GB2572747A

GB2572747A - Bicycle safety apparatus

Info

Publication number: GB2572747A
Application number: GB1804646.6A
Authority: GB
Inventors: Denholm Timothy; Sutherland Milne Timothy
Original assignee: Pelotron Ltd
Current assignee: Pelotron Ltd
Priority date: 2018-03-22
Filing date: 2018-03-22
Publication date: 2019-10-16
Also published as: GB201804646D0

Abstract

A proximity detector that determines an overtaking vehicles proximity (and optionally speed) as it passes the device. The portable device 1 has a video camera 2, a road vehicle proximity detector, a data storage device, a computer processor C and a warning light 4. The device may be used as a bicycle accessory that is fitted on a seat tube, seat post or the seat rails. The device may be powered by a battery 5 and a voltage regulator 6. The vehicle proximity detector may be in the form of two beam-type sensors (3, figure 2). The processor may be programmed to record and overlay the proximity and video data of an overtaking vehicle onto the data storage device. The warning light may be a single red LED or may be a pattern of lights or infrared light. The red rear bicycle light may have a static or moving warning symbol. The video camera may have a front lens that is recessed into a side to protect it from water spray when used on a bicycle. The components of the device may be contained in a housing that is held to the cycle with a bracket 14.

Description

The present invention relates to apparatus and methods for encouraging a driver of an approaching vehicle to pass a cyclist or motorcyclist safely.

Background

In the past, common approaches for discouraging drivers from overtaking cyclists in an unsafe manner have included wearing or carrying various visibility aids such as yellow jackets and lights, the technique of cycling in the middle of a lane when it would be unsafe for motorists to overtake, and also a device known as a “lollipop” “safety flag” or “flash flag”, which is a yellow or orange piece of flexible plastic warning sign arranged to stick out (via a pivoting and typically springmounted bracket) from near the rear axle of the bicycle, which encourages drivers to leave more space when overtaking the cyclist.

The various visibility aids are effective only by way of ensuring the driver has noticed the cyclist and thus are unreliable. A small proportion of drivers still do not consistently give cyclists enough space that they can reliably negotiate potholes and balance safely, potentially causing an accident that could kill the cyclist.

The “lollipop” device is, to date, the best attempt to address this problem. However, despite this it remains extremely unpopular among cyclists. This is believed to be because cyclists generally wish to be seen as athletic individuals and/or enthusiasts, rather than as scared individuals and/or victims. The lollipop in particular tends to be associated with learner cyclists, children, and the frail and elderly, and so there seems little prospect of it ever becoming a popular cycling accessory.

The continued risk that cyclists face due to dangerous overtaking by drivers, as well as the anxiety this causes to cyclists, contributes to the ongoing popularity of cars over bicycles. In turn this contributes to increased road congestion, exhaust fumes, noise pollution, and use of fossil fuels.

Providing a solution to this problem would encourage more people to cycle, and promote a greener world.

Object of the present invention

It is therefore an object of the present invention is to provide a device for, or method of, encouraging a driver of an approaching vehicle to pass a cyclist safely.

According to a first aspect of the present invention there is provided a portable device for a vulnerable non-motorist road user, for encouraging safer driving by motorists, comprising:

- a video camera arranged to record video data comprising a video;

- a road vehicle proximity detector, arranged to determine the proximity of a road vehicle at a time of a road vehicle overtaking the portable device;

- a data storage device;

- a computer processor arranged to record the video data, and the proximity, onto the data storage device; and

- a warning light arranged to warn motorists of the presence of the portable device.

This warns motorists that the device is monitoring recording video and proximity of overtaking, thus incentivising the motorists to drive more safely.

According to a second aspect of the present invention there is provided a device for encouraging safer driving, comprising:

- an image data recorder, arranged to record image data;

- a road vehicle proximity detector, arranged to generate data on the proximity of a road vehicle at the time of the road vehicle overtaking the device;

- a vehicle speed detector, arranged to determine a speed of a vehicle overtaking the vehicle speed detector;

- a data storage device; and

- a computer processor arranged to record the image data and vehicle proximity data, and speed, onto the data storage device.

This has the advantage of enabling identification of the level of safety of overtaking events, which empowers the user to complain to authorities when appropriate, with less doubt regarding the dangerousness of the overtake manoeuvre than is currently the case.

