Search

Controlling the greenhouse environment usually involves bounds (restrictions) on the indoor conditions. In model-based control, these bounds are meant to keep the plant environment away from high risk zones, the effects of which are not sufficiently well described by the model. The objective is to estimate the potential energy saving and gain in profit resulting from relaxing the bounds. The calculations employed a previously developed simulation-optimization program in conjunction with a new, solar-driven evapotranspiration model. Spanning a whole year, the simulations were carried out for a typical Dutch tomato-greenhouse configuration, utilising a gas-fired boiler for both heat and CO2 production, and a water tank for day-to-night heat storage. The main findings are as follows: Provided that the crop is not damaged by the change, the expected gain from increasing the permissible humidity is about 0.74 € m−2y−1 per one percent relative humidity, and from reducing the minimum temperature − about 0.87 € m−2y−1 per degree. Roughly 2% of the energy is saved by a 1K reduction of temperature or a 1% increase of the relative humidity. Adding a heat buffer has no noticeable effect on the total amount of gas used. It does, however, increase the effectiveness of CO2 enrichment, thus increasing the yield and the economic gain (by 3.4 €m−2y−1). Replacing the profit goal by energy-use-minimisation goal, results in a substantial loss (−11.5 € m−2y−1).

Twenty-three lines from a mutagenized Bangladeshi BARI Gom-25 wheat population that included previously identified salt-tolerant lines, and the BARI Gom-25 control variety, were cultivated in a drip-irrigated salinity test field at Salt Farm Texel, Netherlands, to assess their performance during salt stress in European climatic conditions. Lines were tested at irrigation salinity levels of 1, 4, 8, 12, 16, and 20 dS m−1 in four repetitions of plots with 24 plants per plot. Average plant height, tiller number, spike length, frequency of live plants, and total grain weight (TGW) were recorded as functions of seasonal mean pore water salinity in the soil. Increases in salinity triggered reductions in all evaluated variables of the assessed lines and the control variety. However, nine mutagenized lines had at least twofold higher mean TGW than the control variety, 18.73 ± 4.19 g/plot at 1–16 dS m−1 salinity levels. Common models of salt tolerance confirmed this pattern, but there were no clear differences in salinity tolerance parameter estimates between the mutagenized lines and the control variety. Thus, despite the apparent similarity in responses of all lines to salinity increase, we clearly identified lines that tended to have higher TGW at given salinities than the control variety. This higher TGW at the full range of salinity treatments indicates not only a possible higher salinity tolerance but a higher yield potential as well. The mechanisms involved clearly warrant further attention.

Day-to-night heat storage is often practiced in cold-climate greenhouses. It is suggested to manage the heat storage by considering the co-state (virtual value) of the stored heat in the on-line optimization of the greenhouse environment. Examples worked out for a periodic square-wave weather show that a properly selected constant co-state can produce an optimal solution to the control problem. The optimal co-state is shown to change with time over the year. Maximizing the performance criterion can also be achieved by minimizing the time that the heat buffer is either completely empty or completely full.

Crank’s analytical approximations for Fick’s diffusion equation were used to investigate the effect of moisture dependent sample thickness and diffusivity on the behavior of yam (Dioscoreaceae rotundata) cubicles during drying processes. Drying and shrinkage experiments were separately conducted at temperatures of 30, 40 and 50 °C in a cabinet drier. The comparative study of moisture dependent shrinkage and moisture dependent diffusivity justifies the interdependence of diffusivity and shrinkage due to water loss during drying. The behavior for yam is best explained by a combination of fractal moisture dependent shrinkage and moisture dependent diffusion, describing both the drying and rate curves better with good prediction of the high moisture regions. This assertion was reached as a result of low mean square error, standard error, percentage relative deviation, Akaike’s Information Criterion and high coefficient of determination. The results may indicate a varying mobility of water in food matrix of different moisture content in the multilayer and monolayer regimes.

