Railway scheduling

Abstract

Trains 105, 106 travelling along a railway track 102, 103, 124 are scheduled by obtaining stop data representing parameters relevant to a time period for the train to stop at a platform 107, 108 and u

Description

Field of the Disclosure Hie present disclosure relates to railway scheduling. Background of the Disclosure Railway scheduling systems are used to schedule the movement of trains along a railway track. An objective of such scheduling systems is to optimise the utilisation of the railway track by trains, to improve the efficiency of trains’ schedules and timeliness of the trains, whilst ensuring the safety of trains travelling along the railway track, by ensuring that conflicting train movements, e.g., two trains travelling on a same section of track at a same time, is avoided. A particular objective of scheduling systems is to appropriately allocate track time between competing trains to achieve a balance that best serves the overall objective of scheduling efficiency and timeliness of the train portfolio. Summary of the Disclosure A first aspect of the present disclosure provides a method for scheduling trains travelling along a railway track, the method comprising: receiving stop data representing parameters relevant to a time period for a train to stop at a platform of the railway track, computing a time period for the train to stop at the platform based on the stop data, generating schedule data for scheduling the train based on the computed time period, and outputting a schedule signal representing the generated schedule data. The scheduling system may thus make dynamic decisions for the appropriate periods of time for trains to stop at the platforms, referred to in the technical field as the ‘dwell time’ based on various sources of data received dynamically by the scheduling system representing various parameters that have been found relevant to the train dwell time. Thus, the scheduling system can predict in advance appropriate times for trains to stop at platform, and generate schedules for those trains, e.g., dwell times, platform departure times, etc, based on those predicted stop times. The generated schedules may then be used to facilitate control of the trains. In the example of a passenger train, the train is required to stop at platforms for sufficient time to allow passengers to embark and disembark safely. Conversely, it may typically be desirable to minimise the platform dwell time in order to avoid excessively delaying the train, and following trains. It may be desirable therefore that trains stop at platforms for approximately only the time required to allow passengers to embark and disembark. The present scheduling system may thus 2 usefully predict a time required for passengers to embark and disembark trains at the platforms dynamically based on real time data relevant to the passenger embarkation / disembarkation time, for example, the number / distribution of passengers and / or weather. Consequently, the trains maybe scheduled to stop at platforms for only approximately the time necessaiy for passenger embarkation / disembarkation, which may have time-efficiency advantages both for the train and also when scheduling a portfolio of trains running along a same track. In comparison to the present disclosure in which dwell times are determined using the relevant stop data, a conventional approach to scheduling trains is to assume a dwell time at a platform that is agnostic of prevailing conditions. Thus, in the conventional approach, a train may be scheduled and controlled to dwell at a platform for longer than is required, thereby potentially unnecessarily delaying the train and following trains. In an implementation, the receiving stop data comprises receiving image data representing an image of one or both of a passenger compartment of the train and the platform, determining based on the image data a presence of passengers, and the computing a time period involves computing the time period based on the determination of a presence of passengers. Using the image data, the scheduling system may determine whether passengers are present on the platform or in the passenger compartment, and may compute the dwell time based on that knowledge, for example, using attained machine-learning model with the passenger presence as the input. Thus, in the event that no passengers are waiting to embark or disembark, the dwell time may be determined to be short, or the train may even be scheduled not to stop at the platform. Consequently, the scheduling efficiency may be improved. In an implementation, the method comprises determining based on the image data a number of passengers, and the computing a time period involves computing the time period based on the determined number of passengers. It has been recognised that a significant parameter relevant to a time required for passengers to embark and disembark a train is the number of passengers to embark / disembark. A large number of passengers may ty pically take longer to embark / disembark than a smaller number of passengers, for example, due to queuing to pass through train doors. By this feature, the scheduling system may approximate the number of passengers to embark / disembark. Based on that knowledge of passenger numbers, the scheduling system may determine an appropriate dwell time for the train at a platform In an implementation, the method comprises determining based on the image data locations of passengers relative to the field of view, and the computing a time period involves computing the time period based on the determined locations of passengers. It has been further recognised that the distribution of passengers within a train’s passenger compartment and / or on the platform can affect the embarkation / disembarkation time. For example, it has been recognised that where passengers on a platform are bunched together, as might occur near to a platform entrance, it may take relatively longer for those passengers to embark the train than if the same number of passengers were evenly distributed along a platform, because the passengers may tend to queue to use a same train door rather than using all doors of the train. And passengers in a train’s passenger compartment loitering around the train door may