This homework introduces students the material in the course. For the written component to the concept of human/robot interaction of this homework, students have to and learning on a real robot. The students manually calculate a posterior probability of program their AIBOs to play a guessing game the robot's position given a uniform prior distribution by which one player (either the human or the of robot poses in a grid world. AIBO) guesses a sequence of colored markers Mounting a Charging Station. Students use the that the other player (AIBO or human, respectively) object-detection code written in previous makes up ahead of time. The AIBO communicates homework assignments to find a colored bull'seye to the human by a predefined set of and tower beacon. These two landmarks allow the robot to compute the distance and orientation motions. When guessing the colored sequence, of a charging station. The robot needs the AIBO has to reason about the patterns of to search for and then climb onto the charging the colors as well as about the clues given to it station.

Utrecht is more than 1,300 years old and located in the center of the Netherlands, about 40 minutes by train from Amsterdam. School (EASSS 2005) for about 120 students, which was organized by Europe's coordination network for agent systems (AgentLink) and was as successful as previous summer schools in Utrecht, Saarbruecken, Prague, Barcelona, Bologna, and Liverpool. Overall, in the theory and practice of AAMAS 2005 had 778 academic and autonomous agents and multiagent industrial participants from 44 countries systems. AAMAS 2005 is the fourth on six continents. The main room of this can with some justification AAMAS 2005 was held on July building, in which the Treaty of claim to be one of the most active.

This article introduces Pyro, an open-source Python robotics toolkit for exploring topics in AI and robotics. We present key abstractions that allow Pyro controllers to run unchanged on a variety of real and simulated robots. We demonstrate Pyro's use in a set of curricular modules. We then describe how Pyro can provide a smooth transition for the student from symbolic agents to real-world robots, which significantly reduces the cost of learning to use robots. Finally we show how Pyro has been successfully integrated into existing AI and robotics courses.

Learning algorithms have enjoyed numerous successes in robotic control tasks. In problems with time-varying dynamics, online learning methods have also proved to be a powerful tool for automatically tracking and/or adapting to the changing circumstances. However, for safety-critical applications such as airplane flight, the adoption of these algorithms has been significantly hampered by their lack of safety, such as "stability," guarantees. Rather than trying to show difficult, a priori, stability guarantees for specific learning methods, in this paper we propose a method for "monitoring" the controllers suggested by the learning algorithm online, and rejecting controllers leading to instability. We prove that even if an arbitrary online learning method is used with our algorithm to control a linear dynamical system, the resulting system is stable.

Schema learning is a way to discover probabilistic, constructivist, predictive action models (schemas) from experience. It includes methods for finding and using hidden state to make predictions more accurate. We extend the original schema mechanism [1] to handle arbitrary discrete-valued sensors, improve the original learning criteria to handle POMDP domains, and better maintain hidden state by using schema predictions. These extensions show large improvement over the original schema mechanism in several rewardless POMDPs, and achieve very low prediction error in a difficult speech modeling task. Further, we compare extended schema learning to the recently introduced predictive state representations [2], and find their predictions of next-step action effects to be approximately equal in accuracy. This work lays the foundation for a schema-based system of integrated learning and planning.

We present a competitive analysis of Bayesian learning algorithms in the online learning setting and show that many simple Bayesian algorithms (such as Gaussian linear regression and Bayesian logistic regression) perform favorably when compared, in retrospect, to the single best model in the model class. The analysis does not assume that the Bayesian algorithms' modeling assumptions are "correct," and our bounds hold even if the data is adversarially chosen. For Gaussian linear regression (using logloss), our error bounds are comparable to the best bounds in the online learning literature, and we also provide a lower bound showing that Gaussian linear regression is optimal in a certain worst case sense. We also give bounds for some widely used maximum a posteriori (MAP) estimation algorithms, including regularized logistic regression.

Schema learning is a way to discover probabilistic, constructivist, predictive action models (schemas) from experience. It includes methods for finding and using hidden state to make predictions more accurate. We extend the original schema mechanism [1] to handle arbitrary discrete-valued sensors, improve the original learning criteria to handle POMDP domains, and better maintain hidden state by using schema predictions. These extensions show large improvement over the original schema mechanism in several rewardless POMDPs, and achieve very low prediction error in a difficult speech modeling task. Further, we compare extended schema learning to the recently introduced predictive state representations [2], and find their predictions of next-step action effects to be approximately equal in accuracy. This work lays the foundation for a schema-based system of integrated learning and planning.

Learning algorithms have enjoyed numerous successes in robotic control tasks. In problems with time-varying dynamics, online learning methods have also proved to be a powerful tool for automatically tracking and/or adapting to the changing circumstances. However, for safety-critical applications such as airplane flight, the adoption of these algorithms has been significantly hampered by their lack of safety, such as "stability," guarantees. Rather than trying to show difficult, a priori, stability guarantees for specific learning methods, in this paper we propose a method for "monitoring" the controllers suggested by the learning algorithm online, and rejecting controllers leading to instability. We prove that even if an arbitrary online learning method is used with our algorithm to control a linear dynamical system, the resulting system is stable.

We address the problem of learning a symmetric positive definite matrix. The central issue is to design parameter updates that preserve positive definiteness. Our updates are motivated with the von Neumann divergence. Rather than treating the most general case, we focus on two key applications that exemplify our methods: Online learning with a simple square loss and finding a symmetric positive definite matrix subject to symmetric linear constraints. The updates generalize the Exponentiated Gradient (EG) update and AdaBoost, respectively: the parameter is now a symmetric positive definite matrix of trace one instead of a probability vector (which in this context is a diagonal positive definite matrix with trace one). The generalized updates use matrix logarithms and exponentials to preserve positive definiteness. Most importantly, we show how the analysis of each algorithm generalizes to the non-diagonal case. We apply both new algorithms, called the Matrix Exponentiated Gradient (MEG) update and DefiniteBoost, to learn a kernel matrix from distance measurements.

We present a competitive analysis of Bayesian learning algorithms in the online learning setting and show that many simple Bayesian algorithms (such as Gaussian linear regression and Bayesian logistic regression) perform favorably when compared, in retrospect, to the single best model in the model class. The analysis does not assume that the Bayesian algorithms' modeling assumptions are "correct," and our bounds hold even if the data is adversarially chosen. For Gaussian linear regression (using logloss), our error bounds are comparable to the best bounds in the online learning literature, and we also provide a lower bound showing that Gaussian linear regression is optimal in a certain worst case sense. We also give bounds for some widely used maximum a posteriori (MAP) estimation algorithms, including regularized logistic regression.