Day-to-night heat storage in water tanks (buffers) is common practice for cold-climate greenhouses, where gas is burned during the day for carbon-dioxide enrichment. In Seginer, I., van Straten, G., van Beveren, P. (2017). Biosystems Engineering, 161, 188–199, an optimal environmental control approach was outlined for conventional greenhouses, the idea being that a virtual value of the stored heat (its ‘co-state’) could be used to guide instantaneous control decisions. The value of the co-state was heuristically adjusted to minimise the time the buffer was ineffective (being empty or full). Here the same approach is applied to greenhouses with co-generation of heat and electricity (CHP). The parameters-set and weather are characteristic of tomato greenhouses in The Netherlands. The main results are: (1) The heuristic control method is easily adapted to systems with CHP; (2) Buffers are more useful to CO2 enrichment in the summer than to heating in the winter; (3) There is strong synergy between the two production systems – tomatoes and electricity. The tomato crop benefits from the by-products of electricity generation, namely CO2 and heat, sharing this benefit to support low electricity prices; (4) The combined system produces less CO2 pollution than the two production systems operating independently; (5) The contribution of the CHP and buffer to the economic performance of the system is very significant, while that of the thermal screen and boiler is marginal; (6) Flexibility in the system is important. A buffer and/or continuously controlled boiler and CHP are essential to achieving high profitability.

Day-to-night heat storage using water tanks (buffers) is common practice in cold-climate greenhouses, where gas is burned during the day for carbon dioxide enrichment. In this study an optimal control approach is outlined for such a system, based on the idea that the virtual value (shadow price) of the stored heat, its 'co-state', could be used to guide the instantaneous control decisions. If this value is high, the system has an incentive to fill the heat storage (buffer), and vice versa if the co-state is low. The optimal co-state trajectory maximises the net income (performance criterion). To illustrate the method, a system description and a parameter-set roughly representative of tomato greenhouses in The Netherlands is used. The results, for daily-periodic weather, show: (1) The optimal co-state is constant (same value night and day), in contrast to the varying set-points and control fluxes. (2) The optimal solution is associated with minimum time on the storage bounds (minimum time of full or empty buffer). (3) The optimal virtual value (co-state) of stored heat is about the same as the actual cost of boiler heat during winter and about zero in summer. (4) The gain from installing a buffer is highest during spring and minimal in winter. (5) The intensive utilisation of the heat buffer in summer and its low utilisation in winter indicate that the justification of the heat storage practice, under the assumed conditions, is more the need for CO2 enrichment in summer than the need for heating in winter.

Day-to-night heat storage in water tanks (buffers) is common practice in cold-climate greenhouses, where gas is burned during the day for carbon dioxide enrichment. In Part 1 of this study, an optimal control approach was outlined for such a system, the basic idea being that the virtual value (shadow price) of the stored heat (its ‘co-state’) could be used to guide the instantaneous control decisions. The results for daily-periodic weather showed: (1) The optimal co-state is constant in time. (2) The optimal solution is associated with minimum time on the storage bounds (buffer empty or full). With these conclusions as guidelines, a semi-heuristic procedure of optimisation for realistic (i.e. not strictly periodic) weather is developed. The co-state remains constant while the storage trajectory is between the heat storage bounds. It is gradually increased while the buffer is empty, and decreased when the buffer is full, attempting to push the trajectory away from the bounds, thus minimising the time that the buffer is idle. The main outcomes are: (1) No information about the future is required. (2) The algorithm changes the co-state automatically, producing the correct annual variation (high in winter and low in summer). (3) The predictions of yield and heat requirement compare favourably with practice. (4) The gain in performance achievable with the suggested method is probably 75% or more of the true optimum. (5) The procedure can be used in the design stage to determine the optimal buffer size and the usefulness of other modifications of the system.

Worldwide there is a wide variety of greenhouse types en cultivated crops, and consequently control options and control solutions differ. Yet, there are common characteristics that must be considered in every control solution. A versatile framework for greenhouse climate control aiming at economic optimal control is presented. The framework is based on time-scale decomposition. Factors to determine the feasibility of the approach are discussed, as well as options to overcome some of the challenges and issues. The expectation is that with expansion and professionalization of the greenhouse industry the scope for optimal control will become wider.

