Collaborating Authors


CMU School of Computer Science

SIGBOVIK (Special Interest Group on Harry Qatar Bovik) is an annual multidisciplinary conference specializing in lesser-known areas of academic research such as Sticky-Finger Manipulation, Natural Intelligence, and Retroactive Data. SIGBOVIK 2022 is the sixteenth edition of this esteemed conference series, which was formed in 2007 to celebrate the inestimable and variegated work of Harry Queen-o'-the-prairie Bovik. We especially welcome the three neglected quadrants of research: joke realizations of joke ideas, joke realizations of serious ideas, and serious realizations of joke ideas. The proceedings are not yet available as the conference has not yet proceeded. They will maybe be available for purchase at the conference, or at least an IOU ("I Owe You"), i.e., we owe you one proceedings booklet.

Using pgfplots to make economic graphs in LaTeX


In this command, the frame colour is the colour which surrounds the edges of the coloured background. The background colour is the colour behind the text.

An Accelerated Approach to Decentralized Reinforcement Learning of the Ball-Dribbling Behavior

AAAI Conferences

In the context of soccer robotics, ball dribbling is a complex behavior where a robot player attempts to maneuver the ball in a very controlled way, while moving towards a desired target. To learn when and how to modify the robot’s velocity vector is a complex problem, hardly solvable in an effective way with methods based on identification of the system dynamics and/or kinematics and mathematical models. We propose a decentralized reinforcement learning strategy, where each component of the omnidirectional biped walk (𝑣𝑥,𝑣𝑦,𝑣𝜃) is learned in parallel with single-agents working in a multiagent task. Moreover, we propose an approach to accelerate the decentralized learning based on knowledge transfer from simple linear controllers. Obtained results are successful; with less human effort, and less required designer knowledge, the decentralized reinforcement learning scheme shows better performances than the current dribbling engine used by UChile Robotics Team in the SPL robot soccer competitions. The proposed decentralized rein- forcement learning scheme achieves asymptotic performance after 1500 episodes and can be accelerated up to 70% by using our approach to share actions.

Reinforcement Learning for the Soccer Dribbling Task Machine Learning

We propose a reinforcement learning solution to the \emph{soccer dribbling task}, a scenario in which a soccer agent has to go from the beginning to the end of a region keeping possession of the ball, as an adversary attempts to gain possession. While the adversary uses a stationary policy, the dribbler learns the best action to take at each decision point. After defining meaningful variables to represent the state space, and high-level macro-actions to incorporate domain knowledge, we describe our application of the reinforcement learning algorithm \emph{Sarsa} with CMAC for function approximation. Our experiments show that, after the training period, the dribbler is able to accomplish its task against a strong adversary around 58% of the time.

Johan Cruyff Was Right


The hallmark of this was position switching; fullbacks, midfielders, and forwards trading spots in vertical lines through the center or on either flank to minimize their own running. This positional flexibility worked in concert with a willingness to chase the ball and to attempt to win it back closer to the opponent's goal than anyone else had dared, hunting in packs and using a high offside line and a proactive goalkeeper to clean up the open space they left behind their defense. Watch World Cup footage from before and after 1974 and the difference is clear. Before the mid-1970s, players had time to pick out passes or to build momentum when dribbling. The Dutch denied them that time, and every team since has tried to find its own balance between pressing and retreating into a solid defensive formation.