Robot Behaviour: Design, Description, Analysis And Modelling

Robots have evolved impressively since the 3-D manipulator built by C.W. K- ward (1957), the two little electromechanical turtles Elmer and Elsie [Walter, 1950, Walter, 1951], and the ?rst mobile robots controlled by comp- ers, Shakey [Nilsson, 1984], CART [Moravec, 1979, Moravec, 1983], and - lare [Giralt et al., 1979]. Since then, we have seen industrial robot manipu- tors working in car factories, automatic guided vehicles moving heavy loads along pre-de?ned routes, human-remotely-operated robots neutralising bombs, and even semi-autonomous robots, like Sojourner, going to Mars and moving from one position to another commanded from Earth. Robots will go further and further in our society. However, there is still a kind of robot that has not completely taken off so far: autonomous robots. Autonomy depends upon working without human supervision for a considerable amount of time, taking independent decisions, adapting to new challenges in dynamic environments, interacting with other systems and humans, and so on. Research on autonomy is highly motivated by the expectations of having robots that can work with us and for us in everyday environments, assisting us at home or work, acting as servants and companions to help us in the execution of different tasks, so that we can have more spare time and a better quality of life.

"Robot Behaviour - Design, Description, Analysis and Modelling" is the successor to the first textbook published in this field: Scientific Methods in Mobile Robotics and introduces the emerging field of scientific methods in mobile robotics to a wider audience of advanced undergraduate students, postgraduate students as well as researchers in robotics, artificial intelligence, cognitive science, ethology, biology and psychology.

Raising some of the big questions that roboticists should be concerned about - how can (robot) behaviour be described quantitatively?, how can experimental results be compared meaningfully?, is it possible to design faithful and accurate robot simulators@, etc - , this book gives indicators on how these might be addressed, and looks at how behaviour - in this case the behaviour of autonomous mobile robots – can be measured quantitatively, analysed, compared, and modelled. Statistical methods are introduced to compare behaviours and to state statistically significant differences in behaviour. Transparent modelling techniques derived from system identification are presented to obtain models of robot behaviour in closed mathematical form.

With numerous exercises (and solutions) and examples from research, including how robot behaviours can be compared and evaluated, how robot platform-independent code (Robot Java) can be obtained, how to determine analytically which sensors of a robot are important and which are not, and how to obtain robot control code directly from observing a human demonstrator,without any traditional programming, this book will be of great value to researchers and graduate students in mobile robotics, behavioural science and cognitive science.