Background Transcription of genes coding for xylanolytic and cellulolytic enzymes in Aspergillus niger is controlled by the transactivator XlnR. In this work we analyse and model the transcription dynamics in the XlnR regulon from time-course data of the messenger RNA levels for some XlnR target genes, obtained by reverse transcription quantitative PCR (RT-qPCR). Induction of transcription was achieved using low (1 mM) and high (50 mM) concentrations of D-xylose (Xyl). We investigated the wild type strain (Wt) and a mutant strain with partial loss-of-function of the carbon catabolite repressor CreA (Mt). Results An improved kinetic differential equation model based on two antagonistic Hill functions was proposed, and fitted to the time-course RT-qPCR data from the Wt and the Mt by numerical optimization of the parameters. We show that perturbing the XlnR regulon with Xyl in low and high concentrations results in different expression levels and transcription dynamics of the target genes. At least four distinct transcription profiles were observed, particularly for the usage of 50 mM Xyl. Higher transcript levels were observed for some genes after induction with 1 mM rather than 50 mM Xyl, especially in the Mt. Grouping the expression profiles of the investigated genes has improved our understanding of induction by Xyl and the according regulatory role of CreA. Conclusions The model explains for the higher expression levels at 1 mM versus 50 mM in both Wt and Mt. It does not yet fully encapsulate the effect of partial loss-of-function of CreA in the Mt. The model describes the dynamics in most of the data and elucidates the time-dynamics of the two major regulatory mechanisms: i) the activation by XlnR, and ii) the carbon catabolite repression by CreA. Electronic supplementary material The online version of this article (doi:10.1186/s12918-016-0257-4) contains supplementary material, which is available to authorized users.

This work studies the controllability of heat-integrated zeolite adsorption dryers. Mean product moisture content, temperature and vitamin C concentration (representative of product quality) are considered as controlled variables. Set-point tracking and disturbance rejection controllability metrics are considered in addition to energy performance sensitivity. In adsorption dryers, the adsorption system introduces extra degrees of freedom of which some input–output pairs are promising. For corresponding inputs, adsorption dryers are shown to have higher steady-state gains than equivalent conventional dryers due to correlation between dehumidification, adsorption heat and the controlled variables. They also show improved resilience to ambient air disturbances due to adsorbent subsystem-induced self-regulation properties. The encouraging mechanisms of the self-regulation are adsorption heat, kinetic and equilibrium properties of the adsorbent. Due to the high correlation between product moisture content and temperature, improved controllability is observed when vitamin C concentration is used as an output variable instead of product temperature. It is thus proposed that on the availability of reliable soft sensors or state estimators, instead of product temperature, vitamin C or some other temperature-dependent quality measure should be controlled in addition to product moisture in decentralised drying system control. Under perfect rejection of unfavourable disturbances like ambient temperature drop and humidity rise, the energy performance of adsorption dryers is not significantly degraded, whereas, it is for conventional systems.

This study considers two aspects of the implementation of a biomass growth observer and specific growth rate controller in scale-up from small- to pilot-scale bioreactors towards a feasible bulk production process for whole-cell vaccine against whooping cough. The first is the calculation of the oxygen uptake rate, the starting point for online monitoring and control of biomass growth, taking into account the dynamics in the gas-phase. Mixing effects and delays are caused by amongst others the headspace and tubing to the analyzer. These gas phase dynamics are modelled using knowledge of the system in order to reconstruct oxygen consumption. The second aspect is to evaluate performance of the monitoring and control system with the required modifications of the oxygen consumption calculation on pilot-scale. In pilot-scale fed-batch cultivation good monitoring and control performance is obtained enabling a doubled concentration of bulk vaccine compared to standard batch production

The complexities of the dynamic processes associated with bio- and ecological systems offer many challenges for the control engineer. Applying dynamic modelling and control can aid understanding of their complexities. Moreover, using such complex systems as test-beds for new control methods can highlight their limitations (e.g. in relation to system identification) and thus act as a catalyst for methodological advance. This paper continues the theme of exploring opportunities and achievements in applying modelling and control in the bio- and ecological domains.

Sorption isotherms of corn and starch cylinders with immobilised catalase are experimentally determined at different temperatures for use in drying models in optimal control studies. This application of the sorption isotherm requires an accurate prediction of the sorption data at different temperatures for the low water activity range. The GAB equation is used for the prediction of the sorption isotherms. Two major problems are encountered by employing standard procedures, ie prediction of sorption at aw

Singular and non-singular control trajectories of agricultural and (bio) chemical processes may need to be recalculated from time to time for use in closed-loop optimal control, because of unforeseen changes in state values and noise. This is time consuming. As an alternative, in this paper, numerical, nonlinear, static state feedback laws are developed for optimal control on the singular arc that can be applied in closed-loop without the need for iteration. The efficacy of these laws is demonstrated in an example.