obstruct the door and similarly slow the embarkation / disembarkation of passengers. By detecting locations of passengers in the image, the scheduling system may thus more accurately infer the appropriate dwell time. In an implementation, the method comprises receiving further image data representing a further image of one or both of the passenger compartment of the train and the platform, determining based on the further image data a number of passengers in the further image and a sum of the number of passengers in the image and the further image, and the computing a time period involves computing the time period based on the sum of the number of passengers. In an implementation, the method comprises receiving mass data from the train representing a mass of the train, and the computing a time period involves computing the time period based on the mass of the train. The scheduling system may thus usefully infer a number of passengers in the train’s passenger compartment by knowledge of the mass data and an unladen mass of the train. In an implementation, the method comprises receiving weather data representing a weather condition relevant to the platform, and generating based on the weather data the stop data representing the weather condition. It has further been recognised that the prevailing weather conditions may impact the time taken by passengers to embark / disembark, e.g., precipitation may slow passenger embarkation / disembarkation. Thus, by this feature, the scheduling system may more accurately infer the appropriate dwell time. In an implementation, the method comprises receiving position data representing a position of the train and detennining if the position of the train meets position criteria. In an implementation, the generating the schedule data is performed based on a determination that the position of the train does meet the position criteria. By this feature, the scheduling system may determine whether the train is sufficiently proximate to the platform prior to computing the appropriate dwell time and generating the schedule. This may be desirable, for example, to avoid determining the dwell time excessively early, as performing the detennination too early may reduce the reliability of the determination, for example, due to additional passengers arriving at the platform after the detennination has been made. Thus, by this feature, the dwell time may be determined when the train is relatively close to the platfonn. In an implementation, the method comprises determining if the position of the train meets position criteria in response to an initial determination that the position of the train does not meet the position criteria. In an implementation, the method comprises receiving the schedule signal and generating a control signal for controlling movement of the train based on the schedule signal. A second aspect of the present disclosure provides a railway scheduling system configured to perform the method of any one of the preceding statements. A third aspect of the present disclosure provides a computer program comprising instructions, which, when executed by a railway scheduling system causes the railway scheduling system to carry out the method of any one of the preceding statements. A fourth aspect of the present disclosure provide a computer-readable data carrier having the computer program of the preceding statement stored thereon. These and other aspects of the invention will be apparent from the embodiment(s) described below. Brief Description of the Drawings In order that the present invention may be more readily understood, embodiments of the invention will now be described, by way of example, with reference to the accompanying drawings, in which: Figure 1 shows schematically an example of a railway embodying the present disclosure, comprising a railway scheduling system for scheduling trains running along a railway track; Figure 2 shows schematically functional modules of the railway scheduling system; Figure 3 shows schematically hardware for implementing the railway scheduling system; Figure 4 shows schematically a method perfonned by the railway scheduling system for scheduling trains running on the railway track, including an operation for generating a schedule for a train; and Figure 5 shows schematically processes involved in the operation for generating the schedule. Detailed Description of the Disclosure An example of a railway 101 embodying the present disclosure is depicted schematically in Figure 1. An example method of operating the railway 101 embodying the present disclosure will be described herein with particular reference to Figures 4 and 5. Tire railway 101 comprises first and second railway track sections 102, 103, common track section 104, and first and second trains 105, 106 running on the railway track sections 102, 103. The railway track sections 102, 103 each include a passenger platform 107, 108 to allow passengers to embark and disembark the respective train. The railway 101 further comprises a railway control system, which comprises a signalling system including signal controller 109 and signals 110, 111, a monitoring system including monitor controller 112, platform cameras 113 and train passenger compartment cameras 114, and a scheduling system 115. The components of the railway 101 are not depicted to scale in the Figures. For the purposes of the example operation described herein, the trains 105, 106 are described as travelling from the left to the right of the Figure. Each of the 6 trains 105, 106 are substantially like, and for brevity therefore only train 105 will be described herein in detail, on the understanding that substantially the same teachings apply to train 106 also. The train 105 comprises a passenger compartment 116 for accommodating passengers, an onboard motion control system 117 operable to control the motion of the train, a mass sensing system 118 for sensing a mass of the train, and a radio transceiver 119 operable to communicate with the scheduling system 115. The mass sensing system 118 is operable to determine a mass of the train. For example, the mass sensing system 115 may comprise load sensors for sensing the load applied to each axle. The radio transceiver 119 could, for example, comprise a Global System for Mobile communications