LOCAL MEASUREMENTS USING MULTIPLE MOVING SENSORS

Abstract

A method of controlling sampling at a plurality of mobile ambient sensors, comprising: providing a plurality of ambient sensors, each sensor moving periodically along a predefined route; receiving one or more desired measuring locations; and assigning to a first one of the sensors a first sampling location, which is at or near the desired measuring location.

Claims

1. A method of controlling sampling at a plurality of mobile ambient sensors, comprising: providing a plurality of ambient sensors, each sensor moving periodically along a predefined route; receiving one or more desired measuring locations; and assigning to a first one of the sensors a first sampling location, which is at or near the desired measuring location, wherein the assigned sampling locations are output in the form of an instruction to be forwarded to the sensors.

2. The method of claim 1, wherein said assigning further comprises: assigning to a second one of the sensors a second sampling location, which is at or near the same desired measuring location.

3. The method of claim 1, wherein said assigning includes setting a sampling periodicity to apply at the first or second sampling location.

4. The method of claim 3, wherein: each of the sensors is associated with a predefined refractory period, which must elapse between two consecutive sampling instances; and the sampling periodicity is set in view of the refractory period.

5. The method of claim 1, wherein: each of the sensors is associated with a predefined refractory period, which must elapse between two consecutive sampling instances; and at least two sampling locations are assigned to one sensor, wherein the separation of the sampling locations is selected in view of the refractory period and the route of the sensor.

6. The method of claim 5, wherein the at least two sampling locations correspond to respective desired measuring locations.

7. The method of claim 1, further comprising: receiving, from each sensor or an entity associated with the sensor, data representing a route and/or a refractory period of the sensor.

8. The method of claim 1, which is performed by a controller not associated with any sensor, wherein data representing the assigned sampling locations is distributed to each concerned sensor by wireless communication.

9. The method of claim 1, wherein the sensors are mounted on geolocation-enabled vehicles, in particular timetabled geolocation-enabled vehicles.

10. The method of claim 9, wherein one vehicle carries multiple sensors, to which sampling locations are assigned independently.

11. A controller operable to control a plurality of mobile ambient sensors, each sensor moving periodically along a predefined route, the controller comprising: an interface configured to receive one or more desired measuring locations; and processing circuitry configured to assign to a first one of the sensors a first sampling location, which is at or near the desired measuring location, wherein the assigned sampling locations are in the form of an instruction to the sensors.

12. A computer program comprising instructions to cause a controller to perform the method of claim 1.

13. A data carrier with the computer program of claim 12.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Aspects and embodiments are now described, by way of example, with reference to the accompanying drawings, on which:

[0023] FIG. 1 is a flowchart of a method for controlling sampling at a plurality of mobile ambient sensors according to an embodiment of the invention;

[0024] FIG. 2 is a schematic drawing of sensors moving along predefined routes which overlap with desired measuring locations;

[0025] FIG. 3 shows a vehicle carrying two ambient sensors and is in wireless communication with a stationary controller; and

[0026] FIG. 4 shows example vehicles suitable for carrying mobile ambient sensors.

DETAILED DESCRIPTION

[0027] The aspects of the present disclosure will now be described more fully hereinafter with reference to the accompanying drawings, on which certain embodiments of the invention are shown. These aspects may, however, be embodied in many different forms and the embodiments should not be construed as limiting; rather, they are provided by way of example so that this disclosure will be thorough and complete, and to fully convey the scope of all aspects of invention to those skilled in the art. Like numbers refer to like elements throughout the description.

[0028] FIG. 1 is a flowchart of a method 100 for controlling sampling at a plurality of mobile ambient sensors according to an embodiment of the invention. In what follows, the method 100 is will be described when implemented by a controller 320 of the general type shown in the right portion of FIG. 3.

[0029] The controller 320 may be a programmable general-purpose computer equipped with or connected to a communication interface towards the ambient sensors to be controlled. The controller 320 in this embodiment does not move with any of the ambient sensors during the execution of the method 100. In FIG. 3, it is shown located in a building. It is well known that a general-purpose computer includes memory, processing circuitry 324 and a data interface for inputting and outputting data. While the communication interface is drawn simplistically in FIG. 3, as a rooftop antenna 322 directly communicating with the ambient sensors, it is understood that the communication interface may be a connection to a cellular network, where the actual radio link to a sensor is established from a radio base station managed by a network operator. As mentioned above, the communication interface towards the ambient sensors is not essential to allow execution of the method 100; the programming of the sensors in accordance with the method's 100 output data may be entrusted to another party.

