WO2015015217A1

WO2015015217A1 - Location-based navigation

Info

Publication number: WO2015015217A1
Application number: PCT/GB2014/052358
Authority: WO
Inventors: Benjamin HIGGINBOTTOM; Julian COOMBES; Alistair JACKSON; David Robertson; Kenneth MARRITT
Original assignee: Barclays Bank Plc
Priority date: 2013-08-02
Filing date: 2014-07-31
Publication date: 2015-02-05
Also published as: GB201313880D0; GB2516875A

Abstract

A system for selecting one of a plurality of service points at bank or other premises located at a respective plurality of different service point locations for a mobile user, comprising at least one security camera (12) at each of the premises sensing the queue at the respective service point (10) and generating a respective output signal; a queue predictor device (16) arranged to predict the queuing length or time at the service point from the respective camera output; a GPS location device arranged to detect the current location of the mobile user; a navigation system arranged to predict the travel time of the user to each of the service point locations; and a selection device for selecting one of the service points based on both the travelling time to it and the queue time at it.

Description

LOCATION-BASED NAVIGATION

[0001] This invention relates to location-based navigation to one of a plurality of queue points. Standing in queues, also known as lines, to obtain goods or services is economically inefficient both for the queuing person (who is unable to use their time usefully) and for the goods or services supplier who is not maximising throughput. It is usually also not enjoyable for the queuing person, and if it colours their opinion of the supplier, they may not return.

[0002] Many premises, for example supermarkets, have multiple co-located service points and multiple lines are provided, one for each service point. People arriving at the premises will then choose the line that appears the shortest. Sometimes, for example at Passport Offices, customers are given tickets allocating them to one queue or another. Sometimes, for example at bank branch premises, a single line is provided and people are directed sequentially to the next available service point.

[0003] In supermarket premises, the person in charge of the checkout counters may manage the queues, for example by opening more checkouts when there are long queues or closing unused checkouts. Infrared Integrated Systems Ltd, and others, have supplied systems which employ cameras and image processing software to automatically estimate the length of the queue, for example by identifying and counting the number of individuals in each queue. Their EP2521101-A discloses a system of this kind, in which the queuing time (rather than just length) can be estimated by using additional data from the service points to determine the time taken by each transaction.

[0004] However, such systems only operate within a single premises. A mobile customer may have a choice of several premises (for example, supermarkets, coffee shops or banks) where equivalent goods or services can be obtained. Whereas at a single premises it is possible for a customer to visually select the shortest queue, this is impossible as between geographically dispersed premises.

[0005] Recently, an iPhone™ application called WAITBOT has been proposed which will attempt to give users a list of the wait times at premises (for example restaurants) near them, so as to allow users to select the premises with the shortest queue. They apparently propose to ask the businesses by telephone, and also to allow people in the area to supply the wait time data (i.e. "crowdsourcing"). However, such data will inevitably be subjective and there will be no checks on its validity. It is unclear whether the proposed methods will ever achieve their objective, as the principals explain in the video at http://vimeo.com/62119153. [0006] The present invention aims to provide a system, and in another aspect a corresponding computer program, to allow selection of one of a plurality of geographically dispersed queues to which a user can travel. The present invention in one aspect does so based on data representing, for each one, the precicted queue length (or time) taking into account the travel time required to reach the queue.

[0007] By taking into account both the queuing time and the travel time, the user can select (or be allocated) the line which he or she will finish first, not merely the shortest line, or the closest one. It is possible that a premises which currently has a long queue may be predicted on the basis of historical data to have a shorter queue by the arrival time. Merely displaying the current queue data, as proposed in the WAITBOT application, would lead the user to select inappropriately in this case even if the data supplied by that application were accurate (which as discussed above is doubtful).

[0008] The predicted arrival times are calculated using knowledge of the current user position, (preferably from a GPS or other location service), and of the positions of the queuing points, and of the expected speed of the user. Conveniently, the invention may implement a navigation or routing program (such as those supplied by Garmin™ or Tom Tom™, or Google Maps™) which employs knowledge of the roads and speed limits (where the user is in a vehicle) or an indication that the user is on foot, to calculate a route to each queuing point and calculate the arrival times at each point therefrom.

[0009] Preferably, the system is implemented in two parts: a first part comprising cameras and computer programs at (or associated with) the premises, and a second part comprising a mobile position location and communications device associated with a user (for example a smartphone, tablet computer or PDA). In another embodiment, the second part may comprise a standard computer , for example running a conventional browser program, arranged to allow the user to input a position.

