/media/entity-images/sima.json

Computer Vision Simulation

Remote challenge

These rules may not be the final ones, the organizers have the right to modify them as they see fit. If there are any changes we'll let you know at once.

-------------------------------------------------------------------------------------------------------------------------------------------------

Changelog

19.05.2021: 2. Evaluation, evaluation method slightly modified;

                    Challenge duration changed from 36 hours to 46

05.05.2021: 3. Scoring, judging criteria modified;

                    Added the themes of the challenge

26.04.2021: First publishing

-------------------------------------------------------------------------------------------------------------------------------------------------

  1. THE GOAL OF THE CHALLENGE
    1. The team/participant must solve a given task in the timespan of 46 hours;
    2. Using a simulator software, the team/participant must design a robot and write the code for it such that the task is solved.
       
  2. EVALUATION
    1. After the time expires, each team must make a repository on GitHub and upload their project there.
    2. Each team must join a call with the judges and present their work and answer questions if there are any.
    3. Participants are also required to record a 3-5 minute demo of their simulation and send it to the organizers prior to the judging process.
  3. SCORING
    The team will be judged based on several criteria. The maximum score for each criteria is 5 points.

The categories are explained in more detail below:

    1. Complexity is judged based on the level of design & code. If the robot, for example, is hard to construct or contains advanced solutions and components this gives more points. Smart exploitation of simple hardware and unusual but brilliant solutions also gives more points.
    2. Scalability is judged based on how well the project functions when changed in size or volume so that it meets the end users needs.
    3. Innovation is judged based on how innovative the robot concept is. The robot must be a good example of new thinking to score more points in this category.
    4. Completness is judged based on the state the project is in. Obviously, a finished project will score a higher amount of points than one that isn't complete.
    5. Performance is judged based on how well the robot performs the task at hand.

No objections shall be declared against the jury’s decisions.

The following are the themes tackled in this challenge:

  1. Intersection/Object Detection;
  2. 3D Object Detection;
  3. 3D Object Reconstruction;
  4. Under-Drivability Concept;
  5. Preview of the Road;
  6. Point Cloud Segmentation;
  7. Construction Site Detection;
  8. Monocular Camera Depth Estimation;
  9. Pedestrian intention Detection

Recommended simulators: CARLA, Gazebo, Webots