Unsupervised real-time seed detection for precision agriculture seeding machine

In modern precision agriculture all phases and operations in croop growing and harvesting must be optimized in order to maximize crop yield. The seeding process plays a relevant role in achieving high standards yields, both in terms of quality and quantity. Ideally it is necessary to precisely deliver one and only one seed in predetermined equispaced positions. The goal is to reach this result in eachone of the parallel seeding rows where modern seeders deliver the seeds.  Seeds are separed and individually picked by a mechanic-pneumatic system in which the pressure comand must be suitably controlled to ensure proper performances of the seeder. To implement such feedback control, it is necessary to recognize, from images collected by a high-speed camera system, when seeds are correctly picked and separed. Such information is then used to control the seeder vacuum pressure. 

 ERC Sector:

  • PE7_1 Control engineering
  • PE7_10 Robotics

Keywords:

  • Machine control
  • Image analysis
  • Precision agriculture
  • Automatic control
  • Image processing

Research groups

Contact