Projects

PhD student: Julia Winkler
Supervisors: Leo Marcelis, Ep Heuvelink
WP Task: 1.1 Plant architecture for automated crop handling and optimal production
Use case: Horticulture

Short Project Description:
Interest in automation for greenhouse horticulture production is increasing due to rising labour costs. One area of focus is rmarioobotic harvesting and pruning of high-value crops such as cucumbers. Although, setbacks to commercial adoption of these systems exist including occlusion of the fruit. Leaf pruning and the addition of far-red (FR) (700-800 nm) to supplemental lighting are both known to affect canopy structure, and thus potentially impact fruit occlusion. Optimal leaf area for production is a topic of study for many crop species, leading to pruning of leaves of high-value crops for a target leaf area. The perception of increased FR radiation has been correlated with stem and petiole elongation. Additionally, the elongation of peduncles (flower petioles) has been mentioned as a method to improve the robotic harvestability of cucumbers; however, limited research on peduncles exists. This project will begin with a literature analysis of leaf production for optimal leaf area in major crop species. We will then assess the impact of leaf pruning and supplemental FR on plant architecture experimentally and in silico via incorporation into an existing Functional Structural Plant Model (FSPM). The model will be utilized to optimize these conditions and plant architecture (i.e., peduncles) for a reduction in fruit occlusion with minimal yield implications.

PhD student: Samikshya Shrestha
Supervisors: Leo Marcelis, Elias Kaiser, Silvère Vialet-Chabrand
WP Task: 1.2 Acclimation of plant processes
Use case: Horticulture

Short Project Description:
Plant leaves in nature and greenhouses experience fluctuating light (FL) as a significant part of their daily light exposure. Over timescales of hours to days, plants acclimate to their prevailing environment through adjustments in their physiology, morphology, and anatomy. Such adjustments enable the plant to maximize photosynthetic efficiency and are crucial for fitness in a given environment. Acclimation is not instantaneous; it can be assumed that plants which acclimate quickly to changes show improved light use efficiency under dynamic environmental conditions compared to individuals which acclimate slowly. However, little is known about acclimatory mechanisms, genotypic variation, and environmental modulation of acclimation under FL. The main objective of this study is to explore the two aspects of photosynthetic acclimation to FL to obtain a better understanding of the mechanisms that drive acclimatory responses. The first aspect involves intraspecific genotypic variation studies with the aims to identify genes through genome-wide association studies (WP1) and to identify trends in acclimation capacity due to breeding history (WP2). The second aspect involves exploring the acclimation to environmental conditions such as complex FL patterns (WP3) and combination of [CO2] levels with FL (WP4). The experiments will be conducted in climate chambers and Arabidopsis (WP1) and cucumber (WPs 2-4) will be used. Various growth, morphological, and physiological parameters, as affected by acclimation will be assessed.

PhD student: Margreet Edens
Supervisors: Jan Dijkstra, Myrthe Gilbert
WP Task: 1.3 Novel rumen fermentation models
Use case: Dairy

Short Project Description:

To improve feed resource efficiency and to reduce the environmental impact of dairy farming, improving feed efficiency has gained increasing interest. Recently, a breeding company (CRV, Arnhem, The Netherlands) started with routine recording of individual feed intake of cows on commercial dairy farms in The Netherlands. These measurements help to utilize individual variation between cows in feed efficiency. Research suggests that (part of) the variation between cows in feed efficiency could be explained by differences in digestibility between cows. However, inter-individual variation in feed digestibility measured at large cohorts of animals is scarce and limited knowledge is available on the relation between feed intake, feed intake pattern, feed digestibility and feed efficiency. During the last decades, improvements in nutrition together with genetic selection led to increasing milk yield and feed intake of cows. At the same time, cows’ ability to digest feed, in particular that of fibre has decreased, which is undesirable considering the role of a cow in a circular food system, where cows convert human-inedible fibre-rich diets into high quality human-edible animal products. Therefore, a high capacity to digest fibres should be (genetically) preserved, or if possible, improved. Hence, in close collaboration with a PhD candidate from the Animal Breeding and Genomics group of Wageningen University and Research (WP1.4), the current project aims to improve our fundamental understanding of (genetic) differences between cows in feed digestibility by collecting and analysing faecal samples of approximately 400 cows from which individual feed intake is recorded. Special attention will be given to nitrogen, as nitrogen excretion is one of the major issues in dairy farming nowadays, especially in The Netherlands. Furthermore, possibilities to use fast and effective methods to estimate digestibility and feed efficiency via spectroscopy measurements will be explored. Finally, possibilities to improve rumen fermentation models (using the obtained data) might be explored.

