Patent Application
20080316491
The present invention relates generally to sensors, and in particular, to a sensor and method of detecting the condition of a turf grass.
Turf grass, such as a part of landscaping or on a golf course, can often vary greatly over a given section. While some variations in the turf grass may be visible, other variations may not be obvious to an individual observing the turf, or may be in their early stages and will soon develop into significant variations. Further, some sections of turf grass may be extremely large, making it impractical for an individual to inspect all of the sections of the turf.
More importantly, many conventional devices currently employed to inspect a section of crops have a number of disadvantages. For example, many conventional devices use collimated illumination to detect the condition of crops. However, such illumination requires sensors which can be costly. Also, conventional devices employing the sensors are installed on equipment which do not provide uniform samples of the crop to analyze. For example, a sensor attached to a piece of farm equipment which does not provide an even vegetation condition will lead to results which are less reliable. That is, such conventional farm equipment does not provide consistent length or physical orientation of the crop, eliminate dew drops on the crop, reduce the effects of tire tracks, etc.
Accordingly, there is a need for an improved sensor for turf grass and method of detecting the condition of a turf grass.
A method of detecting the condition of a turf grass is described. According to one aspect of the invention, the method comprises steps of attaching a sensor to a mower; traversing a section of turf grass; and processing the output of the sensor.
According to another aspect of the invention, a method comprises steps of attaching a sensor to a mower; providing an illuminating pattern of light to the turf grass; and receiving a reflected pattern of light which is coincident with the illuminating pattern of light.
According to a further aspect of the invention, a method comprises steps of providing a plurality of illumination devices; coupling a feedback circuit to adjust the output of each illuminating device of the plurality of illuminating devices; detecting the irradiance of each illuminating device of the plurality of illuminating devices; and adjusting the irradiance of an illuminating device of the plurality of illuminating devices.
A method of detecting the condition of a turf grass according to a further embodiment comprises the steps of attaching a sensor to a mower; sensing the condition of the turf grass; and detecting the geographic position of the sensor while sensing the condition of the turf grass.
Finally, a method of detecting the condition of a turf grass according to a further embodiment comprises the steps of attaching a sensor to a mower; sensing the condition of the turf grass; detecting the geographic position of the sensor while sensing the condition of the turf grass; processing the sensor data and the geographic position data; and creating a map.
A device for detecting the condition of turf grass is also disclosed. The device comprises an array of illuminating devices generating a pattern of illuminating light; a detecting device receiving a pattern of reflected light which is coincident with the pattern of illuminating light; a detecting device adapted to detect stray light from the array of illuminating devices; and a feedback loop controlling the array of illuminating devices. A system employing the device is also disclosed.
The system of the present invention also preferably provides automatic data transfer using wireless technology and a relational database with transaction support. Therefore, the system ensures data integrity and timely transfer of data. The sensed data detected by the sensors on the mower and GPS data received are coupled to a receiver 120 having an antenna 122 by way of a communication link 124. The receiver 120 could be coupled to or incorporated in a computer 128 in a base station 130. Such a base station could be in a garage or office of a golf course. Alternatively, the sensed data and GPS data could be coupled to a portable computer 132, such as a laptop computer using a wireless PC card, either directly by way of a wireless communication link 134 or by way of a wired communication link 136, or indirectly by a separate wired or wireless communication link 138.
Data from the turf sensor(s) 106 and 108 that is collected is preferably displayed using application software program that alerts the user to any new data sets that have become available since the last time the program was run. Such application software could be run on a processor of the data logger 110 in real time or otherwise, or could be run on a computer 128 or 132, or deferred. Once a data set is loaded into the application software, it can be viewed and manipulated any of a number of ways. The user can selectively filter data points based on the sensor ID, the state of the mower reels (active, inactive, etc), magnitude of the signal, time, or location. Time filtering is preferably employed for separating mowing events for a region in cases of multiple passes (e.g. double mowing). Further, a user specified transform can be applied to the filtered data, allowing after-the-fact calibration adjustment. A common use for this transform is “sensor balancing”, whereby it is assumed that the mean value of all filtered data for a particular region is the same for all sensors, and each sensor's transform is adjusted accordingly.
