ARLINGTON, Va.--With support from the Office of Naval Research (ONR), researchers at the Georgia Institute of Technology have created an artificial intelligence software program named Quixote to teach robots to read stories, learn acceptable behavior and understand successful ways to conduct themselves in diverse social situations. "For years, researchers have debated how to teach robots to act in ways that are appropriate, non-intrusive and trustworthy," said Marc Steinberg, an ONR program manager who oversees the research. "One important question is how to explain complex concepts such as policies, values or ethics to robots. Humans are really good at using narrative stories to make sense of the world and communicate to other people. This could one day be an effective way to interact with robots."
Robots assisting the disabled or elderly must perform complex manipulation tasks and must adapt to the home environment and preferences of their user. Learning from demonstration is a promising choice, that would allow the non-technical user to teach the robot different tasks. However, collecting demonstrations in the home environment of a disabled user is time consuming, disruptive to the comfort of the user, and presents safety challenges. It would be desirable to perform the demonstrations in a virtual environment. In this paper we describe a solution to the challenging problem of behavior transfer from virtual demonstration to a physical robot. The virtual demonstrations are used to train a deep neural network based controller, which is using a Long Short Term Memory (LSTM) recurrent neural network to generate trajectories. The training process uses a Mixture Density Network (MDN) to calculate an error signal suitable for the multimodal nature of demonstrations. The controller learned in the virtual environment is transferred to a physical robot (a Rethink Robotics Baxter). An off-the-shelf vision component is used to substitute for geometric knowledge available in the simulation and an inverse kinematics module is used to allow the Baxter to enact the trajectory. Our experimental studies validate the three contributions of the paper: (1) the controller learned from virtual demonstrations can be used to successfully perform the manipulation tasks on a physical robot, (2) the LSTM+MDN architectural choice outperforms other choices, such as the use of feedforward networks and mean-squared error based training signals and (3) allowing imperfect demonstrations in the training set also allows the controller to learn how to correct its manipulation mistakes.
Sequential decision making problems, such as structured prediction, robotic control, and game playing, require a combination of planning policies and generalisation of those plans. In this paper, we present Expert Iteration (ExIt), a novel reinforcement learning algorithm which decomposes the problem into separate planning and generalisation tasks. Planning new policies is performed by tree search, while a deep neural network generalises those plans. Subsequently, tree search is improved by using the neural network policy to guide search, increasing the strength of new plans. In contrast, standard deep Reinforcement Learning algorithms rely on a neural network not only to generalise plans, but to discover them too. We show that ExIt outperforms REINFORCE for training a neural network to play the board game Hex, and our final tree search agent, trained tabula rasa, defeats MoHex 1.0, the most recent Olympiad Champion player to be publicly released.
In this paper we present a broad overview of the last 40 years of research on cognitive architectures. Although the number of existing architectures is nearing several hundred, most of the existing surveys do not reflect this growth and focus on a handful of well-established architectures. Thus, in this survey we wanted to shift the focus towards a more inclusive and high-level overview of the research on cognitive architectures. Our final set of 84 architectures includes 49 that are still actively developed, and borrow from a diverse set of disciplines, spanning areas from psychoanalysis to neuroscience. To keep the length of this paper within reasonable limits we discuss only the core cognitive abilities, such as perception, attention mechanisms, action selection, memory, learning and reasoning. In order to assess the breadth of practical applications of cognitive architectures we gathered information on over 900 practical projects implemented using the cognitive architectures in our list. We use various visualization techniques to highlight overall trends in the development of the field. In addition to summarizing the current state-of-the-art in the cognitive architecture research, this survey describes a variety of methods and ideas that have been tried and their relative success in modeling human cognitive abilities, as well as which aspects of cognitive behavior need more research with respect to their mechanistic counterparts and thus can further inform how cognitive science might progress.
Artificial Intelligence (AI) is the science and engineering of making intelligent machines, and the art of using computers to understand human intelligence. When it comes to healthcare, AI can bring about radical changes to the way doctors and physicians access, communicate, and interact with patient data. Crucially, this can result in improved patient outcomes and even the possibility of using an individual patient's history - such as previous illnesses, allergies or even genetic makeup - to predict their exact treatment requirements. Associate Medical Copywriter Kulveer Singh explains how the introduction of AI can change the landscape of healthcare, why it's an exciting time for the development of new technologies, and the challenges that implementing new AI may pose. In 2011, IBM's supercomputer WATSON competed against and obliterated a long-term game show champion.