PhD student: Eugénie Guennoc
Supervisors: Henk Bovenhuis
WP Task: 1.4 Longitudinal phenotyping of cows
Use case: Dairy

Short Project Description:

Because of the economic importance of feed on dairy farms (~ 50% of the production costs), studies has been directed toward improving feed efficiency, and this trait is currently included in breeding objectives. Although it has been shown feed efficiency is partly explain by digestibility, it is still unknown how they are related. Moreover, there is little knowledge on digestibility from a genetic perspective as it has been studied only at small scale. This study aims to analyse feed digestibility on a larger scale (~ 450 cows from commercial farms) in collaboration with Animal Nutrition group (WP 1.4). Genetic parameters (heritability, genetic variance and correlation with other traits (e.g., dry matter intake; fertility)) will be estimated to (1) determine the effects of current selection on digestibility and to (2) assess the potential of selection for feed digestibility. In addition, a genome-wide association study (GWAS) will be carried out to provide information on the genetic architecture of feed digestibility. Eventually, a method for fast and on-farm estimation of feed digestibility will be investigated, using a portable near-infrared spectroscopy (NIRS) device.

PhD student: David Kottelenberg
Supervisors: Jochem Evers
WP Task: 1.5 Niche differentiation and weed suppression
Use case: Arable

Short Project Description:

One of the major benefits of intercropping is its ability to decrease weeds. In this research, we aim to understand the mechanisms of weed suppression in different intercrop systems through a combination of field experiments and functional-structural plant modelling. We aim to understand why intercrops experience enhanced weed suppression and which traits are most responsible for this. Furthermore, we aim to understand how the design of the intercrop system influences its weed suppressive ability. Finally, through optimisation modelling we aim to combine this knowledge to find improved intercrop systems that have enhanced weed suppression and increased yield.

PhD student: Zohralyn Homulle
Supervisors: Bob Bouma
WP Task: 1.6 Disease suppression in crop mixtures
Use case: Arable

Short Project Description:

The aim of this research is to study the effect of intercropping on disease reduction, and to improve our understanding of the specific workings of the disease suppressive mechanisms at play. I will start with a meta-analysis on the effect of intercropping on plant diseases, in order to find generic patterns across pathosystems how intercropping can reduce diseases. Then I will use field experiments to focus on the effect of (strip)intercropping on potato late blight specifically, focussing on the extent to which different factors and mechanisms contribute to disease reduction in this pathosystem. Some of the mechanisms I will be looking at in the field experiments are host dilution, a barrier effect and changing microclimate. For the last part of the research, I plan to focus on ecological optimisation and technology for intercrop systems.

PhD student: Don van Elst
Supervisors: Andrea Fiore
WP Task: 2.1 Integrated NIR sensors
Use case: Horticulture

Short Project Description:

In this task we will develop, optimize and demonstrate a new generation of ultracompact near-infrared (NIR) spectral sensors for determining the ripeness of fruit and for monitoring the nutrients (e.g. nitrogen, phosphorus, calcium) in leaves. The sensors will be based on a technology under development at TU/e and will consist of arrays of photodetectors integrated with filters spanning the NIR range (1000-1650nm). The photodetector array is based on a InGaAs/InP heterostructure bonded to a Si wafer using a polymer layer (BCB). This InP-on-Si technology allows scaling up fabrication to large substrate sizes. This solution is completely integrated and extremely compact, with a single chip area of ≈2.25 mm^2 and no moving parts, making it perfectly suited for low-cost, portable spectral sensors. After development and optimization, we aim to demonstrate the functionality on applications relevant to other work packages.

PhD student: Yueyu Lin
Supervisors: Simona Cristescu
WP Task: 2.2 Infrared (IR) gas sensing
Use case: Dairy

Short Project Description:

Within Synergia project, we will develop and characterize an optical-based sensor for analyzing methane in breath of cows, next to other volatile organic compounds analyzed with mass spectrometry. We will also build a novel spectrometer using a broadband light source to monitor liquid biomarkers related to early detection of diseases in dairy cow and their health condition. These data will add to the golden standard biomarkers (i.e. in blood). In addition, we will evaluate the performances of the spectrometer for monitoring fast chemical reactions in liquid phase by using attenuated total reflectance (ATR).

