Fuel recommendation system

Abstract

A method, and computer system, for providing driving recommendations to drivers of vehicles of a fleet comprises obtaining driving data 121 from vehicles 101 of a fleet being driven by respective drivers along a set route; analysing the driving data in comparison with a recommendation model 103; and in dependence on the analysis of the driving data, generating a driving recommendation 123 for each driver of the fleet of vehicles whilst the given vehicle is being driven along the set route. The recommendation model is generated based on previous driving data from a plurality of vehicles previously driven along the set route. The driving data can be obtained by sensors on the vehicle. The generated driving recommendation may optimise driving to reduce fuel rate usage. The driving route may comprise segments, with the recommendation model also having segments based on previous driving data for the specific segments. The driving recommendation is preferably presented to the driver and recommendations can be continually generated. The previous driving data for the route can be amalgamated to generate the recommendation model, with the present driving data being amalgamated with the previous driving data and updating the recommendation model. Vehicles may be mining trucks.

Description

This invention relates to a method for providing driving recommendations to drivers of vehicles. Different drivers may drive a vehicle in different ways with different techniques. Different driving techniques can be more or less efficient, for example in terms of fuel usage of the vehicle, wear on the vehicle, and / or journey time of the vehicle. Driving with a more efficient technique can result in a reduction in the amount of fuel being used, reduction in the amount of wear of the vehicle, and / or a reduction in the journey time. Vehicles may be driven over the same set route repeatedly. In particular, vehicles which form part of fleet of vehicles may be driven over the same set route repeatedly. For example, trucks used in mining are often driven over the same set route within the mine site. Some drivers of the fleet of vehicles may have a more efficient driving technique than other drivers. It is preferable that all drivers drive with an efficient driving technique. It is also desirable for the fleet as a whole to be driven more efficiently. It is also desirable to be able to see how a driver is performing, and in comparison with a previous shift, and in comparison with other drivers. It is also desirable to see if a driver is following recommendations. It is therefore desirable for there to be an improved method of providing driving recommendations to the drivers of the vehicles. According to a first aspect of the present invention there is provided a method for providing driving recommendations to drivers of vehicles. The method comprises: obtaining driving data from a vehicle being driven by a driver along a set route; analysing the driving data in comparison with a recommendation model, the recommendation model generated based on previous driving data from a plurality of vehicles previously being driven along the set route; and, in dependence on the analysis of the driving data, generating a driving recommendation for the driver of the vehicle whilst the vehicle is being driven along the set route. According to a second aspect of the present invention there is provided a method for providing driving recommendations to drivers of vehicles of a fleet of vehicles, the method comprising: obtaining driving data from a plurality of vehicles in a fleet being driven by respective drivers along a set route; analysing the driving data from each of the plurality of vehicles in comparison with a recommendation model, the recommendation model generated based on previous driving data from a plurality of vehicles previously being driven along the set route; and in dependence on the analysis of the driving data from each of the plurality of vehicles, generating a driving recommendation for each of the respective drivers of the vehicles whilst the vehicles are being driven along the set route. The driving data may comprise one or more of: the location of the vehicle, the fuel rate usage of the vehicle, the incline of the vehicle, the payload of the vehicle, the velocity of the vehicle, the acceleration of the vehicle, the gear usage of the vehicle, the retarder usage of the vehicle, the retarder temperature of the vehicle, the brake usage of the vehicle, the engine speed of the vehicle, and the throttle position of the vehicle. The driving data may be obtained by one or more sensors located on the vehicle. The generation of the driving recommendation may comprise generating a driving recommendation which optimises the driving data so as to reduce the fuel rate usage of the vehicle. The recommendation model may comprise an optimal fuel rate usage driving data range. The driving recommendation may be generated for the driver of the vehicle in dependence on driving data from the vehicle being outside of the optimal fuel rate usage driving data range. The generation of the driving recommendation may comprise generating a driving recommendation which optimises the driving data so as to shift the driving data into the optimal fuel rate usage driving data range. The driving recommendation may be generated for the driver of the vehicle in dependence on the analysis of the driving data indicating that the vehicle is on the set route. The set route may comprise a route between a first point and a second point. The set route may comprise an error margin permitting deviation of the vehicle from the set route between the first point and the second point. The recommendation model may comprise segments of the set route. Each segment of the set route may comprise specific previous driving data. The driving recommendation may be generated in dependence on which segment of the set route the driving data indicates that the vehicle is in. The driving recommendation may comprise one or more of: a velocity recommendation for the vehicle, an acceleration recommendation for the vehicle, a gear recommendation for the vehicle, a retarder usage recommendation for the vehicle, a retarder temperature recommendation for the vehicle, a braking recommendation for the vehicle, an engine speed recommendation for the vehicle, and a throttle position recommendation for the vehicle. The driving recommendation may be provided to the driver. Providing the driving recommendation to the driver may comprise one or more of: a visual recommendation on a screen in the vehicle, an audible recommendation through a speaker in the vehicle, and a physical recommendation through a component of the vehicle. The driving data may be continually obtained from the vehicle, and one or more driving recommendations are continually generated for the driver of the vehicle. The method may further comprise obtaining the previous driving data from each of the plurality of vehicles previously driven along the set route. The method may further comprise amalgamating the previous driving data from each of the plurality of vehicles to generate the recommendation model. The method may further comprise analysing the previous driving data from each of the plurality of vehicles in comparison with the previous driving data from each of other plurality of vehicles to determine if each of the previous driving data is an outlier. The method may further comprise, in dependence on the analysis of each of the previous driving data determining that the previous driving data is not an outlier, amalgamating the previous driving data with the other previous driving data to generate the recommendation model. One or more of the plurality of vehicles may be distinct from one another. The driving data obtained from the vehicle being driven by the driver along the set route may be amalgamated with the previous driving data from each of the plurality of vehicles previously driven along the set route to update the recommendation model. The method may further comprise analysing the driving data obtained from the vehicle in comparison with the recommendation model to determine if the driving data is an outlier. The method may further comprise, in dependence on the analysis of the driving data determining that the driving data is not an outlier, amalgamating the driving data with the previous driving data from each of the plurality of vehicles previously driven along the set route to update the recommendation