A mathematical formulation of a realistic optimal greenhouse open-loop control problem for achieving a specified lettuce nitrate concentration is expounded. Like many complex agricultural and (bio)chemical optimal control problems, this problem consisted of non-linear differential algebraic equations, affected by high frequency changing uncontrollable inputs, possibly conflicting path constraints and end-constraints. By reformulating the problem into a problem with isoperimetric constraints in order to deal properly with the path constraints it was cast in a form suitable for a numerical solution with the ACW-gradient optimisation. The optimisation was successful, as is shown in an example.

Safe drinking water without toxic chemicals is crucial for people's health. A recently developed sensor for the detection of toxic components in water is the microbial fuel cell (MFC)-based biosensor. In this biosensor, substrate consumption rate and metabolic activity of bacteria are directly related to the electric current. A reduction in current under otherwise similar conditions is an indication of toxic inhibition. Under steady state conditions, current can be described by the Butler–Volmer–Monod (BVM) model. Knowing which parameters of this model change under toxic contamination can give an indication on the type of toxicity. The model requires that the substrate concentration is known. It is shown in this paper that is not possible to estimate both the substrate concentration as well as the BVM parameters on-line from current data at constant overpotential. However, it appears that substrate concentration and substrate consumption rate can be estimated on-line, and that after a linear reparametrization the BVM parameters can be estimated by ordinary least-squares techniques from a polarization curve that is generated as soon as a suspect change in current occurs. Analysis shows that a weighted least-squares method is necessary to secure a good fit at the overpotentials where current is most sensitive to changes in kinetic parameters. A protocol for on-line detection of toxicity and for detection of the type of kinetic inhibition is provided.

Magnetic resonance imaging (MRI) offers unique opportunities to monitor moisture transport during drying or heating of food, which can render unexpected insights. Here, we report about MRI observations made during the drying of broccoli stalks indicating anomalous drying behaviour. In fresh broccoli samples the moisture content in the core of the sample increases during drying, which conflicts with Fickian diffusion. We have put the hypothesis that this increase of moisture is due to the stress diffusion induced by the elastic impermeable skin. Pre-treatments that change skin and bulk elastic properties of broccoli show that our hypothesis of stress-diffusion is plausible.

Developments in low temperature drying of food products are still an interesting issue; especially with respect to the energy efficiency. This research studies the energy efficiency that can be achieved by a dryer using air which is dehumidified by zeolite. Experimental results are fitted to a dynamic model to find important variables for the drying operation. The results show that ambient air temperature as well as the ratio between air flow for drying and air flow for regeneration, affect the energy efficiency significantly. Relative humidity of used air, and shift time have a minor effect on the dryer performance. From the total work, it can be noted that the dryer efficiency operated at 50-60°C achieves 75 percent, which is attractive for drying of food products.

This work assesses the energy efficiency of dehumidification drying vis-à-vis conventional convective drying techniques. Mathematical models are developed by means of which the energy efficiencies of different dehumidification dryer types are expressed in terms of that of a conventional convective dryer operating at the same temperature. This permits the isolation of important design and operational parameters specific to each dryer type which when optimized, improve energy efficiency for the same product quality requirement and ensure better product quality for the same efficiency as a conventional dryer. Desiccant dehumidification systems have the advantage of providing further opportunities for beneficial heat integration.

One of the challenges in the reconstruction of genetic network is to find experimental designs that maximize the information content in the data. In this work the information value of time course experiments (TCEs) is used to rank experimental designs. The study concerns the dynamic response of genes in the XlnR regulon of Aspergillus niger, whereby it was the goal to find the best moment to administer an extra pulse of inducing D-xylose. Low and high trigger concentrations are considered. The models that govern the regulation of the target genes in this regulon are used for simulation. Parameter sensitivity analysis, Fisher Information Matrix (FIM) and the E-modified criterion are used for the design performance assessment. The results show that the best time to give a second pulse of a low concentration trigger of D-xylose is when the D-xylose concentration from the first pulse is not yet completed reduced. Secondly, pulses with high trigger concentrations were simulated, parameter sensitivities computed, and the experimental designs evaluated. Overall, after the first pulse of 1 mM D-xylose, using a second pulse of 5 (or 10) mM D-xylose yields the best experimental design - leading to improved parameter estimates.

