Patent Grant
7448548
The present invention relates to the counting of traffic such as people using light beams.
The problem of counting people traffic with beams of light is known. Typically, people counters are used at doorways in places of public accommodation such as stores and other buildings to roughly count occupancy and correspondingly control ventilation, heating and air conditioning systems. People counts have other purposes as well; in retail establishments, people counters may be used in store aisles or other locations to determine interest in those particular areas, and may be used to generate statistics such as total traffic through a store or particular aisle, and to perform data mining when combined with other data, e.g., by using register transaction counts to test the efficiency with which sales are being consummated from visiting potential customers in the store or particular aisles.
The most common approach to people counting has been to produce a light beam across a passageway, to count the number of persons passing through the passageway, represented by the number of times the light beam is broken.
A battery powered people counting system that uses this broken-beam approach is described in U.S. patent application Ser. No. 10/635,403, filed by the applicant hereof. This patent application describes an object counter that uses an infrared (IR) light source that generates and detects brief pulses, using very fast emitter/sensor devices and reducing the data cycle to approximately 20 microsecond of IR emission for every 1/16-second operation. This is a power on-to-off ratio of approximately 1 to 300, permitting low power consumption and long-term battery-powered operation. The applicant's U.S. Pat. No. 6,721,546, which is hereby incorporated herein by reference, describes additional low-power techniques that use a processor for a brief period of time.
Single-beam people counters such as disclosed in the above patent, can readily track a beam break, but cannot readily determine the direction of movement of an object or a person that caused the beam break. When counting movements through separate entrance and exit doors in a building, the location of the beam indicates whether the person is entering or exiting. However, when monitoring a passageway that is bi-directional, or where a common door is used for entry and exit, a single beam is not typically able to discriminate between the entry of a person and the exit of a person. For such applications, therefore, it has been known to use a directional people counter.
Directional people counters a retro-reflective target and two narrow beam emitter/sensor assemblies to produce two physically separated beams. The beams must be narrow enough such that the two sensors do not see each other's beams as they are reflected back from the retro-reflective target. Referring to
It is necessary in these typical directional object counting systems, that the emitted beam from the emitters EA and EB be sufficiently narrowly focused that, when mirror 14 is properly positioned, the respective beams A and B from EA and EB will illuminate only one of the corresponding sensors SA and SB. Thus, the shaded area in
Directional people counters thus detect direction of motion by the sequence in which beams are broken and signal lost at sensors. If direction 16 is the direction of entry and direction 18 is the direction of exit, then a break of beam 16 first means an entry, and a break of beam 18 first means an exit.
It will be noted that this method of dual-beam people counting requires optically precise emitters EA and EB, that emit a beam with a relatively narrow aperture angle α, and optically narrow field of view sensors, so that the field of view of sensor SA cannot see stray light from emitter EB emitter and the field of view of sensor SB cannot see stray light from emitter EA. If the field of view and aperture angle α of the sensor and emitter are excessively large for the application, then the beams A and B returning to sensors SA and SB will activate both sensors, as shown in
Typically the width of the passageway is several feet and the emitter-sensor center-to-center separation is only a few inches. As a result an emitter beam divergence of far less than 30 degrees would result in both sensors having a view of both emitters. In this circumstance, the signals received at the sensors SA and SB will be a function of the signals transmitted from both emitters EA and EB, and as a result, both beams EA and EB must be broken before either sensor will lose signal. Thus sensors SA and SB will lose signal simultaneously or nearly so, and only when both beams are broken, and it will be difficult to determine the direction of motion because the beam are not clearly and unambiguously broken at different times, as is the case when the beams have a sufficiently narrow aperture angle as shown in
The reason that the beam generating/sensing assemblies EA/SA and EB/SB are separated by only a few inches is that a smaller package is the better for object counting applications, as object counters are typically mounted on door frames and walls. Thus, the emitter-sensor separation, and the width of the passageway being monitored, are relatively fixed. As a result, beam emitter/sensor assemblies must have particularly narrow beams and fields of view for such applications. This means the beam generating and sensing assemblies are optically precise and complex in design, with a lens and collimator required to produce the small viewing angle necessary so that the sensors SA and SB do not see the beam from the opposite emitter EB and EA. These precision optics result in a high manufacturing cost.