PdEng: Tasfia Kabir
Supervisors: Martijn Heck / Kevin Williams
WP Task: 2.3 Integrated spatial imaging
Use case: Horticulture

Short Project Description: Within the Synergia project, we propose a new class of scanned-light-based detection and ranging using array-scalable, wavelength scanned and steered chips which allow light to be scattered off crops (leaves, stems) to monitor movement, growth, and changes in surface reflections. Nanostructured surface grating arrays will be used as stimulus and monitoring ports in combination with specially designed broadband tunable lasers and interferometers operating over a 50 nm wavelength range, centered within 1500-1600 nm. The design and measurement of the broadband tunable source is the main focus of the PdEng project. Wafer-scale processing technologies will be implemented to ensure a clear route to volume production with the project industrial partners. Compliance with optical sensor packaging technologies will all be ensured using surface-normal optical coupling.

PhD student: Maryem Tanveer
Supervisors: Piyush Kaul / Marion Matters
WP Task: 2.4 THz electronic spectroscopic imaging
Use case: Arable

Short Project Description:

Currently there is no real time sensor at subTHz/THz frequencies. that’s used in agriculture. Neither such a sensor has been designed specifically for material parameter detection/extraction including measuring water content of plants, although there are such electronics-based sensors in other application areas such as communications, and radars. The aim of this project is to identify techniques that allow simplified design of electronics subTHz/THz based sensor for material parameter detection/extraction in agricultural environment. Furthermore, the research would focus to explain a relationship between temporal quantification of water and material parameters using different modelling approaches and experiments. This would facilitate in quantifying parameters such as water potential and stress of different plant species in real time.

PhD student: Menno van Zutphen
Supervisors: Duarte Antunes
WP Task: 3.3 Collaborative decision making
Use case: Arable

Short Project Description:

This task focuses on the scalability of the decision support systems by investigating how several autonomous agents (e.g., drones and small robots) with independent processors can collaborate in order to make local decisions that minimize a certain cost (e.g., uncertainty in a crop/barn world model, shortage of nutrients, etc). The two main innovative aspects are providing (approximately) optimal methods for collaborative decision making on (i) crop-level interventions and (ii) system level observations/parameter- and state estimation/data collection with limitations on sensing resources. The developed methods will be used in WP4 when multiple agents (e.g. drones) need to actively sense the environment and update a shared world map to reduce uncertainty and make actuation decisions such as spraying in an efficient manner.

PhD student: Ioannis Panagopoulos (successor of Ashkan Sebghati)
Supervisors: Tamas Keviczky, Simon van Mourik
WP Task: 3.4 Data-driven stochastic decision-making, optimal control and planning
Use case: Horticulture

Short Project Description:

This project focuses on the development of a data-driven, crop-centric decision-making algorithm. The primary objective is to create a robust approach for planning and implementing climate control strategies within the context of autonomous greenhouse systems. This algorithm is intended to guide the decisions of future optimal climate conditions and the corresponding control inputs, with a focus on maintaining crop states within predefined robust safety boundaries. The proposed next-generation autonomous greenhouse controller should be founded upon a data-driven economic predictive control framework. To achieve this, the initial phase is to address the inherent challenges associated with modeling the intricate and nonlinear dynamics of greenhouse systems, particularly concerning the interaction between climate and crop behavior. Furthermore, this research has to tackle the complexities introduced by exogenous uncertainties and vulnerabilities stemming from the energy market. The control algorithm aims to effectively manage these challenges while simultaneously ensuring sustainable crop production.

PhD student: Maedeh Sadeghi (successor of Ricardo Castro)
Supervisors: Sjoerd Boersma
WP Task: 3.5 Risk-sensitive control methods
Use case: Dairy

Short Project Description:

Milk production faces sustainable challenges regards to the cost of feeding, methane emission and energy balance of dairy cows. Feed management has been shown the potential to address these challenges separately. Since these challenges are in close relationship with each other, we need to consider the impact of feed compositions on all the challenges at the same time. These challenges, however, are in trade off with each other, which leads to different optimal feed compositions based on the differences in farmer’s preferences towards these challenges. In this project, we aim to address these challenges by optimizing the diet of the cow to simultaneously increase feed margin, decrease methane emission and maintain energy needs of cows. To this aim, in this project, we first model physiology, energy and nutrient flows for an individual cow. Then, by implementing a multi-objective optimization, we find the optimal feed composition over the time based on different preferences. Since the energy and nutrient flow of cow change over the time (based on the changes in physiological status), it is important to optimize the diet for the whole life of the cow. Therefore, we need the automation of this optimization over the whole life of the cow based on its physiological stage (pregnancy, milking, dry off). For the automation, we implement a model predictive control (MPC) to optimize the diet.

