The competitions will be held according to 2013 Robocup rules. You can access the rules from here.
This year, we will have 3 technical challenges:
This technical challenge tests the ability of safe navigation in a dynamic environment. As it is shown in the picture, there are eight robots acting as moving obstacles, two stationary and six moving along a straight line. The exact positions and trajectories are depicted in the pictures.
1) Each team participates with two robots.
2) The participating robots must navigate among obstacles as it is illustrated in the picture.
3) A penalty mark would be given to each collision. (-1) point for a moving obstacle. (-2) points for the teammate robot and stationary obstacles.
4) The team would receive a penalty (-1), each time a robot goes out of the specified area.
5) Each harmful clash will result in a penalty of (-4) points. (The referee decides which collisions are harmful).
6) Each navigation cycle for each robot would be scored by (+3).
7) The robots move in opposite direction and obstacle movement are random on the straight lines.
8) Both participated robot have to complete at least one cycle, otherwise team would get no score.
9) The challenge time is limited to 2 minutes and it will be started by “Force Start” signal.
This challenge is designed based on Robocup TC decision about increasing field size in the near future. In addition, this challenge encourages teams to more dynamic play while avoiding useless conflicts.
Here is the 4-camera experimental branch of SSL-Vision that contains all necessary changes for calibrating a 4 camera field. You can check out a copy:
svn checkout https://ssl-vision.googlecode.com/svn/branches/4cam_experimental
Also, here is some initial instructions on how to use the new calibration procedure.
Points will be awarded for the demonstrating the following skills on the double-sized field:
Skill 1: Scoring a goal against a stopped defense consisting of centrally placed goalie and 2 defenders around defense area:
· Without pass and cooperation. (1 point),
· With at least one pass (2 points).
Skill 2: Scoring a goal against a set of 3 moving robots consist of a goalie and 2 defenders:
· Without pass and cooperation. (2 points)
· With at least one pass (4 points).
Skill 3: Scoring a goal against a set of 4 moving robots consist of a goalie, 2 defenders and an active attacker:
· Without pass and cooperation. (4 points)
· With at least one pass (8
Note: Only two goals are scored by skill 1-2 and there would be no limit on the number of goals scored by the 3rd skill.
· The Challenge will be held on a half Double-sized field (8050 mm x 6050 mm).
· Each team can employ up to 3 robots.
· The Challenge starts with “Stop” and “Force Start” commands.
· The Challenge starts from one of the specific points that are shown below. (Teams can choose among the points)
· All free kicks will start from the specific points with “Stop” and “Force Start” commands.
· In any other situation the challenge will continue according to the normal rules of the game.
· The total time of the challenge for each team is 5 minutes active time, or till the team withdraw with the scored goals.
In this technical challenge which is designed to improve Artificial Intelligence level in small-size teams, long-term prediction ability of teams will be evaluated through the following scenario:
Log of a real match is broadcasted by a Test –Predicting- Software (TPS) in the standard format of vision packets. TPS has been designed by the technical committee.
At some point of the match some data from the vision packet related to one team, say team B, will be filtered (meaning the related fields are empty). The participants will be required to predict the state of the team B, t seconds after filtration started. In the illustrated figure a stream of data is depicted which prediction of the red frame is the objective. As it is shown in the figure, the first half of the game log will be broadcasted completely. Just the same as a real match, there will be ten minutes for the half-time. In the second half of the match, at three pre-undetermined points the filtration which was explained above will be done. The results of prediction are assessed based on the similarity of the predicted position and the actual state of the robots at that particular frame in the game log.
- Participant’s prediction results must be sent to TPS in form of one vision packet. Notice that only 5 seconds is granted to teams in order to send their data to TPS.
- It will be announced to participants before the match begins which team (in the game log) is the target.
- Prediction horizon t is not pre-defined and different for each of three filtering processes. This horizon will be selected from 2 to 5 seconds.
- In the similarity scoring, robots ID are not important.
- To avoid conflict with the SSL-vision, packets are sent on the network through another port.