[0010] At this point it is noted that there is a longstanding wealth of literature on the "travelling salesman" problem, to find the best route for a travelling agent (rather than a customer) who has to travel to each of a number of geographically dispersed customers to deliver a service. US2010/0169147-A describes a system of that sort for doctors visiting a list of patients in different geographical points within different zones. However, it will be apparent that the problem addressed herein is quite different to the travelling salesman problem, in that the user (rather than the supplier) is mobile and is choosing a single queue from a plurality of queues, rather than compiling an ordered list of consumers all of whom must be visited. [0011] Where the service points are at retail goods or service premises such as banks or shops, the invention preferably comprises one or more cameras pointing towards the service point to estimate the queue length and/or queuing time. Conveniently, as the queue changes relatively slowly, the cameras may also be used at the same time as security cameras providing a view of the service point to security staff.

[0012] Preferably, the queue length is estimated using a Markov Model stochastic technique, in which data is stored indicating which of a plurality of queue length states the queue is currently in, and the probabilities that it will transition to that same state, and to others of said states. The states may correspond to queues that are empty, short, medium or long for example. Use of a small number of states has been found to allow more effective prediction than use of the actual number of people in the queue (which would imply a very large number of states and hence render prediction difficult).

[0013] Other embodiments, aspects and preferred features of the invention will be apparent from the following description and claims, with advantages that will be apparent hereafter.

[0014] Embodiments of the invention will now be illustrated, by way of example only, with reference to the accompanying description and claims in which:

Figure 1 is a block diagram showing the elements of the system of a preferred embodiment of the invention;

Figure 2 is a schematic diagram illustrating a queue at premises forming part of Figure 1: Figure 3 is a diagram illustrating a Bayes Network representing queue states in the embodiment of Figure 1 ;

Figure 4a is a state transition diagram illustrating the transitions and transition probabilities for a first, longer, time interval;

Figure 4b is a corresponding state transition diagram for a second, shorter, time interval; Figure 5 is a flow diagram illustrating the operation of the embodiment of Figure 1;

Figures 6a-6c are screen displays for a mobile telephone in the embodiment of Figure 1 ; Figure 7 is a screen display for a computer terminal in the embodiment of Figure 1 ; and Figure 8 illustrates a conventional computer system which implements the computers of Figure 1. Preferred Embodiment of the Invention

[0015] Referring to Figure 1, a system according to the preferred embodiment comprises a mobile terminal such as a GPS -enabled smartphone 2, in radio communication with the Internet 4 via a radio network 6, in possession of a user. At a number of premises 8a, 8b, 8c ... in a geographical area such as a town, there are respective service points 10a, 10b, 10c .... (shown in Figure 2). The premises in this embodiment may be branches of a bank, and the service points may be teller counters. At each branch premises 8a, 8b, 8c, ... is a respective branch computer 14a, 14b, 14c ... in communication via a secure link carried over the Internet 4 with a central bank computer 16.

Queue Length Detection

[0016] Referring to Figure 2, at each premises 8, in front of the service point 10 (or multiple several service points) is a space for queuing. A security camera 12a, 12b, 12c ... is provided at each location. Preferably it is a commercially available 360 degree field of view camera. A single camera 12 can be used to detect, report and predict on more than one queue by partitioning the viewable image.

[0017] The digital video output of the camera 12 is connected to the branch computer 14 on the premises, which runs a queue length detection program. The branch computer performs identification of each of the separate people in the queue, using standard CCTV computer vision technologies which detect pixel changes to minimise the effects of occlusion on the faces or bodies of the individuals (see for example "Hallucinating Multiple Occluded CCTV Face Images of Different Resolutions", Kui Jia & Shaogang Gong, in Proc. IEEE International Conference on Advanced Video and Signal based Surveillance (AVSS'05), September 2005). Commercially available programs include the Queue Length detection tool from AxxonSoft Software, Berezovaya alley, estate 5a, building 5, 127273 Moscow, Russia or the IRISYS SMARTLANE system available from Infrared Integrated Systems Ltd, Park Circle, Tithe Barn Way, Swan Valley, Northampton, UK.

[0018] The detection is performed repetitively, for example every 15 seconds (a period slightly shorter than a typical transaction time). As a result, the branch computer has, at 15 second intervals, a list of individual people in the queue, from which can be calculated the associated number of such individuals, defining the instantaneous queue length in terms of numbers of people. Where there is more than one service point 10 at the premises 8, the branch computer establishes these data for each service point. The branch computer stores the numbers of people in long term storage to allow statistical processing later. Each branch computer 14 also sends the state to the central computer 16.