protocol device (GSM). The onboard motion control system 117 may further comprise a positioning system for determining a position of the train 105. For example, the positioning system could include an onboard antenna for communicating with trackside beacons located along the railway track, to thereby determine a position of the train by reference to its proximity to a trackside beacon. Hie signalling system is operable to control the signals 110, 111 to provide visual signals, for use by drivers of the trains 105, 106 to control movement of the train. The signal controller 109 is functional to receive signalling instructions from external system, for example, from the scheduling system 115, the monitoring system 112 and / or other systems of the railway control system. The signal controller 109 comprises a radio transceiver 121 which could, for example, comprise a Global System for Mobile communications protocol device (GSM). Additionally or alternatively, the signalling system could provide signal information in another way. For example, the signalling system could send signal information to the trains for display to the driver. The monitoring system is operable to capture video imagery of each of the platforms 107, 108 and the passenger compartment 116 of each of the trains 105, 106. For this purpose, each of the platforms 107, 108 is provided with a pair of cameras 113 arranged to view respective portions of the platform, and each of the trains 105, 106 comprises a camera 120 arranged view the passenger compartment 116. The monitor controller 112 is functional to operate the cameras to capture video imagery, and transmit the imagery to other systems of the railway control system, e.g., to the scheduling system 115. The monitor controller 112 also comprises a radio transceiver 122 which could, for example, comprise a Global System for Mobile communications protocol device (GSM). The scheduling system 115 is provided for the purpose of scheduling trains running on the railway track, e.g., trains 105 and 106. An objective of the scheduling system 115 is to optimise the 7 utilisation of the railway track by trains, to improve the efficiency of trains’ schedules and timeliness of the trains, whilst ensuring the safety of trains travelling along the railway track, by ensuring that conflicting train movements, e.g., two trains travelling on a same section of track at a same time, is avoided. A particular objective of the scheduling system is to appropriately allocate track time between competing trams to achieve a balance that best serves the overall objective of scheduling efficiency and timeliness of the train portfolio. The scheduling system 115 comprises onboard processing resource forprocessing data and a radio transceiver 123 for communicating with the trains 105, 106, with the monitor controller 112, the signal controller 109, and with other systems of the railway control system. The radio transceiver 123 could, for example, comprise a Global System for Mobile communications protocol device (GSM). The scheduling system 115 is configured for making dynamic scheduling decisions for the trains based on real-time data. Using various sources of real-time data, the scheduling system 115 can predict in advance factors affecting movement of the trains, and generate schedules for those trains ahead of time taking into account those factors. In particular, the scheduling system 115 is operable to predict dwell times for trains, e.g., trains 105 and 106, at platforms along the railway track, e.g., platforms 107 and 108. That is, the scheduling system 115 is operable to predict appropriate periods of time for trains to stop at the platforms based on various sources of data received dynamically by the scheduling system representing various parameters that have been found relevant to the train dwell time. The scheduling system 115 may then generate schedules for the trains, that is routing plans defining appropriate movements of the trains along the railway track with respect to time, based on those predicted dwell times. The generated schedules may then be used to facilitate control of the trains. For example, the signal controller 109 may use the schedules to control the signals 110, 111, and / or the trains’ motion control units may use the schedules to control drive units of the trains. In the example of a passenger train, the train is required to stop at platforms for sufficient time to allow passengers to embark and disembark safely. Conversely, it may typically be desirable to minimise the platform dwell time in order to avoid excessively delaying the train, and following trains. It may be desirable therefore that trains stop at platforms for approximately only the time required to allow passengers to embark and disembark. The scheduling system 115 may thus usefully predict a time required for passengers to embark and disembark trains 105, 106 at the platforms 107, 108, dynamically based on real time data relevant to the passenger embarkation / disembarkation time. Consequently, the trains may be scheduled to stop at platforms for only approximately the time necessary for passenger embarkation / disembarkation, which may have time-efficiency advantages both for the train and also when scheduling a portfolio of trains running along a same track. For example, a significant parameter relevant to a time required for passengers to embark and disembark a train is the number of passengers to embark / disembark. A large number of passengers may typically take longer to embark / disembark than a smaller number of passengers, for example, due to queuing to pass through train doors. It may thus be useful for the scheduling system to be able to approximate the number of passengers to embark / disembark. Based on that knowledge of passenger numbers, the scheduling system may determine an appropriate dwell time for the train at a platform, for example, using a trained machine-learning model with the passenger number as an input. Further, it has been recognised that the distribution of passengers within a train’s passenger compartment and on the platform can affect the embarkation / disembarkation time. For example, it has been recognised that where passengers on a platform are bunched together, as might occur near to a platform