Abstract: The objective of the study was the assessment of the potential of laser distance sensors for measuring morphologic crop parameters to perform site specific crop management. In a first step five pre-selected laser sensors were tested. Based on these results a triangulation laser sensor was chosen and modified. The modified sensor was tested in terms of the accuracy for distance measurements depending on the inclination angle of the plant surface and the rate of failure measurements depending on illumination conditions and measuring distance. Under defined lab condition the mean distance standard error was 2.8 mm. The inclination angle of the plant surfaces did not reduce the measuring accuracy. Bright sunshine and intensive reflecting surfaces (e.g., sandy soils) without crop coverage caused increased failure measurements. Taking into account all test results a potential of laser distance sensors for measuring of morphologic crop parameters was proved.
INTRODUCTION
The agriculture of the future should be both competitively and environment friendly. This aim can be achieved only by using less natural resources and by integrating more information in the production processes. To ensure the competitiveness of agricultural enterprises, the information should be acquired at low costs. New sensors solutions are necessary to meet this demand.
In agricultural production processes the crop height, the coverage level and the crop biomass density are important morphologic parameters for the assessment of crop stands. Based on them, expected crop yields can be appraised and the use of fertilisers and pesticides for the site specific crop management can be optimised. Furthermore, working parameters in harvesting machines such like velocity or the rotation speed of working units may be adapted according the time and site specific crop situation.
The measuring of crop height was performed under research conditions. Several approaches such as manual handling (straight edges), ultrasonic rangefinders (Hutching et al., 1990; Scotford and Miller, 2003) and radar techniques (Paul and Speckmann, 2002) have been reported for this. In addition, the coverage level of crops can be measured by optical methods (Reynier et al., 2004; Dammer, 2005). For surveying crop biomass distribution, manual methods (Gonzales et al., 1990; Lokhorst and Kasper, 1998), aerial photography (Pearson et al., 1976; Tucker et al., 1981) and vehicle based methods (Jaynes et al., 1995; Hansen and Jorhensen, 2001; Ehlert et al., 2003) have been discussed. Under practical farming conditions, surveying of spatial variability is possible by yield mapping in harvesters (Auernhammer et al., 1994).
In the last years new laser sensors for the distance measurement were developed and introduced in the market, mainly for industrial applications. The laser distance sensors are based on two measuring methods:
| • | Light delay time principle |
| • | Triangulation principle (Fig. 1). |
Applications of laser distance sensors in horticulture have already been reported for measuring the canopy volume and structure (Tumbo et al., 2002). For measuring of morphologic crop stand parameters under field conditions, the sensors are required to meet some specific demands. As crop parameters would be detected vehicle based on tractors, self propelled agricultural machines and tool carriers, sensors should be robust against atmospheric conditions (e.g., moisture, dust, variable temperatures, sunshine …) and mechanical vibrations. There is no need for a high distance measuring accuracy, less than 0.01 m has been found to be sufficiently. The effective measuring distance range should be not less than 1.5 m to detect crops in the relevant growth stages.
| Fig. 1: | Measuring principle of laser triangulation |
Furthermore, their utilisability should be given round the clock, ranging from darkness up to bright sunshine conditions. As every single measurement is a random result, the measuring frequency should larger than 1000 Hz to ensure a high usability for the readings.
Some of market available laser distance sensors are low cost products, have small light beam cross sections and are working at high frequencies. Thus, important prerequisites are given to deeply penetrate the crop stands deeply and to get clear reflection signals at a sufficient rate. For measuring morphologic parameters of crop stands like e.g. crop height or crop biomass density, the sensor should be moved in a constant distance above the ground. While horizontally movement, the beam is reflected from both crop surfaces and the soil. Conclusions about the morphologic dimensions of crop stands can be derived for an optimised crop management after a statistical evaluation of the measured reflection distances. It can be assumed that the measured reflection distances will decrease basically in the case of increasing crop biomass densities and plant heights with a high statistical probability (Fig. 1).
To investigate the basic measuring features of laser distance sensors for measuring morphologic crop parameters, the objective of this study was:
| • | Assessment of pre-selected and tested laser distance sensors |
| • | Test and assessment of a modified laser distance sensor according the |
| • | Accuracy of distance measurements, |
| • | Influence of inclination angle on accuracy, |
| • | Influence of crop species and illumination intensity on the rate of failure measurements |
| • | Relation between self indicated and not self indicated failure measurements |
| • | Measurements |
MATERIALS AND METHODS
Pre-selection and test of distance lasers: In a first step market available laser distance sensors were analysed for their suitability with a view to the above defined demands for the acquisition of crop parameters. The initial selection was based on the available technical data. Sensors which were chosen after pre-assessment are listed in Table 1.
As developed for industrial applications, the sensors are mostly working in the red light band. Due to the visible light band, the laser beam can be controlled and adjusted without any specific glasses. To prevent health risks for the operator, the lasers should have the maximum classification II.
To investigate the principal suitability of laser distance sensors for measuring of morphologic crop parameters listed in Table 1, investigations have been conducted in the vegetation periods 2003 and 2004.
Based on the measuring principle and the extremely irregular shaped reflecting crop surfaces failure measurements were expected. To analyse this aspect, each of the listed sensors was mounted (inclined) in front of the basic vehicle (tractor, tool carrier) and focussed on the crop stands.
| Table 1: | Collection and technical specification of chosen laser distance sensors |
| 1) triangulation principle 2) light delay time measuring principle | |
| Table 2: | Laser distance measuring errors under defined lab conditions |
| 1) matt black metal plate | |
Each sensor was installed considering its specific maximum measuring distance. Thereby the maximum plant heights were considered for preventing reflection distances below the minimum measuring distance. All sensors had a high measuring frequency in the kH-range, therefore only short time tests were conducted. Because of the known plant heights it was possible to identify apparent failure measurements and to consider them in the following assessments.
From the pre-tests it was concluded that the investigated laser distance sensors did show considerable functional problems. Mainly bright sunlight caused functional problems. All the tested sensors did not meet the demands for measuring morphologic crop parameters under field conditions. But nevertheless based on the low level and self indicated failure measurements, the sensor LASE ODS 1400 HT did show a potential. To improve the suitability of the sensor necessary modifications were performed (Ehlert, 2005).
Modified laser distance sensor: Based on the test results of the pre-selected distance laser sensors (according Table 1), an advanced laser sensor was developed and tested in 2005. The sensor was a triangulation type ODS 1600 HT 2select from the Danish company LASE. It was modified to meet better the demands for measuring morphologic crop parameters under field conditions. The main technical data were:
The sensor had an increased measuring distance, a higher measuring frequency and more power than the previous ODS 1400 HT. To investigate the potential of the laser distance principle, the sensor should work under sunny conditions. Therefore a sensor of the class 3 b was used.
To assess the suitability of the sensor, the following parameters were investigated:
| • | Accuracy of sensor distance readings, |
| • | Influence of inclination angle, |
| • | Influence of crop species and illumination intensity on the rate of failure measurements, and |
| • | Relation between self indicated and not self indicated failure measurements |
An optical bench was used to investigate the accuracy of the laser distance measurement (Fig. 2). The sensor readings were gathered by the universal digital voltage measurement unit DM 2020 (manufacturer: VEB Werk für Fernsehelektronik, Berlin) with a maximum error of 0.1%. The reference distances of the plant leafs (Ficus, grass, rape) were estimated by us of a ruler. To adjust the inclination angles (0; 30 and 60°) a three-jaw chuck with angular spacing was used. The plant leafs were glued on a black painted metal plate to ensure an aligned surface of the plant surface.
The pre-tests showed better results (low failure rate) for laser sensors under low illumination conditions (e.g., twilight). Therefore experiments were performed to estimate the failure rate depending on reflection media (grass, sand soil, concrete and granite road) and reflection distances (near the minimum measuring distance: 900 mm; mean measuring distance 1600 mm; near maximum measuring distance 2300 mm) for bright sunshine conditions.
| Fig. 2: | Optical bench for distance accuracy measurements |
| Fig. 3: | The influence of measuring distance on differences of readings (all values) |
| Fig. 4: | Example for demonstration of not self identified measurements |
The tests were performed at August 2nd, 2005 from 12.00 to 13.00 o'clock under full sun shine illumination. The beam of the laser sensor was perpendicular directed to the ground. The recording frequency was 5000 Hz.
Using a digital Lux-Meter "Minilux" the reflection intensity was measured in a distance of 1 m above the reflection medium. In previous investigations it was observed that in addition to the self identified and indicated failure measurements the laser distance sensors had not self identified failure readings additionally. Because of the known specific test conditions, it was possible to discriminate the not self identified failure readings.