[0019] The branch computer 14 also stores data specifying the transaction time or speed. This may be a global average figure derived by human measurement, or in some cases could be derived independently from the outputs of apparatus at the service point (for example a vending or ATM machine) to establish the average time between each transaction. The product of the line length and the average transaction time per individual defines the queuing time at the service point. The transaction time is also sent to the central computer 14. Queue Length Prediction

[0020] The central computer 14 then quantifies each queue length into one of a predetermined number of states. In this embodiment, there are four states, HIGH (i.e. long queue, more than 10 people), MEDIUM (i.e. 5-10 people), LOW (i.e. 3-5 people) and EMPTY (i.e. 2 people or less).

[0021] The queue within the sampling time is assumed to be able to transition from its current state to any other of these states (including its current state). It is possible to treat the state chains as Markov transitions. Thus, the central computer 16 stores the queue state transition probability matrix, the transaction time or speed, and a set of past states for the queue(s) at each branch. Periodically (for example every week), and after any change in the queuing geometry of the premises, the branch computer re-estimates the queue state transition probability matrix.

Estimating the transition probabilities

[0022] To initially estimate the transition probabilities, the queue lengths stored at each sample time (i.e. every fifteen seconds) are used. From the set of past values, the probability matrix containing the probabilities of the state changing to each other state (as illustrated in Figure 3) can be calculated, by calculating the number of transitions from each state i to all states (for i=l to N where N is the total number of states, in this case 4, including state i itself), divided by the total number of transitions from state i. Thus, for example, if the EMPTY state occurs 20 times in the stored past data and the immediately following state is FULL on 3 occasions, the EMPTYIFULL transition probability is 3/20 = 0.15 or 15%.

[0023] Figure 4a illustrates the transition probabilities calculated from sample data taken over the period from 08:30 to 16:30 at a branch on 18^th April 2013. [0024] The initial probabilities of each state are easily obtained from the raw data (by summing the number of occurrences of each state in the sample and dividing it by the total number of sampled states) and shown in Table 1.

Table 1

[0025] And for ease, the transitional probabilities of Figure 4a are shown below in Table 2.

Table 2

[0026] What this shows is that for 86% of the operating day the queues are effectively EMPTY (see the initial state probabilities) with a 93% probability of it remaining in an empty state (see the transition probabilities). Likewise should the queue be in a HIGH state, then it will with a 75% likelihood remain in a HIGH state, with a 25% probability of it moving to a MEDIUM state. The queue however is only in a HIGH state for 0.001% of the time the queue is operational.

[0027] However, queues tend to behave differently at different times of day. For example, there are busy periods (lunchtime) and periods which are busy for particular activities (businesses depositing coins in the morning).

[0028] The data for the whole day is thus only useful for predicting medium to long term trends, effective for management information but less so from a customer perspective where the more short term and immediate trends are useful. Taking data from a 20 minute period between 11:50 and 12: 10, the shorter term state transition model is shown in Figure 4b.

[0029] The initial probabilities are again easily obtained from the raw data and shown in Table 3. State Prob'y

P(e] 0.94

P(l) 0.05

P(m) 0.008

P(h) 0.00

Table 3

[0030] And for ease, the transitional probabilities of Figure 4b are shown below in Table 4.

Table 4

[0031] So from this data it can immediately be seen that over this shorter period, the queue is never in a HIGH state, and even if it were in a MEDIUM state it would immediately transition to a LOW state on the next (t₊ 15 second sample time.

[0032] The transition probability matrices are therefore periodically re-estimated based on a time window. Queue length data from the same time of day on the same day of the week for previous recent weeks may also be used to improve the accuracy of the probabilities.

Predicting the future queue length

[0033] The central computer 16 is able to determine the probability of the next queue state in the next sample period, and so on forwards in time. The starting point is the current state (sample to) of the queue as observed by the cameras 12. Working forward to the next time sample t₊₁, for each state, the initial probabilities calculated above are used.

[0034] Working forward to the next time sample t₊₂, for each state, the transition probability into that state from each other is calculated, and multiplied by the initial probability of being in that state.

[0035] Thus, to calculate the probability of a MEDIUM queue length at the next time sample t₊i, the initial probability for the EMPTY state is multiplied by the elm transition probability; the initial probability for the LOW state is multiplied by the llm transition probability; the initial probability for the MEDIUM state is multiplied by the mlm transition probability; the initial probability for the HIGH state is multiplied by the him transition probability, and then the four products are summed to give the total probability that the queue will be in the MEDIUM state.