entrance, it may take relatively longer for those passengers to embark the train than if the same number of passengers were evenly distributed along a platform, because the passengers may tend to queue to use a same train door rather than using all doors of the train. And passengers in a train’s passenger compartment loitering around the train door may obstruct the door and similarly slow the embarkation / disembarkation of passengers. In the example, the scheduling system 115 may usefully detect a presence, number, and locations of passengers to embark / disembark using the monitoring system, by imaging the platforms 107, 108 and the passenger compartments 116 of the trains 105, 106 using the cameras 113 and 120 respectively. Known image processing techniques may then be employed by tire scheduling system 115 to infer the presence of passengers, number and locations of passengers in the image, from which the scheduling system may approximate a dwell time for the train at the platform. Alternative or in addition to imaging the train passenger compartment, the scheduling system 115 could make use of the mass sensing system 118 onboard the trains, to determine a mass of the train, from which a number of passengers onboard the train may be inferred. Alternatively or additionally to imaging the platform or sensing a mass of the train, the scheduling system 115 may receive data from a ticketing system regarding a number of tickets that have been sold for a train journey, and / or data from platform entrance equipment, e.g., ticket barriers, regarding a number of people who have entered a platform through the barriers. From each of these data sources the scheduling system may approximate a number of passengers to embark and / or disembark at a platform, from which an appropriate dwell time for a train at a platform may be 9 inferred. For example, a trained machine-learning model may use model weights that are varied in dependence on the number and / or location of passengers. Further it has been recognised that other parameters may affect passenger embarkation / disembarkation times. For example, it has been recognised that the prevailing weather conditions may impact the time taken by passengers to embark / disembark, e.g., precipitation may slow passenger embarkation / disembarkation. Similarly, parameters such as platform geometry may affect passenger embarkation / disembarkation times, for example, because a relatively short platform may cause passengers to bunch together and queue to embark / disembark the train at a same door. Tire scheduling system 115 may thus usefully receive data representing prevailing weather conditions at the platform and / or other relevant contextual data, which may then be used by the scheduling system 115 to infer an appropriate dwell time at the platform. For example, a trained machine-learning model may use model weights that are varied in dependence on the incoming weather data. In comparison to determining train dwell times dynamically using real-time data, a conventional approach to scheduling trains is to assume a dwell time at a platform that is agnostic of prevailing conditions, for example, the number / distribution of passengers and / or weather. In such a system, it is generally necessary to assume a worst-case scenario dwell time, for example, a number of passengers to embark / disembark that matches the full capacity of the train. Thus, in the conventional approach, a train may be scheduled and controlled to dwell at a platfomr for longer than is required, thereby potentially unnecessarily delaying the train and following trains. A particular advantage of the presently disclosed scheduling system may be observed in the context of the example scenario depicted in Figure 1. In the scenario, the track sections 102, 103 converge to common track section 124, and thus the trains 105, 106 should be scheduled appropriately by the scheduling system 115 to avoid conflicting train movements, e.g., both trains 105, 106 entering the common track section 124 at the same time. Unis, in the scenario, the trains 105, 106 should be scheduled by the scheduling system to depart their respective platforms 107, 108 at staggered times. Consider the scenario where train 105 has previously been scheduled to arrive and depart platform 107 two-minutes earlier than train 106 is scheduled to arrive and depart platform 108, but in which train 105 has been delayed by two-minutes at some time since generation of the initial schedule. Hence, trains 105 and 106 may in reality arrive at their respective platforms at a same time. The scheduling system is thus required to reschedule one or both of the trains to avoid both trains 10 entering the common track section at a same time. A conventional scheduling system agnostic to prevailing parameters affecting the train dwell times at the platforms may simply select at random one of the trains to hold at the platfonn. However, in the present disclosure, the scheduling system 115 may infer using the monitoring system 112 that train 105 has relatively few passengers to embark and disembark, whereas train 106 has a relatively large number of passengers to embark and disembark, and may thus determine that train 106 may require a longer dwell time at the platform than train 105. Consequently, the scheduling system 115 may schedule a relatively short dwell time at the platform 107 for train 105, and a relatively longer dwell time at the platform 108 for train 106. Thus, train 105 may be permissibly scheduled to depart before train 106, notwithstanding their previously conflicting schedules, thereby allowing train 105 to recover its delay relative to the initial schedule. Referring next to Figure 2, in examples the scheduling system 115 comprises processor 201, memory 202, radio transceiver 123, input / output device 203, and system bus 204. Processor 201 is configured for execution of instructions of the computer program for scheduling trains on the railway track. Memory 202 is configured for non-volatile storage of the computer program, defining machine-readable instructions, for execution by the processor, and for serving as read / write memory for storage of operational data associated with computer programs executed by the processor 201. Radio transceiver 123 is configured for communicating by radio with trains running on the track, e.g., trains 105, 106, and for communicating with the monitoring system 112 and the signalling system 109. Input / output interface 203 is configured for connection of the scheduling system to other external systems, for example, to a weather forecasting system for receiving weather data representing prevailing weather conditions at the platfonn. The components of the scheduling system 115 are in communication via system bus 204. Referring next to Figure 3, the hardware of the scheduling system 115 is configured to support five functional modules used for a method of scheduling trains on the railway track. In the Figure, the arrows depict schematically data flows between the modules. Data retriever 301 is functional to retrieve train data, for example data identifying a train such as a unique identification number, a planned route of the train, a current position of the train, and platform assignments for the train at stations along the route. Data retriever 301 is also functional to receive stop data, that is data relevant to the task of scheduling trains. Such stop data may include contextual data, such as weather data received from an external weather forecasting system, and platform and / or train passenger compartment video data from the monitoring system 112. For example, the data retriever 301 could be configured, upon receiving train identification and routing data, to determine platforms to be visited by the train, and then retrieve the data relevant to dwell times of the train at platforms on the train’s route such as data representing prevailing weather conditions at the platforms, and / or video imagery from the monitoring system 112. Data retriever 301 could for example output a request for data to an external weather system and / or the monitoring system 112, or such data could be spontaneously provided to the scheduling system 115 by the external weather system and / or the monitoring system 112 and stored in memory 202, whereby data retriever 301 could retrieve the stored data from memory 202. Passenger detector 302 is functional to receive platform and / or train passenger compartment imagery provided by the monitoring system 112. The passenger detector 302 is functional to process the imagery to infer a presence, number, and locations of passengers in the imagery. Various image processing techniques suitable for this task are known in the art, such as techniques reviewed in “Hou, L., Wan, W., Hwang, J., Muhammad, R., Yang, M., &Han, K. (2017). Human tracking over camera networks: a review. EURASIP Journal on Advances in Signal Processing, 2017(1). https: / / doi.org / 10.1186 / sl3634-017-0482-z”. Schedule generator 303 is functional to generate schedules for trains running along the railway track based on the contextual data from the data retriever 301 and / or the passenger detection data output by the passenger detector 302. In examples, schedule generator 303 may run one or more machine-learning models trained to predict dwell times for trains at platforms appropriate to allow passenger embarkation / disembarkation based on the various sources of input data. The schedule generator 303 is functional to determine the appropriate time period for the trains to stop at platforms, and generate schedule data for the trains based on those time periods. Hie schedule data could, for example, define departure times for trains from the platforms. The schedule data may then be output by the scheduling system 115 for use in controlling the trains. Railway monitor 304 is functional to collect data regarding the actual operation of the railway, e.g., the actual dwell time of the trains and the platforms. The collected data may then be used as feedback to further train the schedule generator 303. For example, the railway monitor 304 may receive platfonn and / or train compartment imagery from the monitoring system 112, and may thus determine whether a train was able in fact to complete its passenger embarkation / disembarkation within the scheduled dwell time, or whether contrarily the scheduled dwell time was insufficient. The model updater 305 is functional to receive the feedback data from the railway monitor, and modify the operation of the schedule generator, for example, update parameters of machine-learning models employed by the schedule generator 303, based on the feedback data. Referring next to Figure 4, in examples the computer program for scheduling trains on the railway track implemented by the scheduling system 115 comprises four operations 401 to 404. At operation 401, the computer program causes the processor 201 of the scheduling system 115 to retrieve data relevant to scheduling the trains. For example, operation 401 could involve the data retriever 301 sending a request to trains 105, 106 for identity, position, and route data, and a request to the monitoring system 112 for platform and / or train compartment imagery. At operation 402, the computer program causes the processor 201 of scheduling system 115 to generate schedules for trains running on the railway track. Processes involved in the scheduling system 115 generating tire schedules will be described in further detail with reference to Figure 5. At operation 403, the computer program causes the processor 201 of the scheduling system 115 to output the schedules generated at operation 402. For example, the scheduling system 115 may output the schedule data to the signalling system 109 to cause the signals 110, 111 to hold the trains 105, 106 at the platforms 105, 106 until tire determined departure time, and / or could be output to the onboard motion control system 117 of the trains 105, 106 to cause the train to remain at the platform until the determined departure time. At operation 404, the computer program causes the processor 201 of the scheduling system 115 to monitor the operation of the railway using the railway monitor 112 to collect feedback data and use the feedback data to update operations of the schedule generator 303. Referring next to Figure 5, operation 402 performed by the scheduling system 115 for generating schedules fortrains running on the railway track involves eight processes. At process 501, the computer program causes the processor 201 of the