The need for a narrow field of view sensor also increases power requirements on the IR emitter. The power emitted by the IR emitter, combined with the sensitivity of the IR sensor, determine the sensing range of the assembly. The more IR power emitted, the greater the range will be for a given IR emitter sensitivity. However, the requirements of a retro-reflective dual beam directional object counter, require sensors with a narrow field of view and a consequently lower sensor sensitivity. That lower sensitivity of the sensor, must be compensated by a higher power IR emitter to achieve a desired sensing range. The power required to operate such a system is typically too high for battery powered operation for reasonably long periods. As a result, wires must be used to supply electrical power to the beam counting system, and to communicate beam break sequence data to a location where it can be incorporated into a higher-level application such as retail traffic monitoring. Wiring costs are high in many installations and often contribute more to overall cost than the beam sensor.
It is an object of the present invention to provide an accurate directional people counting system that does not require precise optics and the attendant expense therefor, and which can operate on battery power for suitably long periods of time thus eliminating the need for wiring to a central location.
In accordance with principles of the present invention, these objects are met by a directional object counter that uses two or more light sources to generate light paths, and one or more sensors to detect the light, in which the light sensor receives light from both sources. Although both light sources illuminate the sensor, the manner in which the illumination is performed permits a processor connected to the sensor to determine whether the first source is or is not illuminating said light sensor, independently of whether the second source is illuminating the sensor. Thus, the processor can count movement of an object through the light paths in an identified direction based upon those determinations.
In the disclosed specific embodiment, there are two light sensors, and the processor separately determines whether the first light source is illuminating the first light sensor and whether the second light source is illuminating the second light sensor, to establish two separate light paths, so that movement of an object through those light paths can be counted, by detecting blockage of one light path and then the other, followed restoration of the light paths.
In this particular embodiment, the light sources generate light pulses, and the processor detects whether a light source is illuminating the light sensor based upon the reception, or lack thereof, of the pulses. The pulses used in the specific embodiment described herein are a pulse generated by the first light source, followed by a pulse generated by the second light source, followed by a pulse generated by the first light source. With pulses formatted this way, the processor can detect that light generated by the first light source is illuminating the light sensor based upon the receipt of a light pulse followed by a light pulse two pulse widths later, and can detect that light generated by the second light source is illuminating the light sensor based upon the receipt of a light pulse followed by a light pulse one pulse width later, or based upon the receipt of a light pulse not followed by a light pulse two pulse widths later.
Other pulse-based discriminations are also possible. For example, the light pulses can comprise a long pulse generated by the first light source, and a short pulse generated by said the second light source. In this case the processor can determine that the first light source is illuminating a sensor based upon the receipt of a light pulse that continues for a time longer than the short pulse, and the processor can determines that the second light source is illuminating said sensor based upon the receipt of a light pulse that continues longer than the long pulse, or a light pulse that continues for a time longer than the short pulse but shorter than the long pulse.
The specific embodiment described below uses infrared light, but other forms of directional radiant energy may also be used.
The invention permits low-power operation of a directional object counter, sufficiently low power to use a battery as a power source. Although the use of battery power is not required for all aspects of the invention, it is an independent aspect of the invention to provide a battery powered directional object counter.
The low-power operation provided by the invention, combined with the independence of the pulse discrimination from the pulse generation, also enables the light sources and light sensors to be positioned on opposite sides of a passageway without wiring connecting them across the passageway. Although this particular placement is not required for all aspects of the invention, it is an independent aspect of the invention to provide a directional object counter in which the light sources and light sensors are positioned on opposite sides of a passageway, without the use of a wired connection between the sources and sensors.
The above and other objects and advantages of the present invention shall be made apparent from the accompanying drawings and the description thereof.
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and, together with a general description of the invention given above, and the detailed description of the embodiments given below, serve to explain the principles of the invention.
As elaborated in the above-referenced patent application filed by the applicant hereof, a single beam object counting device, operating in a pulsed fashion, with an off/on ratio of 300, has a small enough average power supply current that a reasonably small battery can provide the required operating power for the device for multiple years. This pulse operation method is uniquely modified herein for use in a dual beam, directional object counter.
The specific implementation of this pulse operation has the unexpected positive consequence of permitting relaxation of the design requirements for the emitters and sensors in a dual-emitter directional object counter. In particular, the aperture angle of the emitters and field of view of the sensors may be wider than is permitted in conventional designs, such that both emitter beams arrive at and are within the field of view of both sensors, which is the condition discussed above referencing
Specifically, as illustrated in
It will be noted that the emitters EA, EB and the sensors SA, SB are located on opposite sides of the passageway, as opposed to being co-located on the same side of a passageway and opposed by a mirror on the opposite side of the passageway as is the case in the prior art systems illustrated in
Emitters EA and EB are electrically controlled by a control circuit 24, which utilizes a clock 26 to periodically generate pulsed emissions from emitters EA and EB, in a manner to be discussed below. Control circuit 24 is thus a low-power circuit utilizing pulsed transmission principles such as are disclosed in the above referenced patent application, and may be operated for long periods of time on battery power from a battery 28.
Sensors SA and SB are similarly electrically controlled by a controller 30, which utilizes a clock 32 to periodically “wake up” the sensors and attempt to detect pulses transmitted from emitters EA and EB. As described in the above-referenced patent application, clock 32 enables sensors SA and SB on a periodic basis with a period that is slightly shorter than the period between transmissions from emitters EA and EB established by clock 26. Thus, sensors SA and SB will “wake up” just prior to an expected pulse transmission from emitters EA and EB. If pulses are detected during the brief “wake up” period of sensors SA and SB established by clock 32, then sensors SA and SB will remain enabled for a period sufficient to capture the transmitted pulses, and boolean variables in controller 30 (herein identified as A and B) will be set to reflect whether beams A and B are broken or unbroken based upon the pulses captured by sensors SA and SB.
On a periodic basis, clock 32 will “wake up” state machine 34 of the controller 30, which will invoke a pass through logical steps (detailed below with reference to
On a periodic basis, the resulting object/people counts will be transmitted, preferably wirelessly by a wireless transmitter 36, to a remote data collection system.
As a consequence of the periodic nature of these various functions of controller 30 and transmitter 36, as explained in the above-referenced patent application, control circuit 30 and wireless transmitter 36 are low-power circuits due to their use of low duty cycle operation, and may be operated for long periods of time on battery power.
As illustrated in
This three pulse cadence permits sensors SA and SB to determine whether a beam is broken, and identify the broken beam, as follows. When either sensor receives a pulse during a “wake up” period of controller 30, then the controller 30 determines whether that pulse is followed by a second pulse received at either sensor, and then by a third pulse received at either sensor.
In the event a sensor receives a pulse followed by two subsequent pulses, then the sensor must be receiving beams A and B from both emitters EA and EB. In the event the sensor receives no second pulse but receives a third pulse, then the sensor must be receiving beam A from emitter EA but beam B from emitter EB is blocked. In the event the sensor receives a first pulse but no second or third pulses, then the sensor must be receiving beam B from emitter EB while beam A from emitter EA is blocked.
A technique of generating three pulses, first from EA then from EB then from EA again, thus permits each sensor to discriminate between receiving either or both emitter beams. The collection of data can thus proceed. For ease of reference, these three pulses will be identified hereafter as EA1, EB2 and EA3.
For an implementation of this scheme, a pulse width was chosen to be 40 microseconds. This width was chosen to allow for detection flutter. In order to synchronize the sensor side to the emitter, a detection process starts whenever either sensor senses a pulse during a “wake up” period of controller 30. The first detected pulse may be any one of the three pulses, i.e., it may be EA1 or EA3 from EA, or EA2 from EB. When a first pulse is detected by either sensor, three measurements of the IR sensor data SA and SB are made at three times relative to the first detected pulse. As illustrated in
In the event the first detected pulse the first pulse from emitter EA, EA1, then T1, T2 and T3 will occur as shown in
It will follow therefore, that any time a signal is present at T1 and T3 the corresponding sensor is in view of emitter EA. For the purposes of monitoring traffic flow, the case of interest is whether sensor SA is exposed to emitter EA, so the reception state of sensor SA is checked only at times T1 and T3, and only if reception occurs at times T1 and T3, sensor SA is determined to be receiving an unbroken beam from emitter EA.