[0036] To calculate the probabilities of each state at t₊₅, the central computer 16 could work forward from the currently observed state and calculate in turn the state probabilities at t₊i, t₊₂, t+3, t+4, t+5. More conveniently, using total probability one can use standard matrix operations to determine the queue state:

IpCS) [0:9411 0.0504 0,0084 0]

Equation 1

[0037] There is therefore 94% confidence that the queue will be EMPTY 5 samples (i.e. 1 minute 15 seconds) forward in time. Thus, where a user is a 5 minute walk away from a branch, the user will arrive at sample t₊₂o. The same calculation as in Equation 1 is performed but raised to a power of 20 rather than 5.

Operation of the embodiment

[0038] Referring to Figure 5, a user seeking to visit a bank branch opens a banking application 22 running on his mobile terminal 2. The mobile terminal 2 has a GPS receiver (24 not shown) which calculates the terminal position and supplies it as input to the banking application (step 102), for example using the GPS Maps™ Applications Programming Interface (API). If GPS information for the user's location is not available, cell base station ranging or current cell can be used, and/or the user can be prompted to enter their postcode if known, or expected arrival time at a particular location.

[0039] The banking application then displays a message asking the user to select a travel mode between walking and driving (step 104). If walking is selected, the banking application accesses stored walking speed data (calculated by timing previous walking journeys whilst tracking the terminal position). It then sends a message containing the terminal position and the travel mode (and walking speed, if available) to the bank central computer 16 (step 106).

[0040] On receipt, the central computer 16 accesses a location database storing the locations of all bank branches and selects those within a 30 minute radius of the user terminal, using the walking speed (if walking mode is selected) or a stored average traffic speed in the area and speed limit data (if driving mode is selected) to calculate the radius (Step 108).

[0041] The central computer 16 then calculates the predicted queue length (EMPTY, LOW, MEDIUM or HIGH) at the supplied arrival time, and transmits back to the central computer 16 a message containing the current queue length, the predicted future queue length and the corresponding predicted queuing time (or range of times) in step 114. The central computer 16 aggregates the messages and sends them back to the application 22 on the user terminal together with the locations of the selected premises 8a, 8b, 8c (step 116).

[0042] The application 22 then displays (step 118), superimposed on a local map, the travel times to, and current and predicted queue lengths at, each of the local bank branches 8a, 8b, 8c. The application also calculates (step 120) the sums of the travel and queuing times for each, selects the lowest and displays the route to it. The user will often choose to follow the suggestion, but sometimes (for example in rain) may select a shorter travel time even at the expense of a longer queuing time.

[0043] As the user travels, the application communicates the terminal position periodically with the central computer 16, which obtains and communicates updated queue data from and travel time to each of the selected branches. The application then periodically refreshes the displayed data, to allow the user to change destination in the event that the queue length changes, or unpredicted traffic or weather conditions render a longer journey undesirable.

User Interfaces

[0044] Figure 6a shows the display produced on a mobile terminal 2 comprising a smartphone. A local map is displayed using Google Maps™. The user position is indicated as a circle in the lower right-hand corner. Three nearby branches are displayed, by eagle symbols denoting the bank concerned, as is an ATM machine. At the top right hand corner of each branch symbol is a ring containing a number indicating the queue length.

[0045] Figure 6b indicates the effect of selecting, via the terminal 2 user interface, one of the symbols. A text box stating the current queue length pops up, together with the address and distance.

[0046] Figure 6c is a text display selectable, via the terminal 2 user interface, as an alternative to that of Figures 6a and 6b. A text list of the nearby branches is displayed, showing in each case the distance, status (open or closed), list of services available (e.g. advice, mortgages, currency etc., indicated by symbols), and queue length for each separate service point (in a circle at the top right hand corner of the symbol or icon for the service concerned).

[0047] In each case, user can select whether the queue length displayed is the instantaneous length or the predicted length.

[0048] Figure 7 shows the display produced on a user terminal 2 (which may be a mobile terminal or a computer, in either case accessed via the internet). At the top of the screen, the user enters a postcode indicating their current position, and is able to select a particular type of service required (e.g. cash machine, business banking counter, mortgage advisor, open on Saturday etc.). In response, as in Figure 6a, a local map is displayed containing the indicated user position and nearby branches, indicated by symbols. To the right of the screen, a list of the branches is displayed (corresponding to that of Figure 6c) ranked in order of distances from the user position.

