In this paper we describe our approach to robotic skill representation and a prototypical implementation of a programming-by-demonstration approach that allows users to generate skills and robot program primitives for later refinement and re-use. We intend to evaluate the applicability of this approach to high-level programming in a user study, which we also explain.
Machine learning is advancing the capabilities of collaborative and industrial robots. Without 3-D sensors or neural networks, robots are blind and one-dimensional. They're restricted to one repetitive task that's been preprogrammed with no ability to account for variables in their environment. This limits a robot's productivity potential. Now, with vision sensors and machine learning capabilities, collaborative and industrial robots are able to achieve far more than they ever could on their own.
One of the most persistent dangers is the inspection of equipment in the plant. Typically, human inspectors are responsible for checking the boilers, tanks, and other equipment in power plants and industrial facilities. The work is hot, dirty, and dangerous. Pittsburgh-based startup Gecko Robotics offers wall-climbing robots as a safer way to handle these equipment check-ups. Gecko Robotics co-founder Jake Loosararian claims that while power plant inspection-related deaths are not as well documented as they need to be, estimates are between 20-30 deaths per year.
Section is a key area of investigation for video game research 2 describes the large amount of background and related (Hendrikx et al. 2013; Togelius et al. 2011). PLG work, both for Angry Birds and adaptive level generation in can be extremely useful for increasing a game's length and general. Section 3 presents our proposed adaptive generation replayability, as it allows a large number of levels to be created method. Section 4 describes our conducted experiments and in a relatively short time. It is also possible to tailor the results. Sections 5 discusses what these results could mean generated levels towards specific user's playstyles, known as for both human players and agents, Section 6 concludes this adaptive level generation, which allows for a unique and personalised work and outlines future possibilities.
Frazier, Spencer John (University of Southern California) | Huang, Chao (University of Southern California) | Kraus, Sarit (University of Maryland and Bar Ilan University) | Chang, Yu-Han (University of Southern California) | Maheswaran, Rajiv (University of Southern California)
As location-based mobile games become more popular, movement becomes an integral part of game play. This provides an opportunity for the game to influence player behavior in the real world, potentially inducing more physical activity (and better health) through intelligent adaptation of the game mechanic. We describe the application of Markov Decision Processes (MDPs) to model the player's behavior in a custom-built location-based zombie fighting game. The game agent uses this model - a user specific optimal policy (USOP) - to adjust the game behavior to encourage as much game play as possible. Our experiments with human subjects showed that game play time was indeed increased over the control condition. We look at how games can be used to model user behavior and then unobtrusively effect agent-determined behavioral change.