Moreover, if the device is provided with a visual indication of it's nature (i.e. being distinctive in design, and it's functionality having been widely advertised) this provides a particularly strong incentive for drivers in the vicinity of the device (rather than an incentive to drivers in general) to avoid dangerous overtaking manoeuvres. Preferably the device also has a light, preferably in the form of a red bicycle light, preferably presenting having a distinctive moving or static shape, such as a warning symbol.

Generally, the device is for mounting on a bicycle, and is for monitoring other vehicles. Thus the term 'vehicle' refers to vehicles that are remote from the device and that are remote from the vehicle (e.g. bicycle) that the device is mounted on, as opposed to monitoring the vehicle (generally, bicycle) that the device is mounted on. However of course it may additionally monitor the vehicle it is mounted on, particularly its speed.

While the invention is predominantly described with reference to use by cyclists and mounted on bicycles it may have utility with some other vehicles, especially mopeds. Optionally the device is for a bicycle or indeed is mounted on a bicycle. Optionally the device is for a bicycle or moped, or indeed is mounted on a bicycle or moped.

One option is to monitor the proximity data and to initiate recording of image data only when proximity is detected. This has the advantage of saving battery power, processor capacity and/or data storage space. Another option is to record image data continously and to selectively transmit the data associated with vehicles coming in close proximity. This has the advantage of enabling the device to capture and transmit events shortly preceding the moment when a vehicle came dangerously close.

Devices and methods of any of the aspects may be advantageously used in conjunction with any combination of features described herein.

Examples of embodiments and features that may be advantageously used in any of the aspects of the invention are set out in the dependent claims.

Proximity detector and overtaking assessment

A preferred approach is that the vehicle proximity detector is a proximity measurement sensor, preferably an echo or reflection based distance sensor, and is arranged to provide a measurement beam extending outwardly from the device to measure the position of a vehicle passing into the beam.

Although there are many options for how to detect vehicle proximity, it is necessary to detect when a vehicle has overtaken, in order that the image data from the overtaking event can be identified for being uploaded, without needing to upload image data from the whole cycling trip. Not only does this overcome the problem that large volumes of data would need to be uploaded, but also it allows the data to be compartmentalised into data relating to each overtaking event. Each compartmentalised section of data can then be analysed (E.g. at the server end) to determine the registration number of the vehicle in order to build up a database of vehicles associated with unsafe overtaking events.

The vehicle proximity detector could for example be:

- an audio sensor (possibly filtered to preferentially be sensitive to the noise of tyres on tarmac, and passed through an algorithm to detect gradual increase and decrease of such noises - it may even utilise a stereo microphone to estimate the direction of passing of the vehicle),

- a computer processor (e.g. the same as the aforementioned one) controlled to detect a passing vehicle from the image data (note that if only a single video camera is used it is still possible to measure distances by detecting a vehicle, estimating it's size, and using the size to determine it's distance, or even by using standard street furniture to accurately calibrate the camera orientation and thus determine the distance of a vehicle by the position of one of it's wheels), or

- an active reflection detection based distance sensor (e.g. ultrasonic or infra-red)

By providing an assessment of overtaking (overtaking distance, or preferably a combined assessment based on overtaking distance and overtaking speed), fewer segments of image/video data need to be uploaded, reducing the transmission requirements (and typically the burden on the user's broadband plan) and further alleviating the requirement to then analyse each overtaking event at the server end based on the image/video data (which would be computational intensive).

The most straightforward approach would be to only upload those overtaking events that met the overtaking manoeuvre criterion (E.g. came closer than a certain distance, or satisfied a formula involving both distance and speed), however other valid approaches are possible, such as always uploading at least the closest 10 overtaking events.

Generally, the vehicle proximity data is sent along with any portion of image data, however in the cases that the proximity data was generated by computer analysis of a camera (e.g. either using a stereoscopic camera, or just based on position of the bottom of a vehicle wheel within the field of view), then it is adequate to just send the image data because the vehicle proximity analysis could be repeated at the server end. Typically the time and date (if possible) and any location or route data is provided too.

Generally the data transmission module is arranged to transmit the image data to the remote server only in relation to cases where the vehicle proximity data (or data derived therefrom) was less than a threshold. This has the effect of reducing the amount of data to be transmitted.

Typically the device is configured to: preferentially store in the data storage device any image data in relation to cases where the vehicle proximity data (or data derived therefrom) was less than a threshold, and to; delete or overwrite from the data storage device any other image data from the image data recorder. This has the effect of reducing the amount of data storage required.

Optionally the vehicle proximity detector comprises an ultrasