Kenji Hatou

Early detection of drought stress in tomato plants with chlorophyll fluorescence imaging–practical application of the speaking plant approach in a greenhouse–

Abstract The chlorophyll fluorescence imaging technique is useful for evaluating photosynthetic functions of plants without actually touching the plant. In our previous study, we developed a chlorophyll fluorescence imaging system for tomato plants cultivated in greenhouses. This imaging system measures the chlorophyll fluorescence induction phenomenon, a dynamic change in chlorophyll fluorescence intensity induced by illuminating a dark-adapted leaf with a stable intensity excitation light, and analyzes the shape of the induction curve, i.e., the temporal course of chlorophyll fluorescence intensity during this phenomenon. The shape of the induction curve is characterized by an initial maximum peak (P), subsequent transient dip (S), and secondary small peak (M). We defined an index, the photosynthetic function index (PFI; fluorescence intensity of P divided by the average fluorescence intensity from S to M), to evaluate the shape of the induction curve. In this study, we applied this system to detect drought stress in tomato plants cultivated in a semi-commercial greenhouse. PFI was clearly lower in stressed plants than in healthy plants. The decreased PFI in stressed plants is probably attributable to photosynthetic dysfunction in these plants.

Intelligent Control for Plant Production System

Abstract For the optimization of long-term plant growth in hydroponics, this paper proposes a hierarchical intelligent control system consisting of an expert system and a hybrid system based on genetic algorithms and neural networks. These two control systems are used appropriately, depending on the plant growth. The plant growth is controlled by the nutrient concentration of the solution. The expert system was used for determining the appropriate setpoints of nutrient concentration through the whole of the growth stages, and the hybrid system for determining the optimal setpoints of nutrient concentration which maximize TLL/SD (TLL: Total leaf length, SD: Stem diameter) only during the initial growth (seedling) stage. In the hybrid system, TLL/SD as affected by nutrient concentration was first identified using neural networks and then the optimal value was determined through simulation of the identified model using genetic algorithms. The setpoints from the expert system were almost similar to those used by a skilled grower. Also, the setpoints from the hybrid system increased the TLL/SD. Thus, this intelligent control technique allowed the optimization of both long-tern and short-term plant growth to be realized. This shows that this control technique is suitable for the optimization of such complex and long-terns processes as the plant-cultivation process.

Computer Integrated Agricultural Production

Abstract Greenhouse production of vegetables has been remarkably developed. Furthermore, vegetables could be cultivated in the factory where such environmental factors as light intensity and temperature are controlled just like in process industries. In the system, many computers are used for environmental control, nutrient control and management of cultivation. Of course, the artificial intelligence is also introduced for the expert system for control and diagnosis of the cultivated vegetables. On the other hand, progress in automated mechanization for seeding and transplanting has made “greenhouse automation” fit for practical use just like “factory automation (FA)” in industries. Now, process industries are rationalized based on the concept of so called “Computer Integrated Manufacture (CIM)”. Therefore, it might be noted that the system in the agricultural production such as the vegetable factory should also be considered based on the concept of CIM. That is the “Computer Integrated Agricultural Production (CIAP)”. In this paper, we examine the vegetable factory from the “CIAP” point of view. Computer network composed of both usual personal computer for environmental control and the special computer for the artificial intelligence is examined. It seems evident that the CIAP discussed in this paper is expected as the most effective system in the coming generation.

Range Image Analysis for the Greenhouse Automation in Intelligent Plant Factory

Abstract In the paper, the range image analysis is examined. For the intelligent plant factory, computer integrated system has been introduced for these years. Now, it proved excellent to develop the intelligent plant factory. Further, it becomes necessary and inevitable that the characteristic of such the range images as crops and fruits distributed in these system should be taken into account to the decision factor of these systems. For the purpose of both designing of the virtual system and putting greenhouse automation including agro-robot into practice, 3-D recognition is an urgent theme. It is difficult, however, to do it for real time use in small computer. But, for agricultural purpose, it may be allowable that the model is limited to polyhedral shape with convex surface. In the condition, we proposes new approach to 3-D shape recognition for real time use. Supposing the parameter ( l /L) given in Figure 6, effective algorithim is found for the agricultural use.

Computer Integrated Plant Factory Based on Artificial Intelligence

Abstract In this study, the applications of AI in the agricultural field are discussed, using the computer integrated plant factory. These are decision support system for the adjustment of optimal control and diagnosis expert system to the trouble of the control system. In the detail, the application of decision support system to the adjustment of optimal setting points of the nutrient solution during the whole processes in the tomato growth is discussed. It may be noted that the optimal values of the EC in the nutrient solution based on this expert system resulted in high quality of the fruits. As for the diagnosis, both water recirculation and ion sensor in the hydroponic system are also discussed. Utilizing the mean squared error between the estimated water temperature obtained from the identification and the observed water temperature as the criterion of production rules in the expert system, we could realize more accurate and useful diagnosis system of the trouble of water recirculation. Furthermore, the trouble of ion sensor could be diagnosed rapidly by predicting the future behaviour of ion concentration based on the same method of the identification.

