Field Robot Event
Our team Eduro (Martin Dlouhý, Jiøí I¹a, Jan Roubíèek, Tomá¹ Roubíèek) participated on the Field Robot Event 2009 in Wageningen, Netherlands. FRE is geared towards robots with agricultural applications. The ultimate goal is a weed control in a maize field. Maize grows in such a way that the normal heavy agricultural machines cannot access it. Here come the robots, which can travel between the maize rows and perform the necessary tasks.
The competition tasks
The competition consists of 3+1 tasks. The results of the main three tasks, described below, are combined into a final ranking. The fourth task is a Free Style, where the teams may show whatever they consider worth showing.
Task 1: Navigation
The focus of this task lays on the intra-row navigation. The robots should travel as far as they can in a three-minute interval. Points are awarded for the distance and by a jury for a "smoothness of operation".
Task 2: Advanced navigation
Very similar to the first task. This time, however, at the end of a row the robots are not switching to the next row automatically, but they should follow a predefined plan (such as switching to the fourth row to the right, then to the second to the left, etc.). Points are awarded in the same manner as in the first task.
Task 3: Weed control
Green golf balls are placed on a water-sensitive stripe of a paper along the maize. The robots need to spray the balls and are given points for accuracy. And for the "smoothness" again.
Team plans and results
As newcomers to the competition, we wanted mainly to gather a knowledge of the competition flow and some data for a possible next year participation. Because of the team timing, we did not have a chance to meet a maize field before the competition. Our original "plan" was to travel one row reliably.
In the beginning, we wanted to compete with a new robot. For various technical and personal reasons it has not been finished in time. Hence we used a backup solution, the Explorer robot. It was the smallest robot in the Event.
Although Explorer has not been designed for such rough conditions (low clearance height), it performed extremely well. In the rainy weather of the competition it also shined being water-proof.
My insect-inspired vision processing turned out to be a bit slow on the robot's motherboard. Our robot was ready just for the competition and we had no time for the software development and testing. For these reasons we decided to move the robot really slowly. This disqualified us from a serious competition for a medal (the rules are oriented towards speed), but we are the only team which has not damaged a single plant during the competition. For this we have awarded ourselves a title The most maize-friendly robot in Europe :-) The robot has traveled one row in every task reliably, according to our plan, and with that we are very satisfied.
Our team Eduro ended up on the eighth place in the Task 1, the eighth place in the Task 2, the fourth place in the Task 3 and eighth altogether out of sixteen teams from across the whole Europe. The fourth place in the weed control is especially notable, as we did not have any sprinkler. It is a pity that we could not get images of a row with golf balls from the competition. This task took place with plastic plants inside a building due to the rainy weather. But next year the golf balls should by replaced by plastic plants anyway.
Personal opinions about the competition
First, let me say I consider this a great idea for a competition. It is one of the few competition which needs to be prepared some time before the event itself and I thank the organizers for that.
Having an experience from other international competitions I must note, though, that I consider the ranking system utterly broken. My objections start from the priorities, continue through the existence of the "smoothness points" and make a heavy finish in the ranking execution.
Broken priorities
What good is a fast robot, which damages the plants? What makes it a precision control, if I can carry the robot several meters?
Smoothness points
The robots have a measurable task (either in the distance traveled or in the spraying accuracy). Why does there need to be a jury to award some undefined points, which even make half of the ranking? Is this some kind of an arts discipline, such as the dance is, that we need a jury awarding points? I thought this is an engineering. During the competition I did not decipher a method used to award the smoothness points. I was even lead to think that it may have something to do with being or not being a team from one of the organizing countries. It may not have been intentional. I myself have a tendency to see the good numbers and ignore the wrong ones, when scrolling results of my experiments. This is another reason why the human factor has to be excluded from evaluation whenever possible.
Ranking execution
I strongly believe (I am pretty certain) that the final ranking is invalid. In the Task 2, the Eyesonic team failed to switch from the second to the third row according to the plan right in front of me. (I pick the case I am most certain of and do not claim it to be the only mishap.) The robot had to be fixed by a series of (penalized) touches performed by the robot leader over a distance of half of the row (they could hardly be considered a single multitouch). Then the leader gave up and let the robot switch to the next row (in the middle of the field, I remind you). The rows being roughly ten meters long and the robot travelling a completely wrong path from the middle of the third row, I do not see, where the assigned 62.3 meters come from. It has traveled no more than 25 meters along the prescribed path. Even if I accepted the assigned distance (which I shall not), then the number of touches used for the penalization is wrong. The counted four touches were performed just at the beginning of the third row only and there were more touches to follow. This team has won the Task 2 and placed second in total.
It makes me feel sad that such a nice competition has been spoiled by the ranking for me. Please, bear in mind, that we did not compete for a medal and there is now chance for us to get one even if the results were "fixed" somehow. It is just my sense of fairness that suffers.
And if you would like to verify the numbers I am tossing around, you have a very little chance to do so. The official table with results has been withdrawn from the website almost immediately after the competition. Maybe so that nobody could complain?
Gallery
The complete gallery, including videos is available here. I would like to highlight the Explorer's ride in the Task 1.