Currently available models describing microbial fuel cell (MFC) polarization curves, do not describe the effect of the presence of toxic components. A bioelectrochemical model combined with enzyme inhibition kinetics, that describes the polarization curve of an MFC-based biosensor, was modified to describe four types of toxicity. To get a stable and sensitive sensor, the overpotential has to be controlled. Simulations with the four modified models were performed to predict the overpotential that gives the most sensitive sensor. These simulations were based on data and parameter values from experimental results under non-toxic conditions. Given the parameter values from experimental results, controlling the overpotential at 250 mV leads to a sensor that is most sensitive to components that influence the whole bacterial metabolism or that influence the substrate affinity constant (Km). Controlling the overpotential at 105 mV is the most sensitive setting for components influencing the ratio of biochemical over electrochemical reaction rate constants (K1), while an overpotential of 76 mV gives the most sensitive setting for components that influence the ratio of the forward over backward biochemical rate constants (K2). The sensitivity of the biosensor was also analyzed for robustness against changes in the model parameters other than toxicity. As an example, the tradeoff between sensitivity and robustness for the model describing changes on K1 (IK1) is presented. The biosensor is sensitive for toxic components and robust for changes in model parameter K2 when overpotential is controlled between 118 and 140 mV under the simulated conditions.

An RTK-DGPS (Real Time Kinematic Differential Global Positioning System) based autonomous field navigation system including automated headland turns was developed to provide a method for crop row mapping combining machine vision, and to evaluate the benefits of a behaviour based reactive layer in a hybrid deliberate systems architecture. Two experiments were performed at the same time: following of pre-planned paths reconstructed from crop row positions based on RTK-DGPS and crop row mapping by combining vision-based row detection with RTK-DGPS information. The standard deviation, mean, minimum and maximum lateral error of the robot vehicle while following a straight path on the field with RTK-DGPS at a speed of 0.3 m s−1 were respectively 1.6, 0.1, −4.5 and 3.4 cm. The standard deviation, mean, minimum and maximum of the heading error were 0.008, 0.000, −0.022 and 0.023 rad. The point-in-polygon algorithm proved to be a suitable method for detection in which part of the field the actuator position coordinates or the field of view of the camera are located. A smooth headland path that connected to the subsequent path along the crop was generated in realtime using a spline based algorithm. The hybrid deliberate software architecture with a behaviour based reactive layer allowed a convenient evaluation of the robot performance. Results from the field experiments showed that the implement can be guided along a defined path with cm precision using an autonomous robot navigating in a field.

