The result of the competition is as follows:

1) Pál Szeiler and Milán Matók (100%)

3) Martin Sallai (99%)

4) Máté Poór (89%)

Congratulations for the students.

Dear MSc and BSc Students,  

It is our pleasure to announce the second Bosch-ELTE programming contest.  😊 This contest is open to every BSc and MSc students

Problem description 

When it comes to navigating a vehicle equipped with autonomous driving function, the ego location shall be determined first. The obvious approach would be to use the GPS, however sometimes it is not a feasible solution (e.g. weak signal) therefore other localization techniques shall be applied. 


The competition is sponsored by the Robert Bosch GmbH. The first, second, and third competitors are awarded: 

  1. First place: 150.000 HUF 
  1. Second place: 100.000 HUF 
  1. Third place: 50.000 HUF 

All applicants will receive a Bosch T-shirt. Students in the Top10 also get a test-driving opportunity. 

Your task 

Create an algorithm which is able to determine where your vehicle is located among predefined areas of Budapest. There are no restrictions regarding the language and tools, only the input and output interfaces are given. 


We will provide you the following training dataset with the following properties: 

  • Synchronized video and lidar recordings
    • 2 cameras with normal field of view (FoV) looking forward 
    • 2 cameras with wide FoV looking to the sides 
    • 1 LiDAR 
  • 17 different well separated locations around Budapest
    • At least five recordings for each location 
    • Recordings were created in different points in time but cut to approximately 100m long from the same start position. 
    • Recordings that belong to the same location might differ as the environment changes (traffic, weather, etc.) 


  • Your software will be given four segments from the same locations cut from unknown recordings, that were not part of the training set 
  • The segments will be given in random order, but the start points will be aligned to the original sequence cutting points (0, ¼, ½ , ¾) 
  • The running of your software will be supervised by the organizers in a classroom 
  • Your software shall give back the result in .csv format right after the processing of the current input is completed, then the organizers will give you the next input 
  • You must localize your vehicle in total 17[locations] × 100[m] / 25[m] = 60 times 
  • The highest score wins.


  • You can give a list of 3 location titles for each 25m of recordings 
  • The first element in the list shall contain the highest confidence result and worth 3 points 
  • The second element in the list shall contain the second highest confidence result and worth 2 points 
  • The third element in the list shall contain the lowest confidence result and worth 1 point 
  • Any incorrect answer is worth 0 point 
  • No answer is worth 0 point 
  • The three results shall be different, therefore one good result can increase your total score only once 
  • The maximum points that are achievable are 17 × 4 × 3 = 204 

Application deadline: 2022.12.16.   2022.12.23. (!!!extended!!!)


Description of the sensor kit can be downloaded, including the camera/LiDAR calibration data, from our webpage. However, calibration parameters can also be separately downloaded. 

The first version of the document for the evaluation can also be downloaded. If you have questions, please ask in the Teams Group or write us an email. If there are contradictions in the text, we will modify it. But we hope that now you can prepare the final implementation of your street localization method.

Contest data 

We have taken several recordings for developing and testing the students’ implementations. Compressed data files are available here. You can observe the change of nature as winter is closer and closer in Budapest. Remark that the final data will be recorded in December. 

Visit frequently this webpage to see when new data are available. 

Example data are contained by this folder. There are four turns, data are zipped in, X={1,2,3,4}. Remark that there are other four files, named as groundtruthX.csv, they contain the ‘ground truth’ sequence indices. As it is written in the evaluation document the orders are not the same for the turns.

MS Teams 

We have created a Teams group; you can join it by this URL. If you cannot join the Teams, write a letter to hajder(at) Novel information will be spread in the team, and you can also ask the organizers if you have any questions about the contest. 


Please email the source code of your application until 23th December 23:59 to hajder(at) If you send us the resulting cvs files for the published data, we can check the correctness of the file format of your result. We will write you a message if something is wrong.

After the deadline, significant modification is not allowed in the source code, we will check it in the Evaluation Day. More details about the evaluation conditions will send to the applicant in the first weak of January.

Contest Schedule

Webpage history 

Competition Starts, 1st and Second Data Set published 12th October 
3rd Data Set Announcement 21st October 
Extra Data Set Announcement 28th October
4th Data Set Announcement 4th November 
Documentation for Evaluation 22nd November
5th Data Set Announcement 23rd November
Last Data Set Announcement 2nd December 
Extra (another one after last) dataset14nd December 
Example files for submission20th December
Submission/Application Deadline 16th 23rd December 23:59
Evaluation Day & Ceremony at Bosch-Campus 11th January (2023)

  • 2022.10.12. Webpage started; proposal, sensor-kit documentation, sensor parameters and two sets of data uploaded. 
  • 2022.10.14. New information (schedule) added.
  • 2022.10.21. Third dataset uploaded. There are problems in Bosch1 and Bosch5 sequences, see the readme.docx file.
  • 2022.10.28. Extra dataset published. The new zip contains data only for sequences #1…#9.
  • 2022.11.04. Fifth dataset uploaded. It is full to the best of our knowledge.
  • 2022.11.07. Bosch1 sequence is corrected in
  • 2022.11.22. Document for Evaluation is available.
  • 2022.12.02. Last, seventh, dataset is uploaded. It will be completed in 2022.12.09. 2022.12.14.
  • 2022.12.14. An extra dataset ( uploaded. Submission deadline is extended to 23rd December.
  • 2022.12.20. Example files for evaluation are uploaded.