RESULTS
Distance measuring accuracy: The distance measurements of the modified triangulation sensor ODS 1600 HT2 select for defined lab conditions (illumination about 0.1 kLux) resulted in standard errors in the range of only a few millimetres (Table 2). There were no clear tendencies for the impact of the inclination angles (0; 30 and 60°) and of the plant leaf surfaces of ficus, grass and rape on the accuracy of distance readings. In the tests the inclination angle was further increased to estimate the maximum inclination angle for signal receiving and distance measurements. A reflection of the laser beam was still registered by the sensor for inclined plant surfaces near 90°. For this extreme angle an accurate reference distance measuring was impossible.
To demonstrate the influence of the reflection distance on the sensor accuracy, the absolute values of the difference between sensor reading and reference distance were calculated. Figure 3 shows that the differences tend to result in higher values for increased reflection distance. Based on a linear regression the functional tendency between reflection distance and differences was calculated. Using these regressions the differences according Table 2 for the minimum (800 mm) and maximum (2400 mm) reflection distance were estimated. Based on both total differences, an increased difference from 1.1 to 3.1 mm for maximum reflection distance was indicated.
Influence of illumination conditions: Under the test conditions, the ground with sandy soil, concrete and granite was quite plain. Therefore all readings with a difference of more than 50 mm from the adjusted heights (900; 1600; 2300 mm) were assigned to not self identified failure measurements. For grass (maximum height 150 mm) all readings less 200 mm and more than 100 mm of the adjusted height were classified as not self identified failure measurements.
| Table 3: | Percentage of the self indicated (and not self identified) failure measurements under bright sunshine conditions depending on reflection medium and distance |
| 1)Example in Fig. 4 | |
| Table 4: | Differences between adjusted reflection distance and mean distance readings with not self identified (and without not self identified) failure measurements |
| 1) Example in Fig. 4 | |
The sensor worked with a high failure rate under bright sunshine conditions (Table 3). The indicated failure measurements do not result in problems. They are self indicated by the sensor and can be eliminated automatically. The rate of the self indicated failure measurements increased with higher reflection distances up to maximum of 72%. The not self identified failure rate was up to 20.3%. In Table 3 the corresponding mean values of the readings are shown. There is a clear tendency that the mean values of percentage of the self indicated and not self identified failure measurements depends on reflection medium and distance.
Comparing the measured mean reflection distances and the adjusted distances, a permanent underestimation of distances was observed (Table 4). The calculated differences ranged from -12 up to-116 mm depending on the reflection medium and the adjusted reflection distance. This phenomena can be explained by the specific working characteristic of the used sensor. Figure 4 demonstrates a section of the test with the maximum of not self identified failure measurements outside the band of 2250 to 2350 mm.
CONCLUSIONS
For the modified sensor version LASE ODS 1600 HT 2 select under lab conditions (illumination intensity of about 0.1 kLux), accuracies were obtained in the mm-range for the investigated plant leafs and inclination angles. Related to the rough soil-and plant conditions this would be a sufficient accuracy. Based on the made experiences, it can be concluded that bright sunshine in connection with intensive reflecting surfaces (e.g., sandy soils) without crop coverage were limiting the function.
A certain ratio of failure measurement is intrinsically tied to the measuring principle of laser triangulation. To cause a failure measurement e.g., the reflection point is covert by crop parts and therefore a reflection signal can not be registered by the receiver (Fig. 5). Moreover, crop surfaces may be extremely inclined so that the reduced reflection can not be registered from the sensor.
| Fig. 5: | Failure measurement caused by a) covering and b) inclination of crop surface |
A third reason is given if beam cross section targets two very different distanced areas e.g., canopy and soil. A certain ratio of failure measurements is tolerable under the prerequisite if either it can be identified and eliminated or the mean value of the readings is influenced negligibly.
Further intensive tests are required under defined field conditions to investigate the functional relations between the readings of laser distance sensors and the plant height, coverage level and crop biomass density. To assure the systematic approach, laser distance sensors in the class 3b should be used.
Provided, there are good correlations, the design of laser sensors should be continued. Mainly the specific demands for measuring of morphologic crop parameters under hard field conditions and the aspect of safety of work should be taken into account.
ACKNOWLEDGMENT
Thanks to A. Anlauff and U. Frank for their support in preparation and execution of the tests.