Application software employed by the present invention preferably provides gridding of geographic data and automatically checks for new data on the base station. The sensed data could be processed and presented in a number of ways, including but not limited to trend maps, raw data, gridding of various sizes, standard deviation analysis, “removing-the-mean” analysis, displaying only data with reels active, displaying only data inside fairway outlines, data with sensor offsets automatically removed, etc. In displaying raw data, each measured data point is shown in the proper size and shape representing that sampled area and location on a 2D map using any of a number of color scales. Details of any specific point, including transformed values, time, location, etc. can be displayed. In displaying gridded data, data can be binned into any grid spacing, and displayed as a map of squares, using various color scales. Statistics for the data included in any target grid can also be displayed.
The system of the present invention provides numerous analytical tools for displaying the condition of a turf grass. One of the challenges in analyzing turf is the removal of the natural diurnal cycle of the turf response from longer term trends that are due to stresses of interest. The application software of the present invention achieves this by “mean removal”, whereby the mean (or median) value for data in a particular region is computed and subtracted. A map of departures remains which, when color scaled appropriately, results in a map in which significant problem areas are readily visible.
Another challenge of the analysis of a turf grass is to highlight regions of otherwise healthy turf that have the beginnings of disease manifested by spots that are small in comparison to the area of one reading. The application software of the present invention achieves this by “MSD analysis”, in which the product of the mean and standard deviation is computed for gridded data, and mapped. This analysis separates locations that are in the early stages of disease, where for example high mean equals healthy turf, and high standard deviation equals spots, from areas in which the problem is in the latter stages and more widespread (i.e. lower means and standard deviations).
A third analytical tool employed by the application software of the present invention is that of trend maps, in which rates of change of normal data (or mean-removed, or MSD) are displayed in a gridded map. The grid spacing and time domain are user selectable. The time history of the grid square in question can be displayed. As an aid to turf management, the application software also supports overlays containing other space and time based data, such as pictures, regions, and points, along with associated text. Additional analyses can be used to try to amplify, or otherwise bring out, areas in the turf that are of declining or poor quality. A few more specific examples of analyses that may be used are: standard deviation of the data points within a gridded area, a two dimensional spatial Discrete Fourier Transform (DFT) of the data points within a gridded area, a two dimensional spatial Fast Fourier Transform (FFT), or a variogram analysis.
Because dirt, grime or grass clippings on sensors and in the sensor's field of view (FOV) could impede the operation of the system, it may be beneficial to mount the sensors such that they are up and out of the debris field, and also provide a smaller mechanical profile to reduce likelihood of collisions. There are a number of advantages to putting the sensors behind the mower. Because the grass has just been cut and consistently oriented, the grass is of uniform length and physical orientation (i.e. bentness). The act of mowing also eliminates dew droplets on the grass, and reduces the effect of tire tracks. The rear placement also keeps the sensors out of the operator's typical working view. The design and operation of the sensor will be described in more detail in reference to later figures.
The sensor also detects stray light from the LED array 502 in a feedback loop. In particular, stray visible light 540 and stray infrared light 542 are detected by feedback photodiodes 544. The orientation of the LED array and the feedback photodiodes will be described in more detail in reference to
The system of the present invention provides constant LED output control by sensing stray illumination light via feedback photodiodes, which allows a repeatable, stable measurement of the reflected light. The output control circuitry is preferably implemented in hardware, including the feedback and control circuit of
Alternatively to measuring the stray light output with photodiodes or other photodetectors, other techniques could be used to measure and create a signal proportional to the light output, such as: directly sensing the amount of light that is directed at the target (e.g. positioning a photodiode such that it intercepts a portion of the illumination light); collecting stray light by use of a light pipe, or; sensing one or more electrical parameters, such as the electrical current that is driving the illuminating LEDs, which is proportional to the radiant intensity of the illumination LEDs. In the case of measuring the LED current, the temperature would also have to be known due to the LED's significant temperature dependence. By utilizing measurements of both the current and the temperature a very good representation of the light output could be calculated. Or by controlling both the current and the temperature, the light output could be controlled to a known radiant intensity. Finally, one of the parameters could be controlled while the other is just measured. This signal or value, obtained in any of the preceding ways, could be used to either control the illumination or be used to normalize the reflected light value.
By employing active illumination of two narrow band wavelengths, the system of the present invention is not dependent on sun/sky/daylight/shade conditions, and therefore can be operated anytime, day or night, by providing its own reference light source. The modulation/demodulation of the present invention allows ability to detect very small signals in the presence of large noise signals (i.e. sunlight, sun fleck, power source).