scheduling system 115 to retrieve train identity and position data from the trains 105, 106. Process 501 could, for example, involve the processor 201 operating the transceiver 123 to receive a radio signal from a train, whereby the tram reports its identity and position. At process 502, the computer program causes the processor 201 of the scheduling system 115 to determine the position of the train based on the position data received at process 501. Process 502 could, for example, involve the processor 201 processing aggregated data received at process 501 to extract the relevant position data, or performing some other processing operation. As an alternative to processes 501 and 502, process 501 could be omitted, and process 502 could involve the scheduling system 115 determining a position and identity of the trains 105, 106 by alternative means, for example, by communicating with other sensors to sense positions and identities of the trains 105, 106. At process 503, the computer program causes the processor 201 of the scheduling system 115 to determine whether tire position of the train 105, 106 determined at process 502 meets particular position criteria. An object of process 503 is to determine whether the train 105, 106 is in the appropriate position with respect to the platform 107, 108 to proceed with generating the schedule, including the dwell time determination. It may be desirable, for example, to avoid determining the dwell time excessively early, as performing the detennination too early may reduce the reliability of the detennination, for example, due to additional passengers arriving at the platform after the determination has been made. Thus, it may be desirable to determine a platform dwell time and generate the train schedule when the train is relatively close to the platform. The position criteria could therefore be a distance from the platform value, and process 503 may thus involve the processor 201 comparing the position of the train to the position of the platform and determining if that distance is equal to or less than the position criteria. Hie position criteria, e.g., the distance from the platform value, could be predefined and stored in the memory 202 of the scheduling system. If at process 503 it is determined that the current position of the train does not meet the position criteria, e.g., that the train is not sufficiently close to the platform, the processor may pause and revert to executing processes 501 and 502 to receive further position data from the train at a later time, and the processor may cycle through stages 501 to 503 repeatedly until process 503 is answered in the affirmative, indicating that the train does meet the position criteria, e.g., that the train is sufficiently close to the platform. When process 503 is answered in the affirmative, the processor 201 may proceed to process 504 of the computer program. At process 504, the computer program causes the processor 201 of the scheduling system 115 to retrieve the stop data for the train(s). Process 504 could, for example, involve the scheduling system 115 retrieving from memory 202 platform and / or train compartment image data received from the monitoring system 112, and / or mass data received from the mass sensing system 118 of the trains, and / or weather data representing prevailing weather conditions received from an external weather forecasting system. At process 505, the computer program causes the processor 201 of the scheduling system 115 to process the image and / or mass data retrieved at process 504 to infer whether passengers are present on the platform and / or in the train compartment, and if so, infer a number of passengers and a location of the passengers. Process 505 could involve the processor 201 determining locations of passengers relative to particular positions in the field of view of a camera. For example, processor 201 could determine whether passengers are closely bunched on the platform, and / or loitering around a door of the train within the passenger compartment. Where the monitoring system comprises multiple cameras imaging a common area, such as the multiple cameras 113 imaging each platform, process 505 could involve the processor 201 summing the passengers counted in each image in order to determine a total number of passengers in the monitored area. Alternatively or additionally, w here train mass data was retrieved at process 504, process 505 could involve the processor inferring a number of passengers in the passenger compartment of the respective train, for example, by subtracting a know n unladen mass of the train and dividing the remaining mass value by a predefined average passenger mass. At process 506, the computer program causes the processor 201 of the scheduling system 115 to determine the time period for the train(s) to stop at the platform(s), in other words the dw ell time, based on the stop data retrieved at process 504 and / or the passenger data determined at process 505. At process 507, the computer program causes the processor 201 of the scheduling system 115 to generate schedules(s) for the train(s) defining departure times for the train(s) from the platform(s), based on the dwell times determined at process 506. At process 508, the computer program causes the processor 201 of the scheduling system 115 to output the schedule(s) generated at process 507. For example, the scheduling system 115 may output the generated schedules to the signalling system 109, and / or to the motion control systems 117 of the trains. In the example, this may involve the radio transceiver 123 of the scheduling system 115 generating a radio transmission for reception by the radio transceiver 121 of the signalling system and / or the radio transceivers 119 of the trains. Subsequently, the signalling system 109 and / or the motion control system 117 may generate control signal(s) based on the received schedule(s) for controlling movement of the train(s). Although the present invention and its advantages have been described in detail, it should be understood that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined by the appended claims.Jn the claims, the word “comprising” does not exclude other elements or steps, and the indefinite article “a” or “an” does not exclude a plurality.