The logic for determining if sensor SB has a view of EB, accounts for two possibilities, the first being that pulse EA1 starts the T1, T2, T3 read cycle, and the second being that pulse EB1 starts the read cycle. When emitter EA is illuminating sensor SB, pulse EA1 starts the measurement sequence, and EB2 will be viewed by SB at T2. When emitter EA is blocked from illuminating sensor SB, and EB2 will starts the T1, T2, T3 measurement sequence, and EB2 will be viewed by SB at T1, but there will be no signal at T3, that is, there will be a signal at time T1 and there will not be a signal at T3. Using this logic, it is possible to determine beam status for beam B between EB and SB, as follows: if there is signal at time T2, or there is signal at time T1 but not at time T3, then sensor SB is determined to be receiving an unbroken beam from emitter EB.
Each of
It will be appreciated that the particular combinations of beam breaks illustrated in
The table of
Having thus established that the arrangement described above accurately reflects, for two cross-illuminating beams, which of the beams is broken and unbroken, focus may now turn on the logic for determining whether a particular sequence of beam breaks should be considered an entry or exit from a passageway.
In a pass through the state machine of
The state machine of
A: boolean (yes/no) variable—indicates beam A was blocked during the last “wake up” cycle of sensors SA and SB.
B: boolean variable—indicates beam B was blocked during the last “wake up” cycle of sensors SA and SB.
In Cycle: boolean variable—indicates whether the recent blocked/unblocked activity of beams A and B indicate a “cycle” of activity that is suggestive of an object/person passing through the beams.
Change A: boolean variable indicating whether the A beam has changed condition during the current cycle.
Change B: boolean variable indicating whether the B beam has changed condition curing the current cycle.
First: boolean variable having the values A or B, indicating whether the first beam change detected during the current cycle was blockage of the A beam or blockage of the B beam.
Last: boolean variable having the values A or B, indicating whether the last beam change detected during the current cycle was blockage of the A beam or blockage of the B beam.
DBF: counter used to determine whether an indicated unblocked condition of the A and B beams is genuine or an artifact of spurious radiation and/or reflections.
In a first step 100 of
If, at some point, an object breaks one of the beams, a cycle will start. For example, if the B beam is blocked while the A beam is unblocked, then processing will go from step 104 to step 106 to step 108, where the InCycle variable will be set to indicate a cycle has commenced, and the First variable will be set to indicate that beam B was broken first. Similarly, if the A beam is blocked while the B beam is unblocked, then processing will go from step 104 to step 110, where it is determined the B beam is unblocked, and then to step 112, where the InCycle variable will be set to indicate a cycle has commenced, and the First variable will be set to indicate that the A beam was broken first. Thereafter, a cycle will have begun and processing will take a different path from step 102.
The logic described above includes inherent error checking. Specifically, if both beams are broken at the same time, this indicates an error condition rather than a trackable object movement. In such a case, processing will move from step 104 through step 110 to step 114, in which all flags and the DBF counter will be cleared. So long as both beams are broken, no cycle will start, but when a beam becomes visible again, as long as only one beam becomes visible, a cycle will start as described in the previous paragraph.
Once a cycle has started as described above, passes through the state machine of
As long as one of the beams is blocked, a cycle will proceed through steps 102-118-120 or 102-118-124-126 as described above, thus verifying in each pass that a beam is still blocked and noting the last beam that was blocked. However, once both beams appear to be visible, processing while in a cycle will pass from step 118 through step 124 to step 130, which increments the DBF (dual beam flash) counter. In step 130, the DBF counter is incremented by one, and then in step 132 the value of the DBF counter is checked. Initially, the DBF counter will have a value of zero (as a result of a reset in one or more of steps 114, 122 or 126), and so DBF will take a value of 1 during the first visit to step 130 of a given cycle. As a result, the first time in a cycle that step 132 is reached, the value of DBF will be less than 4 and the pass through the state machine will end. If the beams remain unblocked, however, in the subsequent passes through the state machine, the DBF counter will be incremented to 2, 3, 4 and 5, and when the DBF counter reaches a value of 5, processing will continue from step 132 to step 134, which is indicative of a “good exit” from a cycle. This sequence ensures that a temporary condition of both beams being apparently visible, which can be caused by spurious radiation and/or reflections, will not cause a false count. Only if both beams are visible for five passes through the state machine, will there be a good exit from a cycle.
Once there is a good exit from a cycle, processing continues from step 134 to steps 136-150 which evaluate whether the detected beam activity in the cycle is indicative of a proper object count.