model. The method may further comprise obtaining driving data from a plurality of vehicles in a fleet being driven by respective drivers along the set route. The method may further comprise analysing the driving data from each of the plurality of vehicles in comparison with the recommendation model. The method may further comprise, in dependence on the analysis of the driving data from each of the plurality of vehicles, generating a driving recommendation for each of the respective drivers of the vehicles whilst the vehicles are being driven along the set route. The driving recommendation for the set route may be the same for each of the vehicles in the fleet of vehicles. The driving recommendation for the set route may be different for one or more of the vehicles in the fleet of vehicles. The method may further comprise analysing the driving data from each of the plurality of vehicles in comparison with the recommendation model by comparing respective locations of each of the plurality of vehicles. The method may further comprise generating the driving recommendation for each of the respective drivers by generating driving recommendations which optimises the locations of each of the plurality of vehicles so as to reduce the overall fuel rate usage of the fleet of vehicles. The recommendation model may be implemented by means of a machine learning algorithm. Each of the vehicles in the fleet of vehicles may be the same type of vehicle. The vehicle may be a mining truck. The set route may be an off-road route. The method may further comprise, in response to the driving recommendation being provided to the driver, obtaining further driving data from the vehicle being driven by the driver along the set route. The method may further comprise analysing the further driving data in comparison with the recommendation model. The method may further comprise, in dependence on the analysis of the further driving data, generating an adherence score for the driver of the vehicle. The further driving data may comprise a plurality of different types of driving data. The analysis of the further driving data in comparison with the recommendation model may comprise weighting the different types of driving data such that the adherence score is a weighted adherence score. The adherence score may be generated in dependence on whether the driving data is within adherence boundaries. The driving recommendation may comprise a target speed profile for the set route. The adherence boundaries may comprise a speed profile above the target speed profile and a speed profile below the target speed profile. According to a third aspect of the present invention there is provided a computing apparatus for providing driving recommendations to drivers of vehicles. The computing apparatus comprises one or more processors and a memory storing in non-transient form data defining program code executable by the one or more processors, wherein the program code is executable by the one or more processors so that the computing apparatus is configured to: obtain driving data from a vehicle being driven by a driver along a set route; analyse the driving data in comparison with a recommendation model, the recommendation model generated based on previous driving data from a plurality of vehicles previously being driven along the set route; and in dependence on the analysis of the driving data, generate a driving recommendation for the driver of the vehicle whilst the vehicle is being driven along the set route. According to a fourth aspect of the present invention there is provided a computing apparatus for providing driving recommendations to drivers of vehicles, the computing apparatus comprising one or more processors and a memory storing in non-transient form data defining program code executable by the one or more processors, wherein the program code is executable by the one or more processors so that the computing apparatus is configured to: obtain driving data from a plurality of vehicles in a fleet being driven by respective drivers along a set route; analyse the driving data from each of the plurality of vehicles in comparison with a recommendation model, the recommendation model generated based on previous driving data from a plurality of vehicles previously being driven along the set route; and in dependence on the analysis of the driving data from each of the plurality of vehicles, generate a driving recommendation for each of the respective drivers of the vehicles whilst the vehicles are being driven along the set route. The present invention will now be described by way of example with reference to the accompanying drawings. In the drawings: Figure 1 shows a flow diagram of a method by which driving recommendations are provided to drivers of vehicles. Figure 2 shows a schematic example of the set route along which a vehicle may be driven. Figure 3 shows an exemplary plot of driving data in a driving recommendation model. Figure 4 shows an exemplary plot of adherence boundaries in a driving recommendation model. Figure 5 shows an example method for providing driving recommendations to drivers of vehicles. Figure 6 shows a computer apparatus for carrying out the method described herein. The following description is presented to enable any person skilled in the art to make and use the invention, and is provided in the context of a particular application. Various modifications to the disclosed embodiments will be readily apparent to those skilled in the art. The general principles defined herein may be applied to other embodiments and applications without departing from the spirit and scope of the present invention. Thus, the present invention is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed herein. The present invention relates to a method for providing driving recommendations to drivers of vehicles. Figure 1 shows flow diagram 100 which illustrates the steps by which the driving recommendations are provided to the drivers of vehicles. A vehicle 101 may output raw telemetry 121. The vehicle 101 may output raw telemetry 121 when the vehicle 101 is in operation. The vehicle 101 may continually output the raw telemetry 121. The vehicle 101 may output the raw telemetry 121 in batches. The batches of raw telemetry 121 may be outputted periodically. The raw telemetry 121 may be outputted at certain, predefined, intervals. The batches of raw telemetry 121 may be outputted in response to a significant change in the readings of the raw telemetry 121. For example, if a value in the raw telemetry 121 increases above a certain threshold, the raw telemetry 121 may be outputted. The raw telemetry 121 may provide driving data from the vehicle 101. The raw telemetry 121 may be in the form of outputs from one or more sensors. The sensors may be located on the vehicle 101. The sensors may take measurements of characteristics about the vehicle 101. The sensors may include location measurement means to determine the location of the vehicle 101. Such sensor may comprise a GPS location device. The sensors may include altitude measurement means to determine the altitude of the vehicle 101. Such sensor may comprise a GPS altitude device. The sensors may include a fuel rate measurement means. Such sensor may comprise a fuel tank gauge, an engine controller configured to estimate the fuel rate, and / or a fuel management system. The sensors may include a vehicle 101 inclination measurement means. Such sensor may comprise an inclinometer. The sensors may include a vehicle 101 weight measurement means. Such sensor may comprise a weight scale, and / or a suspension load sensing device, such as a pressure sensor for the suspension working fluid. The sensors may include a vehicle 101 velocity measurement means. Such sensor may comprise a speedometer. The sensors may include a vehicle 101 acceleration measurement means. Such sensor may comprise an accelerometer. The sensors may include a gear detection means. Such sensor may comprise a gearbox sensor. The sensors may include a brake usage measurement means. Such sensor may comprise an accelerometer and / or and brake sensor. The sensors may include a retarder usage measurement means. Such sensor may comprise a retarder fluid temperature sensor. The sensors may include an engine speed measurement means. Such sensors may comprise an engine tachometer. Such sensors may include a throttle position sensor. Such sensors may comprise a throttle pedal potentiometer. The raw telemetry 121 may be inputted into a telemetry unit 102. The