Claims

1. A method for scheduling trains travelling along a railway track, the method comprising:receiving stop data representing parameters relevant to a time period for a train to stop at a platform of the railway track,computing a time period for the train to stop at the platform based on the stop data, generating schedule data for scheduling the train based on the computed time period, and outputting a schedule signal representing the generated schedule data.

2. The method of claim 1, wherein the receiving stop data comprises receiving image data representing an image of one or both of a passenger compartment of the train and the platform, determining based on the image data a presence of passengers, and the computing a time period involves computing the time period based on the determination of a presence of passengers.

3. The method of claim 2, comprising determining based on the image data a number of passengers, and the computing a time period involves computing the time period based on the determined number of passengers.

4. The method of claim 2 or claim 3, comprising detennining based on the image data locations of passengers relative to the field of view, and the computing a time period involves computing the time period based on the determined locations of passengers.

5. The method of claim 3 or claim 4, comprising receiving further image data representing a further image of one or both of the passenger compartment of the train and the platform, determining based on the further image data a number of passengers in the further image and a sum of the number of passengers in the image and the further image, and the computing a time period involves computing the time period based on the sum of the number of passengers.

6. The method of any one of the preceding claims, comprising receiving mass data from the train representing a mass of the train, and the computing a time period involves computing the time period based on the mass of the train.

7. The method of any one of the preceding claims, comprising receiving weather data representing a weather condition relevant to the platform, and generating based on the weather data the stop data representing the weather condition.

8. The method of any one of the preceding claims, comprising receiving position data representing a position of the train and determining if the position of the train meets position criteria.

9. The method of claim 8, wherein the generating the schedule data is perfonned based ona determination that the position of the tram does meet the position criteria.

10. The method of claim 9, comprising determining if the position of the train meets positioncriteria in response to an initial determination that the position of the train does not meet the position criteria.

11. The method of any one of the preceding claims, comprising receiving the schedule signal and generating a control signal for controlling movement of the train based on the schedule signal.

12. A railway scheduling system configured to perform the method of any one of claims 1 to 11.

13. A computer program comprising instructions, which, when executed by a railway scheduling system causes the railway scheduling system to carry out the method of any one of claims 1 to 11.

14. A computer-readable data carrier having the computer program of claim 13 stored thereon.

Images