Return to Christian Ferrareis's personal page

This project is focused on agricultural robotics. The main goal is to build a robot capable of following crop rows autonomously. In particular, we decided to modify an existing robot (Otto) so as to make it suitable for this purpose. The first modification consisted in replacing the tyres of the traction wheels with off-road ones, this has been done in a previous work. The modifications foreseen in this work consist in replacing the rear wheel with a bigger one, in order to reduce the risk of blockage against even small lumps of ground. This includes elevating its pivot point.

The robot will use computer vision algorithms to identify and track crop rows as it moves through the field, so as to ensure that the robot is always on track.

In addition to identifying the crop rows, the robot will also need to be capable of changing rows after it has completed its work on the current one. It will then need to identify the outer crop rows in order to plan the most efficient and secure path to achieve the transitioning operation.

In particular, the work will start by leveraging data from the ZED stereo camera available in the "Rosario" dataset to obtain a depth map of its view, and therefore a point cloud of the scene. The point clouds are then registered so as to align them in the same coordinate system and create a 3D reconstruction of the whole crop row. The result is then used to both navigate through the crop rows and plan the process of changing rows.