Infants are experts at playing, with an amazing ability to generate novel structured behaviors in unstructured environments that lack clear extrinsic reward signals. We seek to mathematically formalize these abilities using a neural network that implements curiosity-driven intrinsic motivation. Using a simple but ecologically naturalistic simulated environment in which an agent can move and interact with objects it sees, we propose a "world-model" network that learns to predict the dynamic consequences of the agent's actions. Simultaneously, we train a separate explicit "self-model" that allows the agent to track the error map of its world-model. It then uses the self-model to adversarially challenge the developing world-model.
In nowadays commercial and information reach society, filtering strategies have to be combined respecting the availabilities of resources, and additionally the guaranties regarding the response time should be given. The essence of the presented solution is both in the encapsulation of many known searching algorithms inside separate filtering agents, and in the integration of response time aware coordination mechanisms into one manager agent. Experimental results show that the guaranty of always proving results within 100 seconds can be given without sacrificing a user satisfaction.
Human aware planning requires an agent to be aware of the intentions, capabilities and mental model of the human in the loop during its decision process.This can involve generating plans that are explicable to a human observer as well as the ability to provide explanations when such plans cannot be generated. This has led to the notion "multi-model planning'' which aim to incorporate effects of human expectation in the deliberative process of a planner — either in the form of explicable task planning or explanations produced thereof. In this paper, we bring these two concepts together and show how a planner can account for both these needs and achieve a trade-off during the plan generation process itself by means of a model-space search method MEGA.This in effect provides a comprehensive perspective of what it means for a decision making agent to be "human-aware" by bringing together existing principles of planning under the umbrella of a single plan generation process.We situate our discussion specifically keeping in mind the recent work on explicable planning and explanation generation, and illustrate these concepts in modified versions of two well known planning domains, as well as a demonstration on a robot involved in a typical search and reconnaissance task with an external supervisor.
For humans and automation to effectively collaborate and perform tasks, all participants need access to a common representation of potentially relevant situational information, or context. This article describes a general framework for building context-aware interactive intelligent systems that comprises three major functions: (1) capture human-system interactions and infer implicit context; (2) analyze and predict user intent and goals; and (3) provide effective augmentation or mitigation strategies to improve performance, such as delivering timely, personalized information and recommendations, adjusting levels of automation, or adapting visualizations. Our goal is to develop an approach that enables humans to interact with automation more intuitively and naturally that is reusable across domains by modeling context and algorithms at a higher-level of abstraction. We first provide an operational definition of context and discuss challenges and opportunities for exploiting context. We then describe our current work towards a general platform that supports developing context-aware applications in a variety of domains.
In this paper, we study if modeling can help discriminate actions which in turn can be used to select an appropriate behavior for a mobile robot. For human-human interaction, significant social and communicative information can be derived from interpersonal distance between two or more people. If Human-Robot Interaction reflects this human-human interaction property, then interpersonal distance between a human and a robot may contain similar social and communicative information. An effective robot's actions, including actions associated with interpersonal distance, must be suitable for a given social circumstance. Studying interpersonal distance between a robot and a human is very important for assistive robots. We use autonomously detected features to develop such an interpersonal model using Gaussian Mixture Model (GMM) and demonstrate that such a learned model can discriminate different human actions. The proposed approach can be used in a socially-aware planner to weight trajectories and select actions that are socially appropriate for a given social situation.