In addition to providing a means for separating the desired reflected light from other light sources (sun especially), modulation is also used to separate all the different simultaneous measurements. Within a single sensor, a number of different modulation frequencies are used, (for example, one for the infrared, and a second for the visible). The received signals are then demodulated at the correct corresponding frequency. This allows simultaneous measurements of all wavelengths of illumination. This is important to ensure that when calculating the desired output (e.g. IR/Visible), the measurement outputs are both representative of the same area of turf. If the measurements were made one after the other (e.g. time multiplexed), then the measurement outputs could be representative of different areas of turf. This is a source of error in the desired output.
Different modulation frequencies are also used between adjacent sensors. This is done to prevent optical crosstalk between sensors. For instance, if adjacent sensors were using the same frequency for the infrared modulation, then if one sensor's light output was partially illuminating the adjacent sensor's field-of-view, it would be a source of error. Therefore four different frequencies are used for modulation/demodulation within the system. Specifically, one for odd-numbered infrared, a second for odd-numbered visible, a third for even-numbered infrared, and a fourth for even-numbered visible. The sensors are always placed in numeric order so there is a pattern of odd-even-odd-even, etc. So crosstalk is possible from odd to odd sensor, but there is sufficient optical separation to prevent this. Testing has shown that if adjacent sensors are illuminating and viewing the exact same area there is no detectable crosstalk.
The data logger preferably is capable of detecting the type of sensor, including but not limited to detecting a model, a version of a sensor, technology type, a manufacturer, features of the sensor, etc. to enable the data logger to properly communicate with the sensor.
The system also preferably provides sensor on control and sensor off control by way of a serial to parallel converter 560. System on control could enable the system to turn on with key switch, with vibration of mower, by a manual switch local to the system which uses a diode to block the local “on” switch signal from “turning on” the rest of the mower. System off control preferably provides an off sequence beginning with key switch turning off. Data is transferred, if possible, or a software time out occurs, at which time the data logger and sensors turn off. Further, a hardware watchdog could be employed to turn off the system in event of a software failure. The watchdog is preferably activated when key switch is turned off, and then shuts off the system if a software controlled signal stops toggling for a predetermined number of seconds.
The key switch of the mower could also be employed to control the on/off state of the system of the present invention. In order to meet the user's needs, the system must be reliable and require a minimum of user intervention to operate. Ideally, from the user's point of view, the mower mounted portion of the system would be an extension of the host implement. The turn-off sequence is viewed as being more important than the turn-on event because a failure in the turn-off sequence has the potential for greater consequences to the user's equipment and work flow. An example would be a dead battery on the mower. This would cause significant annoyance and inconvenience for the typical user. Although a failure in the turn-on event would cause a lack of collected data, such a failure would not affect the performance of the main intention of the mower, that is, to mow the target turfgrass.
The key switch signal is detected by connecting a wire into the key switch wiring of the mower. This signal is then connected to the data logger of the system. Advantages of detecting the key switch include simplicity of implementation and straightforwardness of use. The data logger only turns on when the mower is in use and the data logger only begins its turn-off sequence when the mower is turned off. By only beginning the data logger turn-off sequence when the mower is turned off, an additional advantage is realized in that the mower will normally be turned off when the mower is in a known locale with respect to the base station-specifically, a locale in which wireless communications between the data logger and the base station can be achieved.
An example turn-off sequence would be as follows. When the turn-off sequence is initiated by the mower key being turned off, the data logger checks whether communication with the base station can be achieved, and also whether the data logger has any data to transfer to the base station. If communication is achieved, but no data is available for transfer, the data logger and sensors turn off relatively quickly (e.g. 2 or 3 seconds). If communication is possible and data is available to transfer, the data is transferred. At the completion of the data transfer, the data logger and sensors turn off. If communication is not established, a software timeout is implemented in the data logger to limit how long the data logger attempts to connect to the base station. At the end of this timeout period, (e.g. a few seconds to a few minutes, although any appropriate time period could be used), the data logger and sensors turn off. Data still resident on the data logger is maintained until it is transferred to the base station.
Various other means could be employed to provide the information about whether to turn on or off the data logger. For example, vibration from the implement, heat from the exhaust manifold or pipe, or electrical impulses in the spark plug wires or other electrical device on the mower could be detected. Detecting vibration would allow the mower mounted portion of the system to be more autonomous, that is, less dependent on inputs from the mower for its operation. The vibration detection circuit could be mounted inside the data logger, making connection of the system to the mower more simple. Similarly, as the exhaust pipe of the typical mower exists from the rear of the mower, attachment of a sensor to the exhaust pipe could entail less wiring than to the key switch.