[0049] As shown, when one of the branch symbols is selected a pop-up text box displays the address and distance, together with the current queue lengths for each available service, and text saying "From your current position, if you left now, the queue length would be:" followed by the predicted queue lengths for each available service at the premises at the calculated arrival time.

Computer Systems

[0050] The branch computer 14 and central computer 16 systems described herein may be implemented by conventional computer systems such as computer system 1000 as shown in Figure 8. Embodiments of the present invention may be implemented as programmable code for execution by such computer systems 1000. After reading this description, it will become apparent to a person skilled in the art how to implement the invention using other computer systems and/or computer architectures.

[0051] Computer system 1000 includes one or more processors, such as processor 1004. Processor 1004 may be any type of processor, including but not limited to a special purpose or a general-purpose digital signal processor. Processor 1004 is connected to a communication infrastructure 1006 (for example, a bus or network). Various software implementations are described in terms of this exemplary computer system. After reading this description, it will become apparent to a person skilled in the art how to implement the invention using other computer systems and/or computer architectures.

[0052] Computer system 1000 also includes a main memory 1008, preferably random access memory (RAM), and may also include a secondary memory 610. Secondary memory 1010 may include, for example, a hard disk drive 1012 and/or a removable storage drive 1014, representing a floppy disk drive, a magnetic tape drive, an optical disk drive, etc. Removable storage drive 1014 reads from and/or writes to a removable storage unit 1018 in a well-known manner. Removable storage unit 1018 represents a floppy disk, magnetic tape, optical disk, etc., which is read by and written to by removable storage drive 1014. As will be appreciated, removable storage unit 618 includes a computer usable storage medium having stored therein computer software and/or data.

[0053] In alternative implementations, secondary memory 1010 may include other similar means for allowing computer programs or other instructions to be loaded into computer system 1000. Such means may include, for example, a removable storage unit 1022 and an interface 1020. Examples of such means may include a program cartridge and cartridge interface (such as that previously found in video game devices), a removable memory chip (such as an EPROM, or PROM, or flash memory) and associated socket, and other removable storage units 1022 and interfaces 1020 which allow software and data to be transferred from removable storage unit 1022 to computer system 1000. Alternatively, the program may be executed and/or the data accessed from the removable storage unit 1022, using the processor 1004 of the computer system 1000.

[0054] Computer system 1000 may also include a communication interface 1024. Communication interface 1024 allows software and data to be transferred between computer system 1000 and external devices. Examples of communication interface 1024 may include a modem, a network interface (such as an Ethernet card), a communication port, a Personal Computer Memory Card International Association (PCMCIA) slot and card, etc. Software and data transferred via communication interface 1024 are in the form of signals 1028, which may be electronic, electromagnetic, optical, or other signals capable of being received by communication interface 1024. These signals 1028 are provided to communication interface 1024 via a communication path 1026. Communication path 1026 carries signals 1028 and may be implemented using wire or cable, fibre optics, a phone line, a wireless link, a cellular phone link, a radio frequency link, or any other suitable communication channel. For instance, communication path 1026 may be implemented using a combination of channels.

[0055] Likewise, the mobile terminal 2 comprises a computer system 1000 which includes a touchscreen display as the user interface, a radio communications interface and a TCP/ΓΡ communications protocol program. [0056] The terms "computer program medium" and "computer usable medium" are used generally to refer to media such as removable storage drive 1014, a hard disk installed in hard disk drive 1012, and signals 1028. These computer program products are means for providing software to computer system 1000. However, these terms may also include signals (such as electrical, optical or electromagnetic signals) that embody the computer program disclosed herein.

[0057] Computer programs (also called computer control logic) are stored in main memory 1008 and/or secondary memory 1010. Computer programs may also be received via communication interface 1024. Such computer programs, when executed, enable computer system 1000 to implement embodiments of the present invention as discussed herein. Accordingly, such computer programs represent controllers of computer system 1000. Where the embodiment is implemented using software, the software may be stored in a computer program product and loaded into computer system 1000 using removable storage drive 1014, hard disk drive 1012, or communication interface 1024, to provide some examples.

Alternative Embodiments

[0058] The above embodiments are described by way of example, and variants, additions and alternative embodiments which may become apparent to the skilled person on reading the above description may nevertheless fall within the scope of the claims. The word "comprising" as used herein means "including" and does not imply the absence of additional features.

