The joint project of St. Petersburg State University and St. Petersburg Phys & Math Lyceum 239 “Cyberphysical laboratory” has started in 2008. As a result of the project the technique of teaching the elements of the control theory at school has been developed. Using the simple devices on the basis of Lego Mindstorms NXT, students of elementary school have started to master a science currently accessible only to high school or university students. In the first part of the chapter some ideas and recent findings of the project are described. In the second part of the chapter the testbed for studying and testing group control algorithms is presented. The main components and the basic principles of the testbed are described. Algorithms for target interception combined with collision avoidance are discussed and two different solutions are given. A comparison between real-life experiments and computer modeling is included.
TopIntroduction
In this chapter some results of collaboration among undergraduate students of St. Petersburg State University and high school students of St. Petersburg Phys&Math Lyceum #239 are described. The joint project called “Cyberphysical Laboratory” started in September, 2008. Its first results were described in Filippov S. et al (2009, 2011, 2012). The first part of the chapter is a continuation of Filippov S. at all (2009, 2011, 2012). In the next sections some further results are exposed mainly focused on application of estimation and control methods for education purposes. Some particular projects conducted by small student groups are described.
There exist two approaches for training schoolchildren to the compilation of control algorithms by robot:
The first approach is applied much more often, but it doesn't uncover diversity of possibilities of control. Its typical representation is the relay regulator based on branching “if”. In an on-off regulator the regulating organ under the influence of a signal from the sensor can accept one or the other extreme position: either “open” or “closed”. Thus correction action on adjustable object can be only maximum or minimum.
The second approach is more difficult for understanding by schoolchildren. However it comprises a key to further development. The approach is based on the concepts of feedback and error-closing control explained by studying the simplest standard controllers – proportional and differential, (for advanced students also integral).
Having gained a good experience of using the first approach (Filippov S. et al. 2009) - we have passed to the second one and have developed a technique of teaching automatic control by elementary examples. Each of examples sequentially dares by means of a relay controller, and then by means of proportional one and its compositions with other controllers. It allows a pupil to feel advantages of the most simple mathematical methods in control and to concentrate attention to them.
Since mathematics has become one of basic elements in robotic technology learning in the lyceum #239, Robolab 2.9.4 for elementary school and RobotC for the high have been selected as programming environments. Standard Lego Mindstorms NXT environment (G-NXT language) doesn't possess explicitly expressed mathematical apparatus, therefore we don't use it. A key point of approach is also using the difference equations of controlled system (plant) well corresponding to the discrete nature of computer based control instead of differential equations Filippov S. et al (2011)..
In this chapter we demonstrate further extension of the approach by a number of examples. The second part of the chapter is dedicated to teaching and learning in the area of multi-agent control algorithms. Various problems concerned with multi-agent control can be found in mathematics, physics, biology. One of the reasons for such a growth of interest is the rate at which computationally more powerful and yet miniscule microcontrollers are developed. This allows for new uses of embedded systems, including large groups of mobile agents that can solve various vital tasks. Examples multi-agent applications in real life include groups of robots used to create a map of a collapsed mine (Centibots: The 100 Robots Project), groups of flying drones that gathers information on forest fires or oil spills (Amelin et al. 2011), Casbeer D.(2006), or groups of agents exploring terrain.
Recently, a large number of various tools for simulated and modeling of multi-agent control algorithms has appeared. However, results of such modeling may significantly differ from thok in real-life experiments. On the other hand real-life experiments often involve expensive machinery (factory tools, cars, various airborne vehicles, etc.). Moreover, there is not only a risk of wear, but also a possibility of some sort of incidents (for example, collisions) which lead to loss of the machinery.