telemetry unit 102 may convert the raw telemetry 121 into driving data input 122. In this case, the driving data input 122 may provide driving data from vehicle 101. Alternatively, the raw telemetry 121 may not be required to be converted. In this case, the raw telemetry 121 may provide driving data from the vehicle 101. The type of raw telemetry 121 may influence whether the raw telemetry 121 needs to be converted. For example, the requirement for conversion may depend on whether the raw telemetry 121 is in the correct format. The raw telemetry 121 may be normalised. The raw telemetry 121 may be normalised by the telemetry unit 102. The raw telemetry 121 may be normalised to account for the inclination of the vehicle 101. The normalisation may be carried out by finding two set routes 211 in opposite directions up and down a slope. In this way, any inconsistences between inclination readings can be corrected. The normalisation may be carried out by the following steps: A batch of raw telemetry 121 may be separated, for example once a day, the batch of raw telemetry 121 may correspond to a small set of truck trips; GPS data in the raw telemetry may be clustered into groups of close positions; Additionally, a direction may be calculated from the change in GPS position in the raw telemetry 121; Data in the raw telemetry 121 at approximately the same position along the route (which may be based on a tuneable parameter), but with the truck moving in the opposite direction, may be used; The average difference in the inclination may be found based on this; This may give the offset which needed to be applied such that the inclination angle is closer in absolute value in both directions. The telemetry unit 102 may be provided on an Internet of Things (loT) platform 108. The loT platform 108 may provide access to a range of converters to be able to convert the raw telemetry into the driving data input 122. The driving data input 122 may be inputted into a recommendation model 103. The driving data input 122 may be continually inputted into a recommendation model 103. The driving data input 122 may be inputted into a recommendation model 103 in batches. The batches of driving data input 122 may be inputted into the recommendation model 103 periodically. The driving data input 122 may be inputted into the recommendation model 103 at certain, predefined, intervals. Alternatively, the driving data input 122 may be inputted into the recommendation 103 due to an external input. For example, an aspect of the raw telemetry 121 may exceed a certain threshold indicating that the driving data input 122 should be inputted to the recommendation model 103. The telemetry 102 may be configured to output and provide the driving data input 122 to the recommendation model 103 periodically. The driving data input 122 may comprise the raw telemetry 121 if conversion is not required. Depending on the format of the recommendation model 103, the raw telemetry 121 may need to be converted into a format which the recommendation model 103 can interpret. The driving data 122 may comprise one or more of: the location of the vehicle 101, the fuel rate usage of the vehicle 101, the incline of the vehicle 101, the payload of the vehicle 101, the velocity of the vehicle 101, the acceleration of the vehicle 101, the gear usage of the vehicle 101, the retarder usage of the vehicle 101, the retarder temperature of the vehicle 101, the brake usage of the vehicle 101, the engine speed of the vehicle 101, and the throttle position of the vehicle 101. Each of the types of driving data 122 may be determined from the respective sensor on the vehicle 101. The vehicle 101 may output the raw telemetry 121 when the vehicle 101 is being driven, or is in operation. The vehicle 101 may be driven by a driver. The vehicle 101 may be driven along a set route. The driving data 122 may comprise a driving data 122 profile. The driving data 122 profile may comprise data points at time intervals throughout the set route. In the case of the driving data 122 comprising vehicle speed, the driving data 122 may comprise vehicle speed points at time intervals throughout the set route. The time intervals may range between milliseconds to tens of seconds, depending on the accuracy requirement on the system. The time intervals may be between 1 second and 5 seconds. Preferably, the time intervals may be 3 seconds. Figure 2 shows a schematic example of the set route 211 along which a vehicle 101 may be driven. The set route 211 may comprise a first point 201. The set route 211 may comprise a second point 202. The first point 201 and the second point 202 may comprise a degree of accuracy. The first point 201 and the second point 202 may comprise respective areas. The vehicle 101 may set off from the first point 201 and arrive at the second point 202. The set route 211 may not directly run between the first point 201 and the second point 202. The set route 211 may follow a road and / or a path between the first point 201 and the second point 202. The set route 211 shown in Figure 2 is a schematic representation of a non-linear route. The set route 211 may be non-linear. The set route 221 may comprise an off-road route. Using the example of a mining truck as the vehicle 101, the set route 211 may follow a path from the loading zone at the base of the mine at point A to the unloading zone outside of the mine at point B. The set route 211 may follow the road and / or path out of the mine. There may be a plurality of set routes 211 for a particular vehicle 101. For example, the present system may include a plurality of different set routes 211. The recommendation model 103 may store each of the set routes 103. Using the example of a mining truck as the vehicle 101, there may be more than one road and / or path for leaving the mine, and so there may be more than one set route 211 for leaving the mine. The respective set routes 211 may apply to respective mines. The recommendation model 103 may be capable of determining which set route 211 the vehicle 101 is being driven on. The recommendation model 103 may determine which set route 211 the vehicle 101 is being driven on in dependence on the driving data 122. The recommendation model 103 may use the determined set route 211 to generate a driving recommendation 123. In an example, the recommendation model 103 may determine which set route 211 the vehicle 101 is being driven on in dependence on the payload of the vehicle 101. The payload may be full on the way out of the mine. The payload may be empty on the way into the mine. The set routes 101 may be split into ‘full’ payload and ‘empty’ payload set routes 211. The recommendation model 103 may determine whether the vehicle 101 is being driven on a 'full’ payload or an ‘empty’ payload set route 211. The present system may also be capable of determining new set routes 211. The recommendation model 103 may determine a new set route 211 from the driving data 122. If the driving data 122 indicates a new route not on a set route 211, a new set route 211 may be generated. The recommendation model 103 may store the new set routes 211. The recommendation model 103 may also use the new set route 211 to generate a driving recommendation 123. In an example, the recommendation model 103 may determine a new set route 211 by means of a clustering algorithm. The new set routes 211 may be determined by a cluster based on one or more of distance, location, energy, payload and elevation change. The clustering algorithm may split the overall journeys of the vehicle 101 into set routes 211 with the same end points 201,202. The clustering algorithm may split the overall journeys of the vehicle 101 into set routes 211 with similar GPS traces. The clustering may be based on density-based spatial clustering, which may use Frechet distances between pairs of GPS traces. GPS traces with similar Frechet distances may be clustered. The set route 211 may comprise an error margin 212, 213 from the set route 211. The error margin 212, 213 may comprise a degree of accuracy either side of the set route 211. For example, the error margin 212, 213 may comprisea band 5m either side of the set route 211. Figure 2 schematically illustrates the error margin 212, 213 over the path of the set route 211. The error margin 212, 213 may vary in magnitude along the set route 211. For example, at points which the set route 211 overlaps and / or intersects another set route 211, the error margin 212, 213 may be smaller. Similarly, at points which the