[0059] In the user nterface, queue length indication could be done as a Red, Amber, Green, (RAG) symbol (e.g. a coloured circle) based on the "EMPTY, LOW, MEDIUM or HIGH" categories, possibly supplemented by white to indicate EMPTY, so rather than indicate exact queue length, a traffic light colour based on the category is shown.

[0060] Where the terminal is a smartphone or computer, in another embodiment the application itself performs the routing calculations, using for example conventional Google Maps™, Garmin™ or TomTom™ GPS technology. In this case, the application locates local branches, and sends the arrival times at each, to the central computer which transmits in reply the predicted and current queue lengths.

[0061] It would be possible for the central computer to select a preferred branch and communicate only this to the user. This may be useful in the event that the user has limited display or processing capabilities which make map display and updating undesirable. In this way it would even be possible to employ an application running on a 2G phone, communicating using SMS messages.

[0062] The queue length prediction mechanism is data-source independent, so it could be implemented using any other queue counting mechanism, and for example infra-red or motion capture techniques could be used, however the computer vision approach described above provides an accurate measurement of the state of the queue. The techniques described in the above-mentioned EP2521101-A (incorporated herein by reference) could also be used.

[0063] In principle, the cameras 12 could all stream their video footage and the queue detection and prediction could be done at the central bank computer. However, it makes a more efficient use of data communications bandwidth to distribute the processing to the local bank computers 14 and transmit only the queue length data to the central bank computer. The queue length prediction could be performed either at the central computer, storing a plurality of models one for each premises, rather than at the branch computers as described above.

[0064] It would be possible to implement a smoothing mechanism by adding 'false' readings for each of the missing transitions called K, which can be accurately determined by experimentation. Other stochastic or Bayesian statistical modelling techniques could be used for prediction.

[0065] It would be possible for each branch computer 14 to continually calculate its future queue states, rather than the central computer 16.

[0066] Whilst banks have been discussed, it is clear that the invention could without modification be employed at any premises offering equivalent goods or services, for example Post Offices, coffee shops (whether or not in the same chain), supermarkets, ATM machines, vending machines and so on.

[0067] Whilst recognition of queuing humans is described, the same principles could be used to detect and predict the lengths of queues of vehicles, thus enabling a driver to select between alternative crowded bridges for example.

[0068] It will be apparent that the features of different embodiments can be separated and/or combined. Protection is sought for any and all novel subject-matter disclosed herein separately and all combinations and sub-combinations thereof.

Claims

1. A system for assisting selection, for a mobile user, of one of a plurality of service points at premises located at a respective plurality of different service point locations, comprising:

at least one detector device at each of said premises sensing the queue length at the respective service point and generating a respective output signal;

a location device arranged to obtain the current location of the mobile user;

a navigation system arranged to predict the respective arrival times of the user travelling from said current user location to each of the plurality of service point locations; and a queue predictor device arranged to predict the queue length or queuing time at the service points from the respective detector output at the respective predicted arrival time.

2. A system according to claim 1, in which the queue predictor is arranged to calculate the predicted queuing time at each said service point at the elapse of the predicted travel time thereto.

3. A system according to claim 2, in which the queue predictor comprises a store storing data defining a stochastic model, comprising a number of states each corresponding to a queue length or queuing time, and is arranged to predict which state the queue will probably occupy at said elapse.

4. A system according to any preceding claim, in which the detector device at one or more premises comprises a video camera.

5. A system according to claim 4, in which the video camera also provides a security camera feed to a monitor.

6. A system according to any preceding claim, in which the location device is a mobile terminal such as a telephone or computer, with an operating system capable of downloading and running applications.

7. A system according to any preceding claim, in which the premises comprise bank premises.

8. A computer adapted for use as a queue predictor device in the system of any preceding claim.

9. A mobile terminal adapted for use as a queue predictor device in the system of claim 6.

10. A computer-implemented method of assisting selection, for a mobile user, of one of a plurality of service points at premises located at a respective plurality of different service point locations, comprising:

sensing the queue length at the respective service point and generating a respective output signal;

obtaining the current location of the mobile user;

predicting the respective arrival times of the user travelling from said current user location to each of the plurality of service point locations; and

predicting the queue length or queuing time at the service points from the respective output at the respective predicted arrival time.

11. A storage medium comprising machine readable instructions stored thereon for causing a computer system to become configured as the system of any of claims 1 to 7, the computer of claim 8 or the terminal of claim 9.

12. A storage medium comprising machine readable instructions stored thereon for causing a computer system comprising one or more computers to perform a method in accordance with claim 10 or a part thereof.