This page is located in archive. Go to the latest version of this course pages. Go the latest version of this page.

Semestral work - 2D Barbie localization


Purpose of the semestral work is to create an environment in which the students will employ the knowledge gained and the implementations created during the first seven labs. The task to be solved is as follows:

  • TurtleBot is deployed in an unknown environment with obstacles (see image below for details).
  • World frame is determined by x and y-axis which are parallel with playground boundaries as shown in the image below.
  • The starting position is always the left bottom corner (coordinates x0=50cm, y0=50cm) of the playground (denoted by the blue circle in the image). Initial robot orientation is up to the team members.
  • The task is to explore the environment in order to localise one Barbie doll as accurately as possible within the 2-minute time limit (see video for details).
  • Playground is rectangular arena (final size 335x415cm), which will contain several obstacles. Each obstacle has a minimum size 20cm (in all dimensions). Distance between obstacles will be at least two times the size of the TurtleBot (there should be enough space for safe manoeuvres). Using any other prior knowledge about the playground is prohibited.


  • Relative position of the barbie with respect to the world frame is published as geometry_msgs/Point.msgs on topic \barbie_position_final (if more than one message is sent, the last message delivered within the 2-minute interval is evaluated).
  • All messages published on this topic are recorded into bagfile, which will be automatically processed by the lab tutor.
  • The maximum number of points from the semestral work is 22. The points will be assigned upon the absolute localisation error (max 16) a short report (max 6) as follows:
  1. Let a be the last position published within 2min time limit on the \barbie_position_final topic.
  2. Let b be the ground truth 2D position of the barbie at the beginning of the attempt in the world frame.
  3. The localisation error is computed as Euclidean distance e = sqrt((a.x-b.x)^2+(a.y-b.y)^2). z-coordinate is ignored.
  4. Points for absolute localization error will be assigned according to the table below. Every collision with the obstacles or playground boundary will be penalized by -0.5 points.
  5. Each team has two attempts (first in the thirteenth week, second in the fourteenth week). At least one attempt is obligatory (no demonstration attempt ⇒ 0 points from the semestral work). To avoid overcrowding, each team can demonstrate its solution only in the lab time in which at least one team member belongs to.
  6. All codes running on the robot during attempts have to be uploaded into Brute system (missing codes or detected plagiarism ⇒ 0 points from the semestral work).
  7. Points for the report: Each team is assumed to submit a short report (max one A4 page including figures). The report should describe the original idea, which is believed that it provided a competitive advantage with respect to other teams. The maximum number of points from the report is 6.
  8. Optionally: Strongly motivated students can get 5 bonus points for demonstrating the ability to capture the barbie by the magnetic holder and transferring it back by at least 10 cm.
Localization error Points
<25cm 16
(25cm; 27cm> 15
(27cm; 29cm> 14
(29cm; 31cm> 13
(31cm; 33cm> 12
>=55cm 0

Reservation policy

Students are assumed to create 3-member teams. Each team can book TurtleBot (R01-R07) and/or Playground using the reservation system. The maximum number of 45min-slots for the reservation of a robot is 4. The maximum number of 45min-slots for the reservation of the playground is 1. Once a slot is expired, the team can book another slot. It is highly recommended to work on the semestral work as soon as possible since the demand for the slots at the end of the term will be high. Slots in which other subjects take place are already booked out, the other slots are available.


Semestral work has been implemented by lab tutors - special thanks go to Ondra Holesovsky for putting all codes together. Video shows an autonomous exploration of the unknown environment. The current map is visualized in Rviz environment, detected Barbie is shown as the purple circular marker.

courses/aro/tutorials/semestral_work.txt · Last modified: 2019/06/03 13:10 by zimmerk