Zusammenfassung:
Swarm robotics tries to apply swarm intelligence known from social insects such as ants, termites and bees to robots: while each robot is relatively simple, intelligent behaviour emerges from local interactions among robots without any kind of centralized control. As these simple individuals are not aware of their environment, their goals or even themselves and cannot reason or plan their actions the tasks they can perform are rather limited. To extend the application areas, current research tries to extend swarm robotics with intelligent agents. Intelligent agents should extend simple reflexive robots to smart robots that are able to think and communicate in order to reach solutions depending on knowledge gained over a longer period of time. In this master thesis, we have explored the suitability of Lego Mindstorms for swarm robotics in general and later on for the incorporation of learning, reasoning and planning. We have evaluated the overall performance of Lego Mindstorms in precision of motors and sensors, durability of power supply, computational power and capabilities for communication and physical interaction. Based on this assessment we have defined and implemented some preliminary work for collective transportation such as single and distributed line-following and heading following. For collective transportation a swarm of robots should push an object too heavy or big for a single one according to a specific course, e.g. a black line on the floor. Since collective transportation requires further exploration of areas like distributed intelligence, we have focused on the sub-tasks mentioned above. A single line-follower thereby recognizes a black line a white surface and follows this line. Distributed line-following extends this principle to pushing an object by several robots along this line. In the last application of heading following, some robots try to align their heading according to a heading given by a line-follower. Overall, we appreciate Lego Mindstorms as an inexpensive platform powerful and extendible enough for prototyping new approaches in swarm robotics, even with the possibility for the inclusion of intelligent agents.
Download:
N/A