While many professionals with a background in agricultural and bio-resource sciences work with models, only few have been exposed to systems and control theory. The purpose of this paper is to elucidate a selection of methods from systems theory that can be beneficial to quantitative agricultural science. The state space representation of a dynamical system is the corner stone in the mainstream of systems theory. It is not well known in agro-modelling that linearization followed by evaluation of eigenvalues and eigenvectors of the system matrix is useful to obtain dominant time constants and dominant directions in state space, and offers opportunities for science-based model reduction. The continuous state space description is also useful in deriving truly equivalent discrete time models, and clearly shows that parameters obtained with discrete models must be interpreted with care when transferred to another model code environment. Sensitivity analysis of dynamic models reveals that sensitivity is time and input dependent. Identifiability and sensitivity are essential notions in the design of informative experiments, and the idea of persistent excitation, leading to dynamic experiments rather than the usual static experiments can be very beneficial. A special branch of systems theory is control theory. Obviously, control plays an important part in agricultural and bio-systems engineering, but it is argued that also agronomists can profit from notions from the world of control, even if practical control options are restricted to alleviating growth limiting conditions, rather than true crop control. The most important is the idea of reducing uncertainty via feed-back. On the other hand, the systems and control community is challenged to do more to address the problems of real life, such as spatial variability, measurement delays, lacking data, environmental stochasticity, parameter variability, unavoidable model uncertainty, discrete phenomena, variable system structures, the interaction of technical ad living systems, and, indeed, the study of the functioning of life itself., Iako mnogi profesionalci s područja agrotehnike i bio-znanosti rade s modelima, tek je nešto malo onih koji su izloženi uporabi teorije sustava i automatskog vođenja. Namjena je ovog rada raščišćavanje izbora metoda s područja teorije sustava koji mogu biti od koristi kvantitativnoj agronomskoj znanosti. Predstavljanje dinamičkih sustava prostorom stanja temeljno je usmjerenje teorije sustava. U agro-modelima je slabo poznato da je linearizacija, koju slijedi evaluacija svojstvenih vrijednosti i svojstvenih vektora matrice sustava, korisna pri dobivanju dominantnih vremenskih konstanti i dominantnih usmjeravanja u prostoru stanja te pri nuđenju mogućnosti znanstveno zasnovanog pojednostavljenja modela. Opis kontinuiranog prostora stanja također je koristan u dobivanju stvarno ekvivalentnih diskretno-vremenskih modela te jasno pokazuje da se dobiveni parametri diskretnih modela moraju pažljivo interpretirati kada se prenose u okoliš s drugim kodnim modelom. Analiza osjetljivosti dinamičkih sustava otkriva da je osjetljivost ovisna o vremenu i ulazima. Mogućnost identifikacije i osjetljivost bitni su pojmovi u projektiranju informativnih pokusa, a ideja trajne pobude koja vodi dinamičkim pokusima a ne uobičajenim statičkim pokusima može biti vrlo korisna. Posebna grana sustavske teorije jest teorija automatskog vođenja. Očigledno, vođenje igra važnu ulogu u agronomskom i bio-sustavskom inženjerstvu, ali se naglašava da agronomi mogu profitirati iz pojmova svijeta automatike, čak i ako je automatizacijska inačica ograničena na olakšavanje uvjeta ograničenja rasta, a ne i pravog vođenja usjeva. Najvažnija je ideja da se uporabom povratne veze smanji neodređenost. S druge strane zajednica je sustavskih inženjera i automatičara izazvana da se više posveti stvarnim problemima života poput prostorne promjenjivosti, mjernih kašnjenja, nedostataka podataka, stohastičnosti okoliša, varijabilnosti parametara, nezaobilazne neodređenosti modela, fenomena diskretizacije, varijabilnosti sustavskih struktura, međudjelovanja tehničkih i živih sustava, i napose studiju funkcioniranja života kao takvog.

Spectroscopic instrumentation is often seen as promising for process analytical technology (PAT) to enhance control of manufacturing (bio)pharmaceuticals. The interpretation of near infrared spectra is challenging due to the large number of wavelengths recorded and the overlapping absorbance features of near infrared spectroscopy. This work applies a controlled random search procedure to select an optimal window of wavelengths giving a good calibration model for biomass concentrations during cultivation of Bordetella pertussis, the causative agent of whooping cough. The proposed wavelengths selection procedure outperforms the traditional calibration procedures. In the second half of the paper, the near infrared based predictions are compared with the estimations obtained from a software sensor for biomass and specific growth rate based in standard measurements of oxygen consumption. Both methods estimate the exponential biomass growth properly. The near infrared predictions depend on the quality of the training dataset, which needs to encompass all possible sources of temporary disturbances like pH and dissolved oxygen. If the training dataset does not comprise such disturbances, then the accuracy and robustness of the near infrared predictions are less favorable than those of the software sensor. Although near infrared has the potential to provide more information than just biomass, the software sensor is the preferred choice for feedback control of biomass and specific growth rate.

Generally, ecosystems modeling is obstructed by the problem of sparse and unreliable data, and lack of knowledge about processes dominating the system. Under these circumstances, set theoretic uncertainty models are an appropriate alternative to probabilistic models. The only requirement is that the uncertainty is pointwise bounded. A newly developed set membership identification procedure is presented and demonstrated by an application to the modeling of shallow lake eutrophication. First, a set of parameter vectors is identified. Analysis of the set reveals a dominant direction spanned by four algal growth and death parameters. Second, on the basis of additional fuzzy set theoretic assumptions, a formal min-max estimation is performed to obtain information about the model validity. If the model appears to be (partially) invalid, the degree of invalidity, affecting the model prediction uncertainty, can be represented by an estimate of the model structure error in addition to the uncertainty contained in the identified set of parameter vectors.