set route 211 is distant from another set route 211, the error margin 212, 213 may be larger. In this way, there may be a lower chance of inaccurate set route 211 determinations. The error margin 212, 213 may be predetermined. The error margin 212, 213 may be generated by the recommendation model 103. The error margin 212, 213 may be determined by a user. The error margin 212, 213 may be determined automatically by the recommendation model 103. The error margins 211, 213 for all of the set routes 212 stored in the recommendation model 103 may be determined by the recommendation model 103 so as to avoid mis-determination of the set route 211 the vehicle 101 is being driven along. The error margin 212, 213 may permit deviation of the vehicle 101 from the set route 211. Using the example of a mining truck as the vehicle 101, the truck may need to temporarily go off the set route 211, i.e., to avoid a pothole or another truck. The temporary deviation from the set route 211 may not result in the recommendation model 103 determining that the vehicle 101 is no longer on the set route 211. The set route 211 may comprise segments 221, 222, 223. The segments 221, 222, 223 may be split across the set route 211. Using the example of a mining truck as the vehicle 101; the first segment 221 may be reversing away from the loading zone A, the second segment 222 may be an incline from the mine, and the third segment 223 may be a flat section to the unloading zone B. The recommendation model 103 may comprise previous driving data. The recommendation model 103 may be generated based on the previous driving data. The previous driving data may comprise driving data 122 from a plurality of vehicles 101 which have previously been driven along the set route 122. The recommendation model 103 may store the previous driving data. The recommendation model 103 may comprise a different model for each of the set routes 122. The recommendation model 103 may comprise specific driving data 122 for each of the segments 221,222, 223 of the set route 211. Using the example of a mining truck as the vehicle 101; the first segment 221 reversing away from the loading zone A may comprise a lower velocity, the second segment 222 of an incline from the mine may comprise a higher fuel rate and lower velocity, and the third segment 223 of a flat section to the unloading zone B may comprise a higher velocity and low fuel rate. The recommendation model 103 may comprise optimum driving data for the set route 211. In other words, there may be an optimum speed, acceleration, and / or braking for the driver to carry out during the set route 211. The optimum driving data may provide the lowest level of fuel rate usage for the set route 211. For example, for a particular vehicle 101, on a particular route 212, there may be an optimum speed, acceleration, and / or braking plan for the driver which minimises the amount of fuel being used. Alternatively, there may be another parameter to be optimised. For example, for a particular vehicle 101, on a particular route 212, there may be an optimum speed, acceleration, and / or braking plan which minimises the journey time. It is also possible for a combination of fuel consumption and journey time to be optimised, so the journey time minimised and the fuel usage is minimised. It is appreciated that other parameters may be optimised. Additionally, for a particular vehicle 101, on a particular route 212, there may be an optimum speed, acceleration, and / or braking plan which minimises the wear on the vehicle 101. The wear on the vehicle 101 may include tyre, brake, clutch and other vehicle 101 wear. The driving data 122 may be compared with the recommendation model 103. The driving data 122 may be compared against a pool of previous driving data. The recommendation model 103 may determine the similarities and differences between the current driving data 122 and the previous driving data. For example, the recommendation model 103 may compare the velocity of the vehicle 101 with the velocity of vehicles which have previously been driven along the set route 211. In particular, the recommendation model 103 may compare the current driving data 122 with the optimum driving data for the set route 211 the vehicle 101 is driving along. The recommendation model 103 may determine the differences between the current driving data 122 and the optimum driving data. If the set route 211 comprises segments 221, 222, 223, the recommendation model 103 may compare the current driving data 122 with the specific previous driving data for that specific segment 221, 222, 223. Using the example of a mining truck as the vehicle 101, in the second segment 222 with a higher incline, the driving data 122 may be expected to include a higher fuel rate which would be matched with the higher fuel rate in the previous driving data. There may be optimum driving data for each of the specific segments 211,222, 223. The current driving data 122 may be compared with the specific optimum driving data for each specific segment 221, 222, 223. The optimum driving data may comprise a range. In other words, there may be an optimal driving data range for the vehicle 101 to be driven in which still provides a significant improvement in efficiency, such as a reduction in fuel usage, vehicle 101 wear, and / or journey time. The recommendation model 103 may receive external inputs from external sources. The external sources may be external to the vehicle101. The external inputs may include information which may influence the driving recommendation 123. The recommendation model 103 may use the external inputs in the generation of the driving recommendation 123. The recommendation model 103 may use the external inputs to adapt the driving recommendation 123. For example, the recommendation model 103 may receive inputs about the driving conditions the vehicle 101 is travelling in, such as the weather. The recommendation model 103 may receive the driving conditions as an external input from an external source, such as weather from a weather forecast provider. Alternatively, or in addition, the recommendation model 103 may receive the driving conditions as an input from the vehicle 101. The recommendation model 103 may use the driving conditions input to adapt the driving recommendation 123 to better suit the driving conditions. For example, if the weather is very wet, very hot or very windy, the driving recommendation 123 may be adapted to lower the vehicle 101 velocity to meet safety requirements. Other forms of external inputs may include: road traffic levels, mapping of the route such as direction and elevation, and any issues with sections of the route such as road damage. The recommendation model 103 may output a driving recommendation 123 for the driver of the vehicle 101. The driving recommendation 123 may be determined in dependence on the analysis of the driving data 122. The driving recommendation 123 may comprise an optimal driving adjustment so as to reduce the fuel rate usage of the vehicle 101. Using the example of a mining truck as the vehicle 101, in the second segment 222 with a high incline, the driving recommendation 123 may suggest a lower gear or a lower speed so as to reduce the fuel consumption for the set route 211. In an example, the driving recommendation 123 may suggest maximising coasting rather than using the accelerator. The driving recommendation 123 may comprise an optimal driving adjustment so as to reduce the fuel rate usage of the vehicle 101, while not reducing the speed below a minimal threshold. The driving recommendation 123 may comprise an optimal driving adjustment so as to reduce the fuel rate usage of the vehicle 101, while not increasing the journey time above a maximum threshold. In this way, the fuel consumption may be reduced without overall slowing the transport process. Alternatively, the driving recommendation 123 may comprise an optimal driving adjustment for another driving parameter. The driving recommendation 123 may comprise an optimal driving adjustment so as to reduce the journey time of the vehicle 101. The driving recommendation 123 may comprise an optimal driving adjustment so as to reduce by the journey time and the fuel consumption of the vehicle 101. It is appreciated that other parameters may be optimised, such as vehicle 101 wear. The driving recommendation 123 may comprise a target speed profile. The target speed profile may comprise data points at time intervals throughout the set route 211. The driving recommendation 123 