Because the system of the present invention could collect data from the sensors whenever the mower is moving and the GPS is providing differentially corrected position information, a method is needed to discriminate between when the turf grass is being sampled under consistent, “known” conditions and when it is not. For example, the system could collect data on a golf course as the mower is driving from the shed to the target fairway, over other fairways to reach the target fairway, is turning around in the rough of the course, etc. According to one embodiment of the invention, an electrical signal exists on the mower that indicates when the reels are in mowing position and the reels are actually mowing. For purposes of this document, this signal is called the “reels active” signal. A reels active signal, which senses when mowing is in progress, could provide a simple method of determining when data is being collected under the appropriate conditions to allow sorting of data between valid data and invalid data. Such a signal also enables a method of including/eliminating collected data offset in distance or time from the actual signal change. For example, if the sensors are viewing an area behind a forward-moving mower, when the reels are raised the sensors have not yet viewed the total mowed area. Likewise, at the instant when the reels are lowered, the sensors are not viewing mowed turf They will view the mowed turf some traversed distance later. So collected data marked as valid may be offset in time or distance to minimize this effect. The offset could be performed in either direction depending on the location of the sensors.
This signal exists on the mowers in the form of a reels solenoid signal. It is typically active (i.e. carrying an electrical signal of approximately 12V D.C) when the reels are in mowing position and the “mow” switch is engaged by the user. The “reels active” signal is typically inactive (i.e. carrying an electrical signal of approximately 0V DC) when the reels are not in the mowing position OR the “mow” switch is not engaged by the user. Use of this “reels active” signal allows the system to determine when the sampling of the turf grass was performed while the turf grass was in a known condition. Without this, the conditions of the turf grass during sampling is not known, which introduces additional ambiguities into the analysis results. The known conditions of the turf grass provided by the “reels active” signal include, but are not limited to: a consistent turf grass length over the entire fairway; the dew droplets having been removed from the turf grass blades; the turf grass has been rolled, or mechanically positioned, by the mower such that the effect of the mower tire tracks is significantly reduced, etc. To implement detection of this signal, electrical wiring is installed on the mower and connected to any of the “reels active” signals, (i.e. the front, rear, or any other “reels active” type signal), and this wiring is routed to the data logger where the state of this “reels active” signal is logged with the sampled data.
As shown in
The present invention provides a novel method of providing such an output without using any lenses, mirrors, or other focusing or collecting devices. Unlike prior art methods which often use collimated illumination and detection arrangements to achieve the above described goals, the system of the present invention uses coincident patterns of light.
In particular, as shown in
The same illumination/detection pattern with the viewed sample at a closer distance, represented by the dashed horizontal line, is shown in
Actual test results can confirm how well this arrangement works. For a distributed circular array 3.25″ in diameter, with a 12° half angle FOV aperture in front of the detector, the distance error in the ratio is less than 0.1% per 6″ change over a range from 42 to 60 inches. The error gets worse as the sample gets closer. For example, there is a 0.6% error in going from 36 to 42 inches. This worsening effect is evident by looking at
It should be clear that as the distributed source is reduced in size the error is reduced. It is therefore advantageous to design the illumination array to be as small as possible and located as close as possible to the detector. If the source could be reduced to a point and located at the exact same location as the detector, the illumination and detector FOV would be perfectly coincident resulting in no error in the ratiometric output as the distance is varied. Because this is not practical, it is the most advantageous to make the ratio of viewing distance to array size as large as possible, while taking the disadvantage of distance into account.
As can be seen in
In addition to mechanical support, the arrangement shown here also provides electrical separation between the illumination and detection sections of the sensor. This arrangement also provides proper light blocking and containment features. Since the second circuit board 1006 is mostly solid except for the LED holes, it blocks most of the reflected light signal from reaching the feedback photodiodes 1016. The arrangement of the circuit boards also keeps most of the stray light from the LEDs contained so it can be received by the feedback photodiodes. Without the light blocking feature, the reflectance of the sample could affect the output illumination by adding a small amount of light proportional to the sample reflectance to the direct stray light that is intended on being collected by the feedback photodiodes.