The optimization of the fruity separation algorithm of accumulating the strawberry automatic harvesting robot

Abstract In this study, the algorithm, which separated the strawberry fruit at high speed using chain code for separating the fruit at high speed, was examined. Harvesting robot that it used in this studyhas installed the manipulator to rectangular coordinate robots of the three-axial style. CCD camera installed itat the tip of the manipulator, and the image of the strawberry is incorporated in the computer using this camera. There is the resolution of the image at length of 320 pixels, width of 200 pixels, and 16 bits per one pixel. The procedure of the image processing was done, as it was shown in the following. 1. The input of the image from CCD camera. 2. The recognition of the group of the fruit by the binarization processing. 3. Noise rejection and compensation of the contour data. 4. The separation of the fruit in the harvest time. 5. The separation of the fruit according to the chain code. It was proven that CCD camera and distance to the fruit were the important factorswhich decide the recognition speed in this experiment. At CCD camera and distance of the fruit, it was proven that the time for the recognition slowed down, when leaving recognition rate lowers, and when it closes. It is considered the reason why the time for the recognition slowed down that the data increased for the image processing, and that therefore, the calculation amount increased. Then, it was possible to shorten the whole processing time by advancing the optimization of the program. By the optimization of the program, the processing time was almost able to be shortened by 3 seconds

COMPUTER SUPPORT SYSTEM FOR TOMATO CULTIVATION IN PLANT GROWTH FACTORY

Abstract A support system of tomato cultivation in plant growth factory was designed and examined. The purpose of the system lies mainly in the support of horticultural operation in the special plant growth factory developed by Idemitsu Kosan Co., LTD. The system is found to be effective in the horticultural operation and management. It seems evident that operation and management for tomato cultivation in such a plant growth factory could not be carried out without a computer-aided cultivation support system. Furthermore, diagnosis of physiological disorder and disease is examined by introducing the artificial intelligence into the cultivation support system. Initial disorder caused by environmental stress and nutrient deficit could be diagnosed in the system. The computer support system as shown in this paper, may be expected to be helpful for any horticultural cultivation in plant growth factories.

Development of Decision Support Application based on the Prediction Model of Tomato Yields in Intelligent Greenhouse

Abstract In intelligent greenhouse, balance of yield with heating cost is one of important point for manager. In this study, we constructed growing model of tomato based on accumulated temperature in greenhouse, predictive model of air temperature inside greenhouse based on weather forecast, and heating load model in greenhouse. Then using these models, we developed the application to support managers decision by predicting yield of tomato, and flower differentiation date, flowering date, harvest start date and harvest end date of each stage, and furl cost of heating.

Analysis of the effects of DIF for the flower setting habit of the tomato in greenhouse

Abstract Currently, Changes in crop yields and in a monthly reduction of crop yields due to the occurrence of defective flower setting and among abnormal flower cluster from disturbance of flowers setting habit in greenhouse. In this study, we went to the measurement of biological information of the plant in order to create a database for diagnostics and environmental control growth based on the Speaking Plant Approach. As a result, Disturbance of flower setting habit of tomato was demonstrated in the following tomato DIF4. DIF of he for suitable cultivation in tomato is DIF5 ~ 7. But, Likely to affect of from DIF7 onward in flower setting habit of tomato is low. The strength of the DIF and cultivation stage is not relevant to tomato. And, DIF will delay flower differentiation.

Optimization of Watering for Minimizing the Inside Temperature of Zero Energy Cool Chamber for Storing Fruits and Vegetables

Abstract A zero energy cool chamber (ZECC) has been developed for storing fruits and vegetables from the viewpoints of low installation and operating cost. The inside temperature of the ZECC is cooled by adding water to a sand and zeolite based filler between the brick walls based on the principles of a natural evaporative cooling mechanism. The objective of this study was to minimize the inside temperature of the ZECC by controlling watering operation using an intelligent optimization technique combined with neural network and genetic algorithm. The objective function was given by the average value of the inside temperature for one day. For optimization, the control process (24 hours) was divided into 8 steps, and the optimal value (8-step ON-OFF intervals) of watering was obtained using neural networks and genetic algorithms. In this method, dynamic changes in the inside temperature of the ZECC, as affected by the watering strategy and outside temperature, were first identified using neural network, and then the optimal value, which minimized the objective function, was determined through simulation of the identified neural-network model using genetic algorithm. The average inside temperature for this optimal ON-OFF control was 5°C lower than that for the simple ON-OFF watering for 24 hours, and was also 8°C lower than that for no watering. The ZECC with the optimal ON-OFF watering strategy extended the shelf-life of untreated tomato from 7 to 16 days. Thus, it was concluded that a ZECC optimized by using neural networks and genetic algorithms is useful for storing tomato with no electric energy.