Getting started

First things first, we had to take the time to find out all the equipment we would need, all the objectives we wanted to reach and therefore all the tasks we had to achieve so as to complete the project properly. Nevertheless we obviously can’t predict everything right now, but we think that the board below gives a rather accurate overview of the project in itself. This board represents our current objectives, so it might not be the last time you see it on this blog section.


Giving you more details about the project’s ideas and mechanism should be a good start. The basic idea of the project is to gather a flock of land units in real-time. This flock would have a leader and all the other land units would have to keep in formation. Now the interesting point: all the other robots following the leader would be “dumb” robot as we called them in the way that they do not have any sensor mounted on their vehicle. The question being was to find a solution in order to balance this lack of sensor, this absence of communication between those machines and the environment in which they progress.

This question brings us to the interesting part of the project: the use of a drone. Indeed, the drone we’re using has two cameras on board (on looking straight forward and the other one pointing at the ground). The camera looking down will do some pattern recognition, locate every vehicle and the movement of each land unit will be adjusted so as to keep the formation. Namely, the drone will have to stay above the flock so as a first idea, it will stay above the leader (an autonomous or remote-controlled unit).

So far, we established a list of all we need (that you will probably see in our incoming videos relating our tests) and split the work in different parts which would be:

  • Pattern recognition: how to match patterns, which patterns to use, which platform to use for the image analysis;
  • Communication : we need to create an interface using the Bluetooth for the computer-land units layer and WiFi for the computer-drone layer;
  • Algorithmic: which strategy would be the most suitable for the trajectory of the land units and the position of the drone;
  • Design: making the robots ergonomic, stable, easily movable (for example, should we go on a omniwheel architecture or not…).

Comments are closed.