By careful placement of the detectors used for feedback within the array of LEDs, offsetting optical effects are used to help even out the feedback contribution from each LED in the array. As a source emits light its irradiance falls off with the square of the distance (i.e. Ee∝1/d2). Therefore LEDs close to the detector have a larger irradiance than those further away. The present invention shows a technique to minimize this undesired effect. Because of this particular arrangement, the LEDs that are close to the detector are also at a significant angle with respect to the normal angle of the detector. A standard photodiode detector has an angular sensitivity proportional to the cosine of the incident angle (i.e. S∝cos(θ) ). So the partial output from the detector is the product of the incident irradiance from a particular source and the sensitivity of the detector to a source from that particular angle (i.e. Oi∝cos(θi)/di2). The total output of the detector is proportional to the sum of all the sources in the field of view (i.e. O∝O1+O2+ . . . +On). Only the LEDs within the detector's field of view contribute to its output.
The distance d shown in
d
i=x·cos(Θi)+√{square root over (x2·cos2(θi)−x2+ri2)}
where r is the radius of the circular array, and x is the distance from the center of the array to the detector.
Substituting into the above equations gives the following for the partial output formula:
and the total output formula is:
In the case of a single circular array (all at one radius) of LEDs, a very good equalization can be done by proper placement of the detectors with respect to the LEDs. For instance with r=1″, and x=0.61″ and including all LEDs within a 120° total field of view, the maximum partial output in normalized units is 1.032, which occurs at 42°. The minimum partial output is 0.973 at 60°. This is only a difference of −5.7%. The standard deviation of 5 sources evenly spaced from 0 to 60° is only 0.78%. A plot of the distance and normalized intensity is shown.
When multiple circular arrays are used at several different radii, the technique is not as successful, but yet still yields favorable results. For instance, with r1=0.97″, r2=1.26″, and r3=1.55″, and x=0.64″, the overall maximum point to point weighting difference is −61%. This is from row 1 at 48° to row 3 at 60°. Within a single row, the results are more favorable. Maximum weighting differences are: row 1=−6.5%, row 2=−14%, and row 3=−23%. Therefore if the weighting between rows is not important, the differences are small. For instance, if each row contains a different color LED, and each color is controlled independently, then equal weighting between rows is not important.
By using two or more detectors, the outputs can be combined to view most or all of the LEDs in the array. As shown in the embodiment of
While the benefits of a circular array have been addressed here, other configurations could also be used. A linear array may be desired to minimize the overlap of successive samples of the sensor. Other shapes or techniques may also be used. Although the even-weighting advantages of the circular arrangement were discussed, the benefits of other arrangements may be deemed more important than the precision of the light output.
Various technologies are used to make reports available in a convenient and timely fashion. Refering to
The positioning system outputs an electronic Pulse Per Second (PPS) that indicates the precise time when a location measurement occurs. The PPS triggers an interrupt to inform the software of the event. By definition, the PPS occurs once at the beginning of every second even though the positioning system outputs location records 10 times per second. In order to synchronize the sensor data with the location records, the software preferably generates 9 pseudo PPS pulses for every 1 PPS as shown in
Each sensor data point and each location record is also marked as to whether it was collected coincident with a PPS as shown in
The following method is preferably used to get the location deque and sensor data deque synchronized. All sensor data and location records are discarded for 1 second. This flushes all data out of the deques and from the buffers in the RS-232 link. All sensor data and location records are discarded until a marked location record is received. All sensor data and location records are then discarded until 3 location records have been discarded. This is to ensure that the next few location records and sensor data will be between PPS pulses. All unmarked sensor data and unmarked location records are discarded. When the next marked sensor data or marked location record is received, they are pushed into the left side of the deques. The right end of the deques are now synchronized. If the deques ever become unsynchronized as shown in
It can therefore be appreciated that the new and novel turf sensor and method of detecting the condition of a turf grass has been described. It will be appreciated by those skilled in the art that, particular the teaching herein, numerous alternatives and equivalents will be seen to exist which incorporate the disclosed invention. As a result, the invention is not to be limited by the foregoing embodiments, but only by the following claims.
This application claims priority to US Patent Application Ser. No. 60/491,780, entitled SENSOR, filed on Aug. 1, 2003 by at least one inventor of the present invention, and to U.S. patent application Ser. No. 10/655,749, filed on Sep. 5, 2003, both of which applications are incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| Parent | 10655749 | Sep 2003 | US |
| Child | 12008402 | US |