PhD student: Maarten de Jong
Supervisors: Giulia Giordano
WP Task: 3.6 Structural control approach to nonlinear ecological network dynamics
Use case: Arable

Short Project Description:

Intercropping systems can be considered networked single-generation ecosystems, where crops affect each other’s growth by modifying the local environment. If we consider an interaction between crops x and y as (x,y), we can distinguish competition (-,-), amensalism (-,0). antagonism (-,+), commensalism (0,+) and mutualism (+,+). The combined effects of multiple pairwise interactions can result in unintuitive net effects. By constructing simple differential equation models of interactive crop growth that implement key mechanisms such as light interception, resource dilution and disease suppression, one can identify emergent system properties relevant to intercropping performance. These emergent properties are logical extensions to the model construction, and may serve to validate proposed hypotheses and models. Properties that are extremely robust to parameter variations, to the point where they persist for all meaningful choices of parameters are called ‘structural’, for instance, a particular equilibrium might be structurally stable or unstable, regardless of the choice of parameters. If the qualitative features of the model system are in agreement with observations, the model system can be optimized for values of interest such as yield.

PhD student: Ruben Beumer
Supervisors: Duarte Antunes
WP Task: 4.1 Structures and algorithms for updating the world model
Use case: Arable

Short Project Description:

In order to perform actuation tasks in the challenging environments of the use-cases, the robots need an adequate description (model) of the world they operate in. This world contains semantically rich objects (crop, leaf, etc.) enabling the robot to understand high-level instructions and carry out tasks. Maintaining a world model entails creating and updating links between the objects and measurements (vision, tactile sensors, etc.). This data association is highly nontrivial since object detection algorithms often miss and mis-detect objects. Providing models for the uncertainty in the world representation will be instrumental to plan and make decisions pertaining to the robot actuation and sensing tasks.

PhD student: Rick van Essen
Supervisors: Gert Kootstra
WP Task: 4.2 Next-best-view planning to optimize information gain (also 2.6)
Use case: Arable

Short Project Description:

Visual search is an important task for agricultural robots. Due to partial observability from a single viewpoint, lots of challenges in visual search tasks require a camera moving in the environment to localize the target object. Active sensing increases the efficiency by selecting a next viewpoint of the camera that is expected to reveal most information. By making use of spatial relations between objects in the environment, visual search tasks can be made more efficient. Although defining such search strategy can be done manually, these manually defined strategies are limited to the domain in which they are defined and cannot adapt easily to a changing environment. Learning such a search strategy is therefore preferred. A method to learn a search strategy is reinforcement learning (RL). Based on interaction with the environment, a RL agent can learn a search strategy. In this PhD, we propose four studies on visual search, gradually increasing the complexity and realism of the search task. The first study will be done in a simplified simulation of a Unmanned Aerial Vehicle (UAV) searching for weeds in a semi-2D environment. To reduce the training time, we aim to add prior knowledge in study 2. Study 3 deals with the reality gap between the simulation and the real world. To this end, we will transfer a search strategy from the simplified simulation to a realistic simulation. Finally, the effect of occlusion on the search strategy is evaluated in a greenhouse use case to localize sweet pepper fruits in a 3D environment. We believe this research will make visual search tasks by robots more efficient and thereby can improve automation in agriculture.

PhD student: Marissa Jonker
Supervisors: Wouter Hakvoort
WP Task: 4.3 Least tactile sensory data interaction control strategy
Use case: Horticulture

Short Project Description:
This research aims to design a system for automatic crop harvesting, where the gripper contains the least number of attached sensors possible while still allowing careful manipulation. We aim to use vision instead of physical force sensors.
Due to hygiene standards in the food industry/agriculture, we use grippers that are easy to clean and replace. Flexure-based grippers offer exactly this: as underactuated grippers they are inherently easy to clean, and since the production cost and effort are both low, easy to replace. To harvest delicate products as food with a high throughput, we desire a firm grasp, while preventing damaging the food products. Therefore, we require knowledge of the forces the gripper exerts on the object. However, to upkeep the cleanability and replaceability properties, we desire to not use force sensors for this, but to estimate the forces using vision. Additionally, we want to estimate object properties and slip/friction (essentially emulating tactile sensing) using this/a similar gripper, after which model reduction then should allow for relatively fast prediction and control.