[0030] Referring to a first step 110 of the method 100, a plurality of ambient sensors 220, 222, 224, 226 are illustrated in FIG. 2. Each sensor moves periodically along a predefined route. The periodical movement of the sensors may include repeated linear movement from an initial point to a final point, reciprocating movement (back and forth) between two endpoints, or cyclic movement along a closed curve. Here, sensor 222 moves along a first route 230, sensors 220 and 226 move along a second route 232, and sensor 224 moves along a third route 234. For clarity of the drawing, only portions of the first, second and third routes 230, 232, 234 are shown in FIG. 2. It is seen that some segments of the routes 230, 232, 234 coincide or run parallel. Because measurements in locations at or near these segments can be effectuated by a freely selectable one of the corresponding sensors, the method 100 has room to optimize the precise sampling locations and thereby the relative timing.

[0031] The sensors 220, 222, 224, 226 are mounted on geolocation-enabled vehicles, such as the truck, bus and construction equipment shown in FIG. 4. The vehicles may be conventional or autonomous (driverless). Generally speaking, “geolocation-enabled vehicles” includes vehicles and vessels with a Global Navigation Satellite Systems (GNSS) receiver, a cellular transceiver positionable in relation to base stations, a wireless local area network receiver positionable on the basis of a known access point location, an inertial measurement unit or other positioning equipment. Sensors which are associated with a positioning functionality are controllable on the basis of location. For example, each sensor may be configured to sample when it receives a signal from the vehicle, on which it is mounted, that it is currently at or near a sampling location assigned to the sensor. Alternatively, the sensor may receive updated geolocation data from the vehicle (or poll the vehicle for such data) and evaluate it against the sampling location(s) assigned to it. As a further alternative, the sensor may be geolocation-enabled in itself; at a possibly higher per-sensor cost, this avoids the reliance on the vehicle's positioning resources and the need to interface with these.

[0032] In the present example, the bus 310 shown in the left portion of FIG. 3 acts as the geolocation-enabled vehicle. The bus carries two sensors 220a, 220b which may for example be urban air quality sensors designed to measure nitrogen oxide pollutants (NO.sub.x) and particulate matter (PM), respectively. The sensors are connected to an internal controller 312 of the bus 310. The internal controller 312 serves as an interface towards resources shared by the bus 310, such as electric power, compressed air, a time signal, positioning services, data connectivity and the like. As suggested by an external antenna 314 in FIG. 3, the internal controller 312 can communicate wirelessly with the controller 320.

[0033] The bus 310 is assumed to be a public transport vehicle serving a route defined in a public timetable, so that the sensors 220a, 220b will move periodically along this route. It is noted that many vehicle types operate according to non-public timetables or have periodic movement patterns for other reasons, which can therefore replace the bus 310 in this example. This includes refuse collection trucks, postal vehicles, street sweepers, mining and construction vehicles.

[0034] From the point of view of the sampling control method 100, the step 110 of providing the ambient sensors 220, 222, 224, 226 includes obtaining up-to-date information indication the numbers and identities of the active sensors and their measuring capabilities.

[0035] In a second step 112, the controller 320 receives the desired measuring locations 240, 250. One may therefore consider that a purpose of the method 100 is to enable the efficient the collection of air quality measurements here. As shown in FIG. 2, the two desired measuring locations 240, 250 in this example have a two-dimensional geometry.

[0036] In a next step 114, the controller 320 receives data from each sensor 220, 222, 224, 226 representing a route and any refractory period of the sensor. Based on this data, the controller 320 is able to ascertain which ones of the routes 230, 232, 234 can be used to cover each of the desired measuring locations 240, 250. In the situation illustrated in FIG. 2, sensors traveling on all three routes 230, 232, 234 can measure in the first desired measuring location 240, while only the second route 232 reaches the second desired measuring location 250.

[0037] In steps 116 and 118, to cover the first measuring location 240, the controller 320 assigns sampling location A to the second sensor 222; sampling location B to the first and third sensors 220, 224; and sampling location C to the second and third sensors 222, 224. (It is noted that sampling location B is assigned to either sensor 220a or sensor 220b. Certainly these are carried by the same vehicle 310 but are treated as separate sensors for the purposes of this method loo; this is all the more justified as the sensors 220a, 200b are configured to measure different quantities.) To cover the second measuring location 250, a sampling location D is assigned to the fourth sensor 226. One or more of the sampling locations may optionally be assigned in conjunction with a sampling periodicity.

[0038] The resulting sampling times follow from the movements of the sensors. In the present example, FIG. 2 shows the positions and directions of the sensors at an instant between time t=n−1 (arbitrary units) and t=n. The approximate sampling times for the respective sensors are given in Table 1.

TABLE-US-00001 TABLE 1 n − 1 n n + 1 n + 2 n + 3 220 B 222 A C B 224 B A C 226 D

[0039] In other embodiments, a corresponding overview of the sampling times may be obtained from timetables or other service plans, possibly with the aid of simulations modeling the effects of non-deterministic traffic flows, weather conditions, boarding and alighting passengers, and other factors causing timetable deviations.

