czech english

ALife III

simulated robots looking for energy in a maze-like environment

Two robots are roaming a maze-like environment, looking for energy. Energy is provided by chargers. However, chargers are scattered all around the environment and it is not so easy for the robots to find them. Moreover, once used by a robot, a charger will be unavailable for a while. Hence, the robot will have to go away and look for another charger. A robot will die if it fails finding an available charger before it runs out of energy. Then, the remaining robot will be declared the winner of the match.


Alex — G-BAGS based controller
The heart of the controller is going to be a graph based map, made according to the G-BAGS (Graph-Based Adaptive Guiding System) described in more detail in Winkler's master theses. For now only example source code for user control of the robot is available along with some descriptions. Author Zbyněk Winkler.

Wolf — controller based on skills
Wolf is s simple controller based on skills. A skill is a simple behavior, that can be easily debugged. The original intention was to have a three-level-architecture, but currently it seems that two levels will be sufficient. The source code is available for download. Author Martin Dlouhý.

Quick Info

Screenshot
Screenshot
From July 7th 2002 until May 1st 2003 you can submit new controllers for the robots at the contest homepage. The controllers can be programmed in Java only. Java security system ensures that the binaries carry no cheating systems. The simulated robots are Kheperas. Hence each robot has a differential wheels basis with incremental encoders, eight infra-red sensors for distance measurement and a color camera (80x60px, 32bits) looking in front.
The scoring is based on a hall of fame. Once a new controller is submited to the web site, it is appended to the bottom. Every day one round is played. If n robots are presents in the hall of fame, n-1 matches will be played each round. The first match confronts the last robot (bottom rank in the hall of fame) to the robot of rank #n-1. If the last robot wins, the two robots swap their positions in the hall of fame. Otherwise, nothing is changed. The next match confronts the robot ranked #n-2 with robot ranked #n-1 and so on until we reach the top of the hall of fame.
This way a robot can theoretically climp up from the bottom to the top position within a single round. However, a robot can loose only one rank per round. This is to encourage new competitors to submit their robots and have a chance to climb up the hall of fame rapidly. A round is played every day during the contest.
It is always possible to introduce a new version of an existing robot controller, by simply uploading the versions of the .class files, erasing any previous ones. When a new version of a robot controller is introduced in the contest, its position in the hall of fame remains unchanged. The next matches are run using the new version.
You can find more information at the contest homepage. The information presented here is adopted from the official contest description. If you have further questions you can use list or you can ask us (in czech or english) through our feedback form.