may comprise a target speed for the halfway point of the segment 221, 222, 223, and / or a target speed for the end point of the segment 221,222, 223. The recommendation model 103 may determine the target speeds. The target speed may be a mean value for the half of the segment 221, 222, 223. Put another way, the driving recommendation 123 may suggest driving at the halfway point target speed for the first half of the segment 221, 222, 223, and / or the driving recommendation 123 may suggest driving at the end point target speed for the second half of the segment 221,222, 223. The recommendation model 103 may output a driving recommendation 123 for the driver when the vehicle 101 is being driven outside of the optimal driving data range. For example, the recommendation model 103 may only output the driving recommendation 123 if the fuel efficiency of the vehicle 101 needs to be improved, if the journey time needs to decreased, and / or if the wear on the vehicle 101 needs to be slowed. The recommendation model 103 may determine if the driving data 122 is within the optimal driving data range, and output a driving recommendation 123 if the driving data 122 is not within the optimal driving data range. The driving recommendation 123 may comprise instructions for the driver of the vehicle 101 which would shift the driving data 122 into the optimal driving data range. For example, the recommendation model 103 may output a driving recommendation 123 which adjusts the driving data 122 to reduce the fuel consumption into the optimal range and / or adjusts the driving data 122 to reduce the journey time into the optimal range. The driving recommendation 123 may be outputted in dependence on the location of the vehicle 101. The driving recommendation 123 may only be outputted if it is determined that the vehicle 101 is being driven on a set route 211. The recommendation model 103 may determine if the vehicle 101 is being driven along one of the stored set routes 212. If the vehicle 101 is being driven along one of the set routes 212, it may be determined which set route 101 the vehicle 101 is being driven along. The driving data 122 may be compared against the relevant previous driving data for that set route 211. The driving recommendation 123 may then be outputted. However, if it is determined that the vehicle 101 is not being driven along one of the stored set routes 212, then the driving recommendation 123 may not be outputted. However, as described herein, a new set route 211 may be generated from the new route. The driving recommendation 123 may only be outputted in dependence on the payload of the vehicle 101. The driving recommendation 123 may only be outputted if it is determined that the vehicle 101 is being driven with the correct payload. The recommendation model 103 may determine if the vehicle 101 is being driven with a payload which matches the previous payload data. The previous payload data may include a payload data range. Alternatively, a particular payload may be inputted by a user for use in checking. If the vehicle 101 is driven with the correct payload, then the driving recommendation 123 may then be outputted. However, if it is determined that the vehicle 101 is not being driven with the correct payload, then the driving recommendation 123 may not be outputted. In this way, the driving recommendation 123 may be tailored for particular journeys of the vehicle 101. The driving recommendation 123 may only be outputted in dependence on the road conditions experienced by the vehicle 101. The driving recommendation 123 may only be outputted if it is determined that the vehicle 101 is being driven in the correct road conditions. The recommendation model 103 may determine if the vehicle 101 is being driven in road conditions for which the recommendation will be effective in improving the vehicle efficiency. In this way, the driving recommendation 123 may only be outputted for particular driving conditions. For example, if the vehicle is being driven after heavy rain, with slippery road conditions, then a driving recommendation 123 may only be outputted if it would improve efficiency in these conditions. The driving recommendation 123 may vary depending on the location of the vehicle 123. The driving recommendation 123 may be determined in dependence on the location of the vehicle 123. The driving recommendation may be determined in dependence on the location of the vehicle 123 along the set route 211. The driving recommendation 123 may depend on which segment 221, 222, 223 of the set route 211 the vehicle 101 is being driven in. As described herein, the previous driving data may vary in each segment 221, 222, 223. Therefore, the driving data 122 may be compared against different previous driving data for each segment 221, 222, 223. Therefore, the driving recommendation 123 may be based on the specific previous driving data for that segment 221, 222, 223. Using the example of a mining truck as the vehicle 101, in the second segment 222 with a higher incline, the driving recommendation 123 may suggest a low gear or a lower speed so as to reduce the fuel consumption. Alternatively, in the third segment 223 with a flat section, the driving recommendation 123 may suggest a higher gear with higher speed so as to reduce the fuel consumption and reduce the journey time. The driving recommendation 123 may comprise one or more of: a velocity recommendation for the vehicle 101, an acceleration recommendation for the vehicle 101, a gear recommendation for the vehicle 101, a retarder usage recommendation for the vehicle 101, a retarder temperature recommendation for the vehicle 101, a braking recommendation for the vehicle 101, an engine speed recommendation for the vehicle, and a throttle position recommendation for the vehicle 101. The driving recommendation 123 may comprise a specific recommendation for certain active parts of the vehicle 101. For example, an engine speed recommendation, a throttle recommendation, and a brake and / or retarder recommendation. The driving recommendation 123 may be outputted whilst the vehicle 101 is being driven along the set route 211. In other words, the driving recommendation 123 may be generated in real-time. In this way, the present system may provide real-time advice to the driver of the vehicle 101. The driving recommendation 123 may comprise a slight lag from when the raw telemetry 121 is received. The driving recommendation 123 may comprise a forward insight into later parts of the set route 211. The driving recommendation 123 may comprise adjustments for upcoming parts of the set route 211. It may be determined where the vehicle 101 is on the set route 211. Based on the current position of the vehicle 101, and the upcoming part of the set route 211, a driving recommendation 123 may be generated which provides improvements for future parts of the set route 211. The driving recommendation 123 may be provided to the driver of the vehicle 101. The driving recommendation 123 may be provided directly to the driver of the vehicle 101. The driving recommendation 123 may be provided to a guidance delivery unit 107. The guidance delivery unit 107 may convert the driving recommendation 123 into a user interface 124. The user interface 124 may include instructions for the driver of the vehicle 101. The user interface 124 may provide one or more of a visual recommendation, an audible recommendation, and a physical recommendation. The visual recommendation may be provided through a screen in the vehicle 101. The audible recommendation may be provided through a speaker in the vehicle 101. The physical recommendation may be provided through a component in the vehicle 101. The component in the vehicle 101 may be configured to interact with the driver. The component in the vehicle 101 may provide haptic feedback to the driver. The component in the vehicle 101 may be a control input device. For example, the control input device may be a steering wheel or a gear stick, and a physical input may be provided to the driver through forced feedback, or similar systems. Alternatively, or in addition, the component in the vehicle 101 may not be a control input device. For example, the component in the vehicle 101 could be a vibration device configured to vibrate the driver, such as a vibrating seat. The driving data input 122 may be continually received by the recommendation model 103. The recommendation model 103 may continually compare the driving data with the previous driving data. The recommendation