The world population is expected to reach 9 billion inhabitants by 2050. Producing food to feed such a large population will significantly increase the strain already in place on production environments and inputs, especially land and water. It furthermore proves to be difficult already now to find skilled and motivated workers and to keep the production costs low. To meet this important societal challenge, research and development in the agri-food sector is of utmost importance. One important direction is towards increasing automation and robotisation of the industry to improve the efficiency of its processes, reduce waste and costs, and improve food quality and safety.

The challenge of natural variation

Where some other industries, such as the automotive industry, are already highly robotized, the agri-food industry is still very labor intensive. This discrepancy is mainly due by the fact that agricultural robots need to deal with natural products in natural environments. The main challenges for robotics in this area lie with the variation in (a) products (between and within species), (b) environmental parameters (light, soil, humidity, seasonal factors etc.), and (c) tasks (pre-harvest maintenance, harvesting, and post-harvest processing). Although automation and robotic system exist for performing specific tasks on specific crops in specific environments, these are often not economically viable due to their limited application. We therefore believe that the main research challenge in agri-food robotics is to improve flexibility and robustness to deal with these variations.

With recent advances in horticulture and food production, agri-food is more and more being produced in semi-controlled environments. Examples include high-density orchards, where fruit literally grows in “fruit walls,” modern greenhouses, etc. Parts of the environments are controlled, such as the type of product, and cultivation parameters,  but on the other hand, there is always the previously-discussed variability of the product and the environment due to the inherent biological nature which is unstructured and uncontrolled. Semi-controlled environments allow the use of top-down model-based design methods, whereas uncontrolled environments need methods that are robust to the variability. The agri-food domain requires robotic research in both areas combined.

Application domains can range from more controlled, e.g., food inspection and packaging, to more uncontrolled, e.g., crop maintenance and harvesting in the open field, where the more controlled domains have potential to develop applications on short term, whereas more long-term fundamental research is required to handle less controlled situations.

Workshop’s aims

In this workshop, we aim to bring together IROS attendees dedicated to or interested in solving some of the more pressing agri-food sector problems, specifically towards improved robustness and flexibility in dealing with variability in products, environments, and tasks. Our goal is furthermore to increase awareness of the important research topics and challenges in agri-food robotics and to promote interaction between researchers from different backgrounds in order to propel research in this area.

Topics of interest

Topics of interest addressed by this workshop include, but are not limited to robotic research in the following application domains:

  • Crop maintenance
  • Crop inspection
  • Weed detection and removal
  • Indoor navigation (greenhouses)
  • Outdoor navigation (open field)
  • Crop thinning and pruning
  • Crop yield estimation
  • Disease detection
  • Harvesting
  • Food-quality inspection
  • Food grasping and manipulation
  • Packaging

We invite researchers working in different areas of robotics to contribute to the workshop

Support by IEEE RAS AgRA TC

This workshop is supported by the IEEE RAS Technical Committee on Agricultural Robotics and Automation (AgRA TC)