[0040] It is derivable from the present Table 1, first, that the time-wise coverage of the first measuring location 240 is adequate, for at least one sample will be taken at each of the shown time instants. The coverage of the second measuring location 250 may be considered optimal in the circumstances, knowing that only the fourth sensor 226 is available but that it is utilized (once, or according to an assigned sampling periodicity) whenever it passes.

[0041] Second, Table 1 allows to establish whether each sensor's refractory period has been observed. The separation of two sampling times by the first and fourth sensors 220, 226 is at least 5 time units. The corresponding number for the second and third sensors 222, 224 is 1 time unit. If this is incompatible with the refractory periods reported in step 114, then the sampling locations A, B, C (and possibly D) are discarded and steps 116-118 repeated until a satisfactory assignment is found.

[0042] Steps 116-118 may be repeated by initiating another execution of a random procedure, an optimization solver run with slightly different initial values, or with another change allowing a different number of sampling locations or differently situated sampling locations to be objectively expected. Another approach to the removal of too dense sampling is to apply modifications at the problematic time instants directly. For the second sensor 222, such a time instant may be t=n, where sampling at location C will occur only one time unit after the sampling at location B; giving up sampling location C may be a viable option, as the first measuring location 240 is anyway covered by the third sensor 224 at sampling location B at t=n. Similarly, it may be justified to un-assign sampling location A from the third sensor 224 at t=n+2, which may jeopardize this sensor's ability to sample at location C one time unit later.

[0043] As explained in an earlier section of this disclosure, clustering of measurements is to be avoided. Table 2 allows to examine to what extent this goal is achieved by an assignment of sampling locations.

TABLE-US-00002 TABLE 2 A B C D 220 n + 1 222 n − 1 n + 2 n 224 n + 2 n n + 3 226 n + 2

[0044] It is clear from Table 1 that there are no simultaneous measurements at any of the sampling locations. The sensitivity to clustering is related to the time unit used. Executing the method 100 with a longer time unit will correspond to a lower time resolution (higher granularity), which tends to identify a greater number of simultaneous measurements. The method 100 will then reject and recompute candidate sampling location assignments relatively more often, so that the clustering tendency is suppressed more vigorously.

[0045] The steps 116-118 may optionally be followed by a programming step and/or a data collection step. In the programming step, the sensors 220, 222, 224, 226 are controlled in accordance with the sampling locations A, B, C, D. In the data collection step, the controller 320 or another data recipient receives the measurement results from the sensors.

[0046] In a variation of the described embodiment, the method 100 may be executed by the internal controller 312. The beneficiary of the collected data may then be the vehicle 310 or a person using or traveling with the vehicle 310. The internal controller 312 may be elected as lead controller by the execution of a leader election algorithm by the simultaneously present sensors associated with analogous controller equipment; the leader election algorithm may be random or pseudo-random. When the method 100 is executed from an internal controller 312 of a vehicle, communication with other sensors (possibly, through the intermediary of the vehicles on which they are mounted) may proceed over a cellular network or short-range wireless. Multi-hop or ‘meshed’ communication, by which a nearby vehicle relays a message towards a more distant vehicle, is a way to alleviate the coverage limitations of short-range wireless, so as to increase the workable size of a measuring zone.

[0047] While clearly not essential to the method 100, some heuristics and automated approaches for assigning the sampling locations (and any sampling periodicities) on the basis of desired measuring locations and refractory periods will now be discussed. One possible approach is to perform a random or pseudo-random assignment of sampling locations, which is evaluated against predefined criteria and repeated if needed. The criteria may include: [0048] completeness of time coverage, [0049] avoidance of clustering, [0050] observance of refractory periods, [0051] completeness of spatial coverage of the desired measuring locations, [0052] resilience (e.g., the number of independent sensors that are engaged to cover each measuring location).

[0053] As explained above, the three first criteria may be evaluated by forming Tables 1 and 2 for the case under consideration.

[0054] Another possible approach is to formulate an optimization problem which is solved numerically, with the sampling locations as output. The above criteria may either be included in an objective function or in boundary conditions. For example, the objective function may be dependent on the sampling locations and may include terms penalizing clustering and rewarding complete coverage. The refractory periods, which are arguably of a more coercive character, may be included as boundary conditions. The objective function of such an optimization problem may further include a term rewarding the total number of samples collected, so as to reflect a desire to optimize the use of a given number of sensors. The choice of the numerical solver is not essential to the present invention; it will be within the skilled person's abilities to select a suitable one of the solvers described in the literature.

[0055] Machine learning methods constitute a still further option for automating the assignment steps 116-118.

[0056] The aspects of the present disclosure have mainly been described above with reference to a few embodiments. However, as is readily appreciated by a person skilled in the art, other embodiments than the ones disclosed above are equally possible within the scope of the invention, as defined by the appended patent claims.