3. Field mapping

General description

For this task, the robots are navigating autonomously. The robots shall detect weed patches represented by pink golf balls and obstacles represented by yellow tennis balls. Task 3 is conducted on the area used in task 2 (Picture 4). The map created in this task will be used in task 4. Up to ten obstacles may be placed in the field, either between rows or in the headland. Obstacles must not be passed regardless of whether the robot can do so without touching them. Up to ten weeds may be placed in the field. All weeds will be placed between rows.

The rules for entering the field, moving the robot, using remote controller etc. are the same as in task 1 and task 2.

Field conditions

As in task 2 random stones are placed along the path, to represent realistic field scenario where the robot should cope with holes etc. The stones are not exceeding the level of 35 mm from the average ground level in the neighbourhood. The stones may be pebbles (diameter <35 mm) laid in the ground and large rocks that push (max 35 mm) out from the ground, both are installed. In other words, the robot must have ground clearance of this amplitude at minimum, and the robot must be able to climb over obstacles of max 35 mm high. No maize plants are intentionally missing in the end of the rows. However, due to circumstances of previous runs by other robots, it is possible that some plants in the end of the rows are damaged.

The weeds are objects represented by pink golf balls randomly distributed between the rows in the soil so that only the upper half is visible. Robots may drive across or over them without a penalty. The weeds are located in a band 60 cm wide between the rows. No weeds are located within rows or on headlands.

Obstacles are represented by yellow tennis balls which will be placed randomly between rows and on the headland. Robots are not permitted to touch or pass the obstacles. Picture 3 shows an example.

Rules for robots

Each team has only one attempt. The maximum available time for the run is 5 minutes.

Points will be awarded for detecting weeds and obstacles and for recording their positions.  The positions must be given in a Cartesian coordinate system with its origin and orientation equal to the starting pose of the robot. The positions (the map) must be provided in a text file similar to the one in table 5. 

Teams can nominate whether they wish to indicate the detection of weeds and obstacles separately from the mapping of their locations. Once the nomination has been made then that method must be used for the task.  By using audible or visual signals it is possible to get points for the detection of weeds and obstacles even if a map is wrong or missing. 

There is no requirement for the robot to travel along every row, provided that all obstacles and weeds are detected, i.e. it is acceptable for example to have a robot with a high mounted camera which is capable of surveying two or three rows at a time.

A single robot navigates between the rows, as in tasks 1 and 2, giving an audible signal when it comes across each weed or obstacle to indicate that it has detected it at that location. The detection of a weed should be indicated by a two second signal and the detection of an obstacle should be indicated by a five second signal. A robot that is capable of surveying more than one row at a time must indicate the row in which it has detected the obstacle or weed.

A robot producing an acoustic signal without any reason will be regarded as a false positive. Failure to produce an acoustic signal when an obstacle or weed is encountered will be regarded as a false negative.

The robot should have some means of storing the locations of the weeds and obstacles as this information will be required to complete task 4.

Assessment

 Each correctly identified and located weed or obstacle (true positives) will be awarded according to the following: 
audible or visible signaling without correct mapping audible or visible signaling with correct mapping correct mapping without audible or visible signaling
4 point 6 points 6 points
 A correct mapping is given, if the recorded location is within a square meter, which is centered in the true location. So the tolerance for x, y is +/- 0.5m.  

Crop plant damage by the robot will result in a penalty of 2 points per plant. Passing obstacles will result in a penalty of 5 points per obstacle.

Manual intervention to move or adjust the robot will result in a penalty of 2 points for each time the robot is STOPPED.

Indicating the presence of a weed or obstacle when none is present in that location (false positives) will result in a penalty of 1 point per occurrence.

Failure to indicate the presence of a weed or obstacle when one is present (false negatives) will result in a penalty of 2 points per occurrence.

The Jury will register the number of true positives, false positives and false negatives.

The total travelled distance will not be assessed.

If a team completes the task in less than 5 minutes (excluding the 2 minutes allowed to produce a map), this time will be used to calculate a bonus factor = total points x 5 minutes / measured time.

The task completing teams will be ranked by the number of points as described above.

The three best teams will be rewarded.

Picture 3 – Possible locations of the weeds and obstacles for task 3 and task 4.
Picture 4 – This is the actual layout of the row pattern. So there are no straight lines this year.
X Y Kind
2.6 3.5 weed
3.8 3.5 weed
4.8 3.5 obstacle

Table 5 – Example for a map file, recorded in Task 3

Print Friendly, PDF & Email