model 103 may continually output a driving recommendation 123. The guidance delivery unit 107 may continually output the recommendations through the user interface 124. In other words, the driver may receive a constant stream of recommendations, should they be required. Alternatively, the driving data input 122 may be received by the recommendation model 103 in batches. The recommendation model 103 may compare the driving data 122 with the previous driving data in batches. The recommendation model 103 may output a driving recommendation 123 in batches. The guidance delivery unit 107 may output the recommendations through the user interface 124 in batches. In other words, the driver may receive a recommendation at certain points over the set route 211. The recommendations may be provided at particularly difficult and / or inefficient parts of the set route 211. The recommendation may be continually displayed, or otherwise provided, to the driver, and the recommendation may be updated periodically. For example, the recommendation may be provided on a screen, and the recommendation on the screen may be updated at various times. The driving recommendations 123 may also be provided to a data science unit 105. The data science unit 105 may comprise driving data 122 and driving recommendations 123 from a range of sites. Using the example of a mining truck as the vehicle 101, the data science unit 105 may comprise the driving data 122 and driving recommendations 123 from a range of different mines. The driving data 122 from different mines may be amalgamated to further improve the recommendation model 103. The data science unit 105 may provide a means for monitoring the performance at fleet level. The user interface 124 recommendations may also be provided to a dispatch unit 106. The dispatch unit 106 may output the user interface 124 recommendations to a range of different vehicles. Using the example of a mining truck as the vehicle 101, the dispatch unit 106 may output the user interface 124 recommendations to a number of trucks in a fleet of trucks. The user interface 124 recommendations may take account of each of the different trucks in the fleet. A fleet of vehicles 101 may be configured to output respective raw telemetry 121. Each of the vehicles 101 in the fleet of vehicles 101 may be driven along the set route 211. Alternatively, each of the vehicles 101 in the fleet of vehicles 101 may be driven along different set routes 212. The raw telemetry 121 may be received by the telemetry unit 102 and converted into driving data input 122, as described herein. The driving data 122 from each of the vehicles 101 may be inputted into the recommendation model 103, as described herein. The recommendation model 103 may analyse the driving data 122 from each of the vehicles 101 by comparing it with previous driving data, as described herein. The recommendation model 103 may output driving recommendations 123 for each of the vehicles 101, as described herein. The driving recommendations 123 may be converted into user interface 124 recommendations for each of the drivers of the vehicles 101, as described herein. In this case of a fleet of vehicles 101, the recommendation model 103 may compare the driving data 122 of each of the vehicles 101. The recommendation model 103 may output recommendations 123 which aim to reduce the overall fuel usage for the fleet of vehicles 101. The recommendation model 103 may compare the respective locations of each of the vehicles 101. For example, if the vehicles 101 are close together, it may be more efficient to slow the later vehicles 101, so that later vehicles 101 stop and start less often. The recommendations 123 may aim to ensure that each of the vehicles 101 in the fleet of vehicles 101 are kept spaced apart. The recommendation 123 for each vehicle 101 in the fleet of vehicles 101 may be the same for the set route 211. Put another way, each vehicle 101 of the fleet of vehicles 101 may receive the same recommendation 123 while on the same set route 211. In this way, each of the vehicles 101 may receive the optimal recommendation 123 at a vehicle level. When each vehicle 101 in the fleet of vehicles 101 receives the same recommendation 123, this may ensure that each of the vehicles 101 in the fleet of vehicles 101 are kept spaced apart. The recommendation 123 for one or more of the vehicles 101 in the fleet of vehicles 101 may be different for the set route 211. Put another way, one or more of the vehicles 101 of the fleet of vehicles 101 may receive a different recommendation 123 while on the same set route 211. In this way, each of the vehicles 101 may receive the optimal recommendation 123 at a fleet level. Each of the vehicles 101 in the fleet of vehicles 101 may be the same type of vehicle 101. Each of the vehicles 101 in the fleet of vehicles 101 may have the same capacity, be the same make and model, and / or the same specification. The capacity for each of the vehicles 101 may range from 40 tons to 500 tons. Each of the vehicles 101 in the fleet of vehicles 101 may have similar characteristics. In this way, the same recommendation 123 may, in theory, have the same effect on the driving (allowing for differences in drivers). Put another way, if the same changes are made by the driver for different vehicles 101, then there should be the same effect of efficiency on each of the vehicles 101. Alternatively, vehicles 101 in the fleet of vehicles 101 may be different types of vehicles 101. The fleet may include vehicles with a different capacity, be different makes and models, and / or have different specifications. The capacity for each of the vehicles 101 may range from 40 tons to 500 tons. In this case, the recommendations 123 may need to be specific for the different types of vehicles 101. Smaller vehicles 101 may need to be driven differently to larger vehicles 101 to optimise efficiency. For example, larger vehicles 101 may have greater efficiency gains by reducing acceleration and deceleration, due to the increased mass of the vehicles. The recommendations 123 may be tailored for the different types of vehicles 101 in the fleet. The raw telemetry 121 from the vehicle 101 may also include the type of vehicle 101. Alternatively, even with different vehicles 101 in the fleet, the same recommendation 123 may be given to all vehicles 101. This may reduce the complexity of the recommendations 123. The recommendation model 103 may be pre-generated before use. The recommendation model 103 may be generated from previous driving data. Previous driving data may be obtained from a plurality of vehicles 101 previously being driven along the set route 211. The previous driving data may be amalgamated to generate the recommendation model 103 for a specific set route 211. The process may be repeated to obtain previous driving data from a plurality of vehicles 101 previously being driven along a range of set routes 212. The previous driving data from each of the set routes 212 may be amalgamated to generate the recommendation model 103 for all the set routes 212. Figure 3 shows an exemplary plot 300 of driving data 304 in a driving recommendation model 103. The driving data 304 in the recommendation model 103 will generally fit a trend 303. In the simple example in Figure 3, the previous driving data 304 fits a line of best fit 303 between driving data values X 301 and Y 302. It will be appreciated that further driving data values 301, 302 may form the recommendation model 103 - for example, vehicle 101 velocity, vehicle 101 acceleration, vehicle 101 fuel consumption, vehicle 101 braking, vehicle 101 incline, vehicle 101 payload, and / or vehicle 101 gear. New driving data 305 may be included in the driving recommendation model 103 if it fits the trend 303, and may be determined to be previous driving data point 305a. However, if new driving data 305 does not fit the trend, it may not be included in the driving recommendation model 103, and may be determined to be an outlier driving data 305b. The new driving data 305 from a vehicle 101 may be compared with the previous driving data 304 to determine if the new driving data 305 is an outlier 305b. If the new driving data 305 is not an outlier 305b, then the new driving data 305a is amalgamated with the previous driving data 304 in the recommendation model 103. If the new driving data 305 is an outlier 305b, then the new driving data 305b is not amalgamated with the previous driving data 304 in the recommendation model 103. This process may be used to initially