PhD student: Robert van de Ven
Supervisors: Ali Leylavi Shoushtari
WP Task: 4.4 Adaptable two tentacle robotic arm
Use case: Horticulture

Short Project Description:

With this research we aim to increase the adaptability of agricultural robotics, and thereby increase the automation in agriculture. One of the main challenges for increasing the automatization in agriculture is the large variety of work done on farms. We will focus on issues related to performing the actions needed to do tomato harvesting, dealing with the handling of fragile tomatoes and the detaching of tomatoes. We will use learning-based control of robots to allow agricultural robotics to perform this task. We will focus on the variability in the task of tomato harvesting to identify the possibilities of applying learning-based robotics in agriculture. We will use learning based on observed expert data, Learning from Demonstration, as it allows the human to demonstrate the specific motions used for detaching tomatoes and thus reduces the time and data required for learning. The four studies vary in the use of sensory feedback in the learning-from-demonstration paradigm. The first study serves as a baseline, using only an observation of the pose of the target object at the start and focussing on learning of fast motions with high acceleration needed to detach fruits from the plant. In the second study, continuous visual observations (feedback) are included in the framework, allowing to better deal with variation and uncertainty in the pose of the object. The third study includes tactile sensory feedback, allowing to deal with variation in the shape and weight of the object. And the last study deals with force feedback as an integral part of the LfD paradigm, allowing to deal with variation in the connection strength.

PhD student: Filip Sunjic
Supervisors: Jan de Jong
WP Task: 4.5 Low-cost, low-maintenance, lightweight, dexterous arm for inspection
Use case: tbd

Short Project Description: The agricultural sector has been dealing with growing food demand, and it is projected that the demand will only increase in the following years as we are expected to reach a population of 9.7 billion people by the year 2050. At the same time, the labour shortage has caused disruptions and by some estimates, more than 10% of the fruits worldwide are left unharvested. As a possible solution to those problems, the Agricultural sector is looking to increase the automation level in order to meet the growing food demand. One aspect that poses a significant challenge is the automation of picking fruit. This project aims to create an end-effector with a gripper that can perform the task of fruit picking, that will achieve a high success rate of picking fruit, while at the same time minimizing the damage to the fruit and the tree. To achieve those goals, the end effector will utilize design principles from flexure mechanics and soft robotics.

PhD student: Lenn Gorissen
Supervisors: Kornelia Konrad
WP Task: 5.1 Anticipation of critical issues in transition pathways
Use case: all

Short Project Description:

The objective of my work is to study expectations and imaginaries and explore possible transition pathways related to the T4E-concept. I will investigate the implicit and explicit concepts and visions of the new forms of farming central to the Synergia project and identify challenges and tensions. These insights will feed into exploration of socio-technical scenarios and transition pathways, which will be conducted together with stakeholders and partners in the project. This will serve as a basis to deduce requirements and options for responsibly governing transformative change.

PhD student: Xuying (Marion) Leo
Supervisors: Arnout Fischer
WP Task: 5.3 Consumer acceptance
Use case: all

Short Project Description:

My work explores public and consumer perceptions of technology in ecological farming, focusing on acceptance or resistance toward technology for ecological farming. Widespread adoption of T4E solutions requires either acceptance or, at the very least, the absence of active opposition. In particular, we examine ambivalence as a potential driver of resistance. Ambivalence arises when individuals hold both positive and negative beliefs and emotions about the domains of technology, agriculture, and ecology. Understanding the root causes of this ambivalence is crucial for mitigating resistance and fostering successful innovation of T4E solutions.

PhD student: Patricia Roost
Supervisors: Ed Nijssen
WP Task: 5.4 & 5.5 Business models & Stimulating farmer adoption
Use case: all

Short Project Description:

Due to several macro-level trends, new precision agriculture technologies will increasingly be offered to customers in the form of integrated customer solutions with new contracting and pricing models. First, the development of these customer solutions will be investigated. Specifically, we will consider issues such as integration and customization, the involvement of customers, suppliers and ecosystem partners, and internal collaboration within solution provider firms. Secondly, to examine the business models accompanying these customer solutions, we will look into selling, pricing and contracting methods.

PostDoc: Monique Mul (successor of Fatima-Zahra Abou Eddahab)
Supervisors: Boelie Elzen
WP Task: WP 6-8 System design and integration of technologies
Use case: all

Short Project Description:

The objective of my work is to go beyond farming practices related to three use cases (i) arable, (ii) horticulture, and (iii) dairy farming systems and propose a next generation design of these systems. The design integrates (i) biological principles, (ii) decision-making mechanisms, (iii) robotic handling, (iv) technological enables, (v) societal acceptance criteria to move towards a more sustainable future. In this project, the design requirements need to be determined based on our novel concept called technolog-4-ecology. It is characterized by the fact that it stands on ecology and the development of technologies that support it.