Current state of agrifood robotics: three opportunities
20 September 2024
In May this year, the book ‘Advances in agri-food robotics‘ was published. In 732 pages, several authors, including various Wageningen researchers, describe the progress that agrifood robotics has made in smart ag in recent years and whether robotics is crucial to feeding 9 billion mouths in the future. Together with Dr Yael Edan from Ben Gurion University in Israel, Prof. Eldert van Henten from Wageningen University & Research edited the book. What does he think are the main advances that have been made?
Size of book speaks volumes about progress robotics in farming
“I feel that the fact that progress has been made is reflected in the size of the book and the many authors from around the world who contributed to it. And even then – partly due to a shortage of skilled authors – we had to leave out chapters in the post-harvest automation of fruit, vegetables and meat. Nevertheless, the book gives a very good and also very diverse picture of the developments and advances in the various application areas. These are arable farming, fruit farming, horticulture and livestock farming. I notice that the essential issues relating to robotisation are now being addressed throughout the chain and that there are also huge developments. Personally, I find it very interesting to see that when you look at the required technologies, there are many similarities between the different application areas. We should make more use of that. In my view, everyone should read the chapters on the current status of robotics in application domains. Developments in the last five years have really accelerated.”
Sense, plan, act
The status and development of technology is described in three phases/steps: sense (observe/detect), plan (plan) and act (implement). “In my opinion, these are the three key and sequential steps and we describe some very interesting examples of each. For example, contributions concerning observation/perception with cameras and other technologies. Personally, I also find the chapter on soft robotics with gripping (implement) fragile products like fruit very interesting. And there is also a chapter on planning. What we do see is that once observing/detecting seems to work, planning an action can still be a big challenge. In the case of picking soft fruits, for example, you want to grip them without damaging them. Which is tricky. In that respect, the integral design of a robotised system is quite a challenge and there are still plenty of opportunities for engineering and system design.”
From fully robotic harvesting to 85% robotic harvesting
Asked whether all the sectors are ready to embrace robotic solutions, Van Henten says “we’re not there yet”. “In horticulture, I think that everything that can be mechanised and robotised relatively easily has already been done. However, it is interesting to analyse where people are still needed to perform tasks. These are apparently tasks and/or operations that are so complicated that robotics is not yet a serious alternative. That is partly due to the occlusion phenomenon. Freely translated, this is the hindrance that the vision part experiences from leaves or branches on a plant to detect a ripe fruit, for example. In this respect, automated work in a natural environment with differing variables is much more challenging than a robotised industrial environment where all the variables are known in advance. Furthermore, unlike humans, most robotic solutions are still highly crop-specific. But change is coming. For example, I notice that the original objective of fully robotic harvesting of ripe vine tomatoes is being abandoned. Having 85 percent harvested by a robot is now also being seen as a very valuable step. Which is a very interesting turnaround! Because it means that humans will continue to pick some of the harvest, creating a need for collaborative robots or cobots, which is something I personally see a great future in. However, the challenge here is in adopting such cobots and embedding them in a business model.”
Three opportunities
As far as Van Henten is concerned, three clear opportunities can be observed for researchers and developers. “I’ve just mentioned the first which is the use of intermediate solutions where humans and technology work together. In the form of cobots but also in the form of augmented reality, for example. The second opportunity I see is to further develop and create a more integral design for the sense-plan-act cycle and incorporate feedback into it. Humans have very fast and effective feedback mechanisms. Something that we don’t yet see much of in arable farming with autonomous vehicles, for example. That’s because the safety and monitoring role of a tractor driver also needs to be automated for a fully autonomous system. The third opportunity is to make production systems less complex and modify them. As has happened in dairy farming with the selection of cows whose udder shape better suits a milking robot. But there are also opportunities in horticulture and fruit growing to adapt cultivation systems, thereby reducing the challenges for new robotic technologies.”
“One of the most interesting aspects of all this is that at Wageningen University & Research we have everything we need to help accelerate those developments. Not only do we have a global leadership position in the field of autonomy and robotics, we also link the university to research. Where the university stops at technology readiness levels (TRL) 4 to 6, the research colleagues take it further from there. We are therefore very good at making the connection between education, research associate and market!”
Photo header: fruit harvesting robot PPS Handsfree production in agrifood (LWV 19178)