The objective of this research is the replacement of hand weeding in organic farming by a device working autonomously at field level. The autonomous weeding robot was designed using a structured design approach, giving a good overview of the total design. A vehicle was developed with a diesel engine, hydraulic transmission, four-wheel drive and four-wheel steering. The available power and the stability of the vehicle does not limit the freedom of research regarding solutions for intra-row weed detection and weeding actuators. To fulfill the function of navigation along the row a new machine vision algorithm was developed. A test in sugar beet in a greenhouse showed that the algorithm was able to find the crop row with an average error of less than 25 mm. The vehicle is a versatile design for an autonomous weeding robot in a research context. The result of the design has good potential for autonomous weeding in the near future.

Singular and non-singular control trajectories of agricultural and (bio) chemical processes may need to be recalculated from time to time for use in closed-loop optimal control, because of unforeseen changes in state values and noise. This is time consuming. As an alternative, in this paper, numerical, nonlinear, static state feedback laws are developed for optimal control on the singular arc that can be applied in closed-loop without the need for iteration. The efficacy of these laws is demonstrated in an example.

The complexities of the dynamic processes and their control associated with biological and ecological systems offer many challenges for the control engineer. Over the past decades the application of dynamic modelling and control has aided understanding of their complexities. At the same time using such complex systems as test-beds for new control methods has highlighted their limitations (e.g. in relation to system identification) and has thus acted as a catalyst for methodological advance. This paper continues the theme of exploring opportunities and achievements in applying modelling and control in the bio- and ecological domains.

One way of guiding autonomous vehicles through the field is using a vision based row detection system. A new approach for row recognition is presented which is based on grey-scale Hough transform on intelligently merged images resulting in a considerable improvement of the speed of image processing. A colour camera was used to obtain images from an experimental sugar beet field in a greenhouse. The colour images are transformed into grey scale images resulting in good contrast between plant material and soil background. Three different transformation methods were compared. The grey scale images are divided in three sections that are merged into one image, creating less data while still having information of three rows. It is shown that the algorithm is able to find the row at various growth stages. It does not make a difference which of the three colour to grey scale transformation methods is used. The mean error between the estimated and real crop row per measurement series varied from 5 to 198 mm. The median error from the crop row detection was 22 mm. The higher errors are mainly due to factors that do not occur in practice or that can be avoided, such as a limited number and a limited size of crop plants, overexposure of the camera, and the presence of green algae due to the use of a greenhouse. Inaccuracies created by footprints indicate that linear structures in the soil surface in a real field might create problems which should be considered in additional investigations. In two measurement series that did not suffer from these error sources, the algorithm was able to find the row with mean errors of 5 and 11 mm with standard deviations of 6 and 11 mm. The image processing time varied from 0. 5 to 1.3 s per image. (c) 2007 Elsevier B.V. All rights reserved.

A brief report is given of the field robot contest held in Wageningen, 5–6 June 2003. The experience of the student competitors sheds light on a number of issues that need to be addressed before autonomous vehicles will become a reality.

The adjustable control-variation weight (ACW)-gradient method proposed by Weinreb [Optimal Control with Multiple Bounded Inputs, Department of Electrical Engineering, Stanford University, Stanford, 1985, p. 148] is put to the test in finding optimal control laws for an optimisation problem with bounds on the inputs and terminal state constraints, presented by loslovich and Seginer [Acceptable nitrate concentraion of greenhouse lettuce: an optimal control policy for temperature, plant spacing and nitrate supply, in: Proceedings of the Agricontrol 2000, Wageningen, The Netherlands, IFAC, Wageningen University and Research Centre, Royal Dutch Institute of Engineers, 2000]. By making certain assumptions they derived properties of the solution in an analytic way. Here, it is shown that the numerical ACW-gradient algorithm is capable of finding solutions without making additional assumptions.

To achieve good product quality and consistency the bioreactor has to be harvested in time. A prediction algorithm is developed which aids the operator in determining and scheduling the right moment of harvest. An observer was build for biomass concentration and specific growth rate from oxygen consumption measurements only. With these estimations and a dual-substrate kinetic model the substrate concentrations are calculated. With the estimated biomass and calculated substrateconcentrations the future trajectory of the cultivation is computed by forward simulation to successfully determine the harvest time on-line. Several experiments with a benchscale bioreactor to cultivate B. pertussis were carried out.