicitly state that the book is designed as "an introductory text for students of biology, behavior, and medicine," but it is hard to imagine any audience, interested in any fragment of neuroscience at any level of sophistication, that would not enjoy this book. The editors have done a masterful job of weaving together the biologic, the behavioral, and the clinical sciences into a single tapestry in which everyone from the molecular biologist to the practicing psychiatrist can find and appreciate his or

Principles of Neural Science

Reinforcement learning offers to robotics a framework and set of tools for the design of sophisticated and hard-to-engineer behaviors. Conversely, the challenges of robotic problems provide both inspiration, impact, and validation for developments in reinforcement learning. The relationship between disciplines has sufficient promise to be likened to that between physics and mathematics. In this article, we attempt to strengthen the links between the two research communities by providing a survey of work in reinforcement learning for behavior generation in robots. We highlight both key challenges in robot reinforcement learning as well as notable successes. We discuss how contributions tamed the complexity of the domain and study the role of algorithms, representations, and prior knowledge in achieving these successes. As a result, a particular focus of our paper lies on the choice between model-based and model-free as well as between value-function-based and policy-search methods. By analyzing a simple problem in some detail we demonstrate how reinforcement learning approaches may be profitably applied, and we note throughout open questions and the tremendous potential for future research.

/pdf/reinforcement-learning-in-robotics-a-survey-4w4397qx9i.pdf

Reinforcement learning in robotics: A survey

Policy search methods can allow robots to learn control policies for a wide range of tasks, but practical applications of policy search often require hand-engineered components for perception, state estimation, and low-level control. In this paper, we aim to answer the following question: does training the perception and control systems jointly end-to-end provide better performance than training each component separately? To this end, we develop a method that can be used to learn policies that map raw image observations directly to torques at the robot's motors. The policies are represented by deep convolutional neural networks (CNNs) with 92,000 parameters, and are trained using a guided policy search method, which transforms policy search into supervised learning, with supervision provided by a simple trajectory-centric reinforcement learning method. We evaluate our method on a range of real-world manipulation tasks that require close coordination between vision and control, such as screwing a cap onto a bottle, and present simulated comparisons to a range of prior policy search methods.

/pdf/end-to-end-training-of-deep-visuomotor-policies-1vqd5uo50d.pdf

End-to-end training of deep visuomotor policies

https://users.dimi.uniud.it/~antonio.dangelo/Robotica/2012/helper/leggedRobot/Ijspeert08NN.pdf

2008 Special Issue: Central pattern generators for locomotion control in animals and robots: A review

Reinforcement Learning in Robotics: A Survey.

/pdf/dynamic-hebbian-learning-in-adaptive-frequency-oscillators-1nnwrz7qpp.pdf

Dynamic hebbian learning in adaptive frequency oscillators

We present a system of coupled nonlinear oscillators to be used as programmable central pattern generators, and apply it to control the locomotion of a humanoid robot. Central pattern generators are biological neural networks that can produce coordinated multidimensional rhythmic signals, under the control of simple input signals. They are found both in vertebrate and invertebrate animals for the control of locomotion. In this article, we present a novel system composed of coupled adaptive nonlinear oscillators that can learn arbitrary rhythmic signals in a supervised learning framework. Using adaptive rules implemented as differential equations, parameters such as intrinsic frequencies, amplitudes, and coupling weights are automatically adjusted to replicate a teaching signal. Once the teaching signal is removed, the trajectories remain embedded as the limit cycle of the dynamical system. An interesting aspect of this approach is that the learning is completely embedded into the dynamical system, and does not require external optimization algorithms. We use our system to encapsulate rhythmic trajectories for biped locomotion with a simulated humanoid robot, and demonstrate how it can be used to do online trajectory generation. The system can modulate the speed of locomotion, and even allow the reversal of direction (i.e. walking backwards). The integration of sensory feedback allows the online modulation of trajectories such as to increase the basin of stability of the gaits, and therefore the range of speeds that can be produced

/pdf/programmable-central-pattern-generators-an-application-to-afemwnoz89.pdf

Programmable central pattern generators: an application to biped locomotion control

The development of robotic cognition and the advancement of understanding of human cognition form two of the current greatest challenges in robotics and neuroscience, respectively. The RobotCub project aims to develop an embodied robotic child (iCub) with the physical (height 90 cm and mass less than 23 kg) and ultimately cognitive abilities of a 2.5-year-old human child. The iCub will be a freely available open system which can be used by scientists in all cognate disciplines from developmental psychology to epigenetic robotics to enhance understanding of cognitive systems through the study of cognitive development. The iCub will be open both in software, but more importantly in all aspects of the hardware and mechanical design. In this paper the design of the mechanisms and structures forming the basic 'body' of the iCub are described. The papers considers kinematic structures dynamic design criteria, actuator specification and selection, and detailed mechanical and electronic design. The paper conclude...

/pdf/icub-the-design-and-realization-of-an-open-humanoid-platform-1znv6e51w6.pdf

iCub: the design and realization of an open humanoid platform for cognitive and neuroscience research

Personal robots can only become widespread if they are capable of safely operating among humans. In uncertain and highly dynamic environments such as human households, robots need to be able to instantly adapt their behavior to unforseen events. In this paper, we propose a general framework to achieve very contact-reactive motions for robotic grasping and manipulation. Associating stereotypical movements to particular tasks enables our system to use previous sensor experiences as a predictive model for subsequent task executions. We use dynamical systems, named Dynamic Movement Primitives (DMPs), to learn goal-directed behaviors from demonstration. We exploit their dynamic properties by coupling them with the measured and predicted sensor traces. This feedback loop allows for online adaptation of the movement plan. Our system can create a rich set of possible motions that account for external perturbations and perception uncertainty to generate truly robust behaviors. As an example, we present an application to grasping with the WAM robot arm.

Online movement adaptation based on previous sensor experiences

Central pattern generators (CPGs) are becoming a popular model for the control of locomotion of legged robots. Biological CPGs are neural networks responsible for the generation of rhythmic movements, especially locomotion. In robotics, a systematic way of designing such CPGs as artificial neural networks or systems of coupled oscillators with sensory feedback inclusion is still missing. In this contribution, we present a way of designing CPGs with coupled oscillators in which we can independently control the ascending and descending phases of the oscillations (i.e. the swing and stance phases of the limbs). Using insights from dynamical system theory, we construct generic networks of oscillators able to generate several gaits under simple parameter changes. Then we introduce a systematic way of adding sensory feedback from touch sensors in the CPG such that the controller is strongly coupled with the mechanical system it controls. Finally we control three different simulated robots (iCub, Aibo and Ghostdog) using the same controller to show the effectiveness of the approach. Our simulations prove the importance of independent control of swing and stance duration. The strong mutual coupling between the CPG and the robot allows for more robust locomotion, even under non precise parameters and non-flat environment.

/pdf/pattern-generators-with-sensory-feedback-for-the-control-of-3438kkeh0i.pdf

Ludovic Righetti

Papers

Dynamic hebbian learning in adaptive frequency oscillators

Programmable central pattern generators: an application to biped locomotion control

iCub: the design and realization of an open humanoid platform for cognitive and neuroscience research

Online movement adaptation based on previous sensor experiences

Pattern generators with sensory feedback for the control of quadruped locomotion