Official rules
The official competition rules for 2024-2025 are available here.
Scenario
The objective of this challenge is to explore an unknown, difficult-to-access, potentially dangerous area, search for people that can be stationary or can wander in this area and guide them out to a safe zone. A typical use case involves investigating the basement of a collapsed building in the dark or in smoke, to locate trapped people and rescue them by guiding them to the exit.
Each team will have a fleet of 10 drones. Each drone will be equipped with communication functions, a laser rangefinder (lidar, measuring distance to obstacles), a semantic sensor (enabling it to determine the nature of an object without data processing) and GPS. It is also equipped with life points that diminish with each collision with the environment or other drones, leading to its destruction when it reaches zero.
The UAVs (Unmanned Aerial Vehicle) will have to manage the limited range of communications resources, collaborate with each other to acquire and share information to optimize exploration, and be able to handle sensor and communications failures and unforeseen events such as the loss of UAVs, to conduct this mission autonomously.
The challenge therefore focuses on multi-drone exploration, navigation and coordination, and does not address issues such as person or object recognition. All scientific approaches (planning, learning, bio-inspiration, etc.) for the proposed solutions are possible.
Intermediate evaluation 1
In this first scenario, a simple environment will be used to evaluate navigation capabilities. The environment will feature a few obstacles, a single fixed wounded person, a single drone and a GPS loss zone.
The score will be the time taken to return the wounded person to the rescue zone, evaluated with and without the GPS loss zone.
Possible solutions for this evaluation range from very simple solutions exploiting random displacements, to much more complex solutions using GPS loss compensation, mapping and planned exploration methods. All these solutions are accepted. The aim is to enable participants to assess the quality and need for improvement of their navigation functions.
Intermediate evaluation 2
In this second scenario, a simple environment will be used to evaluate coordination and communication capabilities with 10 drones, a large number of wounded people (fixed or moving) on a large, unobstructed map with areas of communication loss and drone loss.
The score will be the time taken to bring all wounded people back to the rescue zone, assessed with and without the areas of lost communication and drones.
Possible solutions for this evaluation range from very simple solutions exploiting random displacements, to much more complex solutions using drone coordination mechanisms, task allocation and robustness to communication loss. All solutions are accepted. The aim is to enable participants to assess the quality and need for improvement of their communication and coordination functions.
Final evaluation
The final assessment will be an average of several qualitative and quantitative criteria defined below. The evaluation will be carried out on maps unknown to the participants, grouping together all the difficulties in a complete scenario.
The criteria of this evaluation will be, in limited time:
- The percentage of people brought back to the rescue zone,
- The percentage of the map explored when all people are rescued or the time limit is reached,
- The percentage of life points of the drones that come back to the rescue zone at the end of the mission compared to the life points of drones at the beginning,
- The percentage of time remaining relative to the time limit when all people are rescued.
Finally, the quality of the presentation of the teams’ work to a jury will also be assessed.
See the official rules above for the exact definition of the criteria, and the code for their implementation.