A first criterion for a countable object movement is that a change has been seen in both the A and B beams. Thus, in step 136 the change A flag is evaluated and if it is not set, the cycle is aborted in step 138 (by proceeding to step 114 and resetting all flags and the DBF counter). Similarly, if the change A flag is set then in step 140 the change B flag is evaluated and if it is not set, the cycle is aborted in step 138. If both A and B have changed during the cycle, then processing continues from step 140 to steps 142-148 where the next criterion is evaluated.
The second criterion for a valid object movement is that the first beam change be different from the last beam change. Thus, in step 142 the First flag is checked. If the First flag indicates that the A beam changed first, then in step 144 the Last flag is checked to determine if the B beam changed last. If not, in step 138 the cycle is aborted, but if the B beam changed last there was a good cycle and in step 146 a B count is made (a B count indicates an object apparently passed through the beams, leaving the B beam last). Similarly, if in step 142 the First flag indicates that the B beam changed first, then in step 148 the Last flag is checked to determine if the A beam changed last. If not, in step 138 the cycle is aborted, but if the A beam changed last there was a good cycle and in step 150 an A count is made (an A count indicates an object apparently passed through the beams, leaving the A beam last).
It will be appreciated from the foregoing that the present invention provides an effective and robust object/people counting function using two pulsed beams that are both detectible by each of two sensors.
While the present invention has been illustrated by a description of various embodiments and while these embodiments have been described in considerable detail, it is not the intention of the applicants to restrict or in any way limit the scope of the appended claims to such detail. Additional advantages and modifications will readily appear to those skilled in the art.
For example, it will be noted that the second pulse from emitter EA eliminates ambiguity in cases where both beams are broken, and then one beam A or B is unbroken. Without a second pulse from EA, it would be difficult to determine which beam became unbroken unless the relative timing of the emitted pulses could be determined at the sensor. Such would be readily possible were the emitters and sensors on the same side of the passageway. However, in the embodiment illustrated above, no relative timing information is transferred from the emitters to the sensors, to avoid the need for a common clock for the emitters and sensors, and permit them to be located opposite one another in a passageway. However, an alternative embodiment might transfer relative timing information so as to enable the determination of which pulse is transmitted based upon the timing information.
It would also be possible, using only one pulse per emitter, to determine which of two beams is unbroken beam, from the relative timing of the newly-received pulse and previously detected pulses prior to the beam blockage, but the required timing information may be difficult to maintain during a protracted beam blockage.
Another alternative to producing a third EA pulse, would be to produce IR pulses at EA and EB that are different widths, i.e., “short” and “long”, such that from pulse width alone the source, either EA or EB, could be determined. Specifically, if the “long” pulse is transmitted first, receipt of light for a period at least as long as a “long” pulse would indicate that the source generating “long” pulses is unblocked. Receipt of light for a period at least as long as a “short” pulse but not as long as a “long” pulse, or for a period longer than a “long” pulse, would indicate that the source generating “short” pulses is unblocked. (Either the “short” or “long” pulse could be transmitted first in this approach.) This noted, various sources of timing inaccuracy would require a relatively long “long” pulse to reliably discriminate between the “short” and “long” pulse, potentially, the “long” pulse would need to be longer than two “short” pulse widths. Specifically, background IR components for ambient lighting result in pulse-width flutter in the sensed sensor signals SA and SB. This pulse-width flutter is increased by the microprocessor's cycle time for capturing a sample, which would require further differences in the pulse widths of EA and EB. The total energy requirement of the system is directly proportional to the total time that the IR emitters are on each cycle, therefore, it is desirable to reduce the on time of the IR emitters to a minimum. Nevertheless, the use of a “short” and “long” pulse may be an effective alternative approach to using two pulses on one of the emitters, particularly if the “long” pulse can be less than twice the length of the “short” pulse while maintaining reliability, in which case this alternate approach might achieve lower power consumption than the two-pulse approach described herein.
The invention in its broader aspects is therefore not limited to the specific details, representative apparatus and method, and illustrative example shown and described. Accordingly, departures may be made from such details without departing from the spirit or scope of applicant's general inventive concept.
| Number | Name | Date | Kind |
|---|---|---|---|
| 4874063 | Taylor | Oct 1989 | A |
| 5519784 | Vermeulen et al. | May 1996 | A |
| 6712269 | Watkins | Mar 2004 | B1 |
| 20030179127 | Wienand | Sep 2003 | A1 |