generate the recommendation model 103 before operation. In this case, the driving data 304 may have been obtained from a plurality of vehicles 101 driving over the set route 211. Each of the plurality of vehicles 101 may be distinct. In other words, a number of different vehicles may have driven over the set route 211. However, the same vehicle 101 may have also driven over the route a plurality of times. A combination of different vehicles 101 traveling over the same and different set routes 211 a plurality of times may also be used for the driving data 304. Additionally, this process may be used to update the recommendation model 103 during operation. In this way, the recommendation model 103 may be constantly updated and improved using the most recent driving data. The updates to the recommendation model 103 may be made continually. Alternatively, the updates to the recommendation model 103 may be made in batches. The updates to the recommendation model 103 may be made offline. The updated recommendation model 103 may put into operation at distinct times. The recommendation model 103 may be checked and validated before it is put into operation. The recommendation model 103 may be implemented using an algorithm. The recommendation model 103 may be implemented by means of a machine learning algorithm, such as through a deep neural network. Figure 4 shows an exemplary plot 400 of adherence boundaries 403a, 403b in a driving recommendation model 103. In response to the driving recommendation 123 being provided to the driver, further raw telemetry 121 may be output from the vehicle 101. Once the driving recommendation has been provided to the driver, the raw telemetry 121 may be known as further raw telemetry 121. This further raw telemetry 121 may be different to the raw telemetry 121 if the driver has followed the driving recommendation 123. Put another way, the further raw telemetry 121 is outputted after the driver has been provided the driving recommendation 123. The further raw telemetry 121 may be provided to the telemetry unit 102. The telemetry unit 102 may output further driving data 122 from the further raw telemetry 121. Each of the vehicles 101 in the fleet of vehicles 101 may output further raw telemetry 121 in response to receiving a driving recommendation 123. The recommendation model 103 may analyse the further driving data 122 in comparison with the recommendation model 103. In dependence on the analysis of the further driving data 122, the recommendation model 103 may generate an adherence score for the driver of the vehicle 101. The adherence score will indicate how well the driver has following the driving recommendation 123. For example, a high score may be given for good adherence, and a low score may be given for poor adherence. In this way, drivers who are more fuel efficient may receive a higher adherence score. The adherence score may be output to the data science unit 105 for further analysis, and / or review by driver managers. The further driving data 122 may comprise a plurality of different types of driving data 122. The different types of driving data 122 may be as described herein. The driving recommendation 123 may comprise a plurality of different types of driving recommendations 123. The different types of driving recommendations 123 may be as described herein. The analysis of the further driving data 122 in comparison with the recommendation model 103 may comprise weighting the different types of driving data 122 such that the adherence score is weighted. For example, vehicle speed may have a higher weighting than engine speed and throttle position. The weighting may be predetermined. The weighting may be adjusted depending on the importance of the parameters. The weighting may be adjusted depending on the implementation. The vehicle speed may have the biggest effect on the energy efficiency of the vehicle 101. Therefore, the vehicle speed may be the most important parameter and so have the highest weigthing. The adherence score may be generated in dependence on whether the driving data 122 is within adherence boundaries 403a, 403b. For example, as shown in Figure 4, the driving recommendation may comprise a target speed profile 401 for the set route 211. The adherence boundaries 403a, 403b may comprise a speed profile above 403b the target speed profile 401 and a speed profile below 403a the target speed profile 401. The adherence score may be generated in dependence on whether the vehicle 101 speed profile 402 is within the adherence boundaries 403a, 403b. The adherence score may be generated in dependence on the proportion of the set route 211 that the vehicle 101 speed profile 402 is within the adherence boundaries 403a, 403b. As described herein, the target speed profile 401 and / or the vehicle 101 speed profile 402 may comprise data points at time intervals throughout the set route 211. The adherence boundaries 403a, 403b may also comprise data points at time intervals throughout the set route 211. The adherence score may be generated in dependence on the proportion of vehicle 101 speed profile 402 data points being within the adherence boundaries 403a, 403b data points. The data points are illustrated in Figure 4 by the points at which the target speed profile 401 and the vehicle 101 speed profile 402 values change. The adherence score may be generated over time. The adherence score may be per shift. The driver may be generated an adherence score for each shift. Driver’s adherence scores shift after shift may be compared to see if the driver is improving overtime. In this way, it can be seen if the driving recommendations are improving the driver’s ability. Each of the vehicles 101 in the fleet of vehicles 101 may be given an adherence score. Each of the vehicles 101 in the fleet, and the drivers, may be compared based on the adherence score. Each of the vehicles 101 in the fleet, and the drivers, may be ranked based on the adherence score. In this way, the driving of the different drives can be compared and ranked. The adherence score of a driver may be used to direct further training for the drivers. For example, drivers with lower adherence scores may require further training. Drivers with higher adherence scores may be rewarded. The adherence scores may be normalised per fleet. The adherence scores may be normalised per set route 211. In this way, drivers driving on the same set routes 211 and / or in the same fleet may be compared fairly. The adherence score may be compared across different set routes 211. The adherence score may be compared across different fleets. In this way, the efficiency of different set routes 211 and / or different fleets may be compared. Figure 5 shows the steps of an exemplary method for providing driving recommendations to drivers of vehicles. At step 501, the method comprises obtaining driving data from a vehicle being driven by a driver along a set route. At step 502, the method comprises analysing the driving data in comparison with a recommendation model, the recommendation model generated based on previous driving data from a plurality of vehicles previously being driven along the set route. At step 503, the method comprises, in dependence on the analysis of the driving data, generating a driving recommendation for the driver of the vehicle whilst the vehicle is being driven along the set route. Figure 6 shows an example of a computing apparatus 600 for providing driving recommendations to drivers of vehicles described herein. The apparatus 600 may comprise at least one processor 601, and at least one memory 602. The memory 602 stores in a non-transient way code that is executable by the processor(s) to implement the apparatus 600, in the manner described herein. The applicant hereby discloses in isolation each individual feature described herein and any combination of two or more such features, to the extent that such features or combinations are capable of being carried out based on the present specification as a whole in the light of the common general knowledge of a person skilled in the art, irrespective of whether such features or combinations of features solve any problems disclosed herein, and without limitation to the scope of the claims. The applicant indicates that aspects of the present invention may consist of any such individual feature or combination of features. In view of the foregoing description it will be evident to a person skilled in the art that various modifications may be made within the scope of the invention.

Claims

1. A method for providing driving recommendations to drivers of vehicles of a fleet of vehicles, the method comprising:5 obtaining driving data from a plurality of vehicles in a fleet being driven byrespective drivers along the same set route whilst the vehicles are being driven along the set route;analysing the driving data from each of the plurality of vehicles in comparison with a recommendation model, the recommendation model generated based on 10 previous driving data from a plurality of vehicles previously being driven along the set route; andin dependence on the analysis of the driving data from each of the plurality of vehicles, generating a driving recommendation for each of the respective drivers of the vehicles whilst the vehicles are being driven along the set route.

2. The method according to claim 1, wherein the driving data comprises one or more of: the location of the vehicle, the fuel rate usage of the vehicle, the incline of the vehicle, the payload of the vehicle, the velocity of the vehicle, the acceleration of the vehicle, the gear usage of the vehicle, the retarder usage of the vehicle, the retarder20 temperature of the vehicle, the brake usage of the vehicle, and the engine speed of the vehicle, and the throttle position of the vehicle.

3. The method according to claim 1 or 2, wherein the driving data is obtained by one or more sensors located on the vehicle.

254. The method according to any preceding claim, wherein the generation of the driving recommendation comprises generating a driving recommendation which optimises the driving data so as to reduce the fuel rate usage of the vehicle.30 5. The method according to any preceding claim, wherein the recommendationmodel comprises an optimal fuel rate usage driving data range.13 12246. The method according to claim 5, wherein the driving recommendation is generated for the driver of the vehicle in dependence on driving data from the vehicle being outside of the optimal fuel rate usage driving data range.5 7. The method according to claim 5 or 6, wherein the generation of the drivingrecommendation comprises generating a driving recommendation which optimises the driving data so as to shift the driving data into the optimal fuel rate usage driving data range.10 8. The method according to any preceding claim, wherein the drivingrecommendation is generated for the driver of the vehicle in dependence on the analysis of the driving data indicating that the vehicle is on the set route.

9. The method according to any preceding claim, wherein the set route comprises a route between a first point and a second point, the set route comprising an error margin permitting deviation from the set route between the first point and the second point., 10. The method according to any preceding claim, wherein the recommendation20 model comprises segments of the set route, each segment comprising specificprevious driving data.

11. The method according to claim 10, wherein the driving recommendation is generated in dependence on which segment of the set route the driving data indicates 25 that the vehicle is in.

12. The method according to any preceding claim, wherein the driving recommendation comprises one or more of: a velocity recommendation for the vehicle, an acceleration recommendation for the vehicle, a gear recommendation for the 30 vehicle, a retarder usage recommendation for the vehicle, a retarder temperature recommendation for the vehicle, a braking recommendation for the vehicle, an engine speed recommendation for the vehicle, and a throttle position recommendation for the vehicle.

13. The method according to any preceding claim, wherein the drivingrecommendation is provided to the driver.

14. The method according to claim 13, wherein providing the driving recommendation to the driver comprises one or more of: a visual recommendation on a screen in the vehicle, an audible recommendation through a speaker in the vehicle, and a physical recommendation through a component of the vehicle.

15. The method according to any preceding claim, wherein the driving data is continually obtained from the vehicle, and one or more driving recommendations are continually generated for the driver of the vehicle.

16. The method according to any preceding claim, further comprising steps of: obtaining the previous driving data from each of the plurality of vehicles previously driven along the set route; andamalgamating the previous driving data from each of the plurality of vehicles to generate the recommendation model.

17. The method according to claim 16, further comprising steps of:analysing the previous driving data from each of the plurality of vehicles in comparison with the previous driving data from each of other plurality of vehicles to determine if each of the previous driving data is an outlier; andin dependence on the analysis of each of the previous driving data determining that the previous driving data is not an outlier, amalgamating the previous driving data with the other previous driving data to generate the recommendation model.

18. The method according to claim 16 or 17, wherein one or more of the plurality of vehicles are distinct from one another.

19. The method according to any preceding claim, wherein the driving data obtained from the vehicle being driven by the driver along the set route is amalgamatedwith the previous driving data from each of the plurality of vehicles previously driven along the set route to update the recommendation model.

20. The method according to claim 19, further comprising steps of:5 analysing the driving data obtained from the vehicle in comparison with therecommendation model to determine if the driving data is an outlier; andin dependence on the analysis of the driving data determining that the driving data is not an outlier, amalgamating the driving data with the previous driving data from each of the plurality of vehicles previously driven along the set route to update 10 the recommendation model.152521. The method according to any preceding claim, wherein:the driving recommendation for the set route is the same for each of the vehicles in the fleet of vehicles; orthe driving recommendation for the set route is different for one or more of the vehicles in the fleet of vehicles.

22. The method according to any preceding claim, wherein:analysing the driving data from each of the plurality of vehicles in comparison with the recommendation model comprises comparing respective locations of each of the plurality of vehicles; andthe generation of the driving recommendation for each of the respective drivers comprises generating driving recommendations which optimises the locations of each of the plurality of vehicles so as to reduce the overall fuel rate usage of the fleet of vehicles.

23. The method according to any preceding claim, wherein the recommendation model is implemented by means of a machine learning algorithm.30 24. The method according to any preceding claim, wherein each of the vehicles inthe fleet of vehicles are the same type of vehicle.

25. The method according to any preceding claim, wherein the vehicle is a mining truck, and / or the set route is an off-road route.

26. The method according to any preceding claim, further comprising steps of:5 in response to the driving recommendation being provided to the driver,obtaining further driving data from the vehicle being driven by the driver along the set route;analysing the further driving data in comparison with the recommendation model; and10 in dependence on the analysis of the further driving data, generating anadherence score for the driver of the vehicle.

27. The method according to claim 26, wherein the further driving data comprisesa plurality of different types of driving data, and wherein the analysis of the further driving data in comparison with the recommendation model comprises weighting the different types of driving data such that the adherence score is a weighted adherence score.CO28. The method according to claim 26 or 27, wherein the adherence score is20 generated in dependence on whether the driving data is within adherence boundaries.

29. The method according to claim 28, wherein the driving recommendation comprises a target speed profile for the set route, and the adherence boundaries comprise a speed profile above the target speed profile and a speed profile below the 25 target speed profile.

30. A computing apparatus for providing driving recommendations to drivers of vehicles, the computing apparatus comprising one or more processors and a memory storing in non-transient form data defining program code executable by the one or 30 more processors, wherein the program code is executable by the one or more processors so that the computing apparatus is configured to:obtain driving data from a plurality of vehicles in a fleet being driven by respective drivers along the same set route whilst the vehicles are being driven along the set route;analyse the driving data from each of the plurality of vehicles in comparison 5 with a recommendation model, the recommendation model generated based on previous driving data from a plurality of vehicles previously being driven along the set route; andin dependence on the analysis of the driving data from each of the plurality of vehicles, generate a driving recommendation for each of the respective drivers of the vehicles10 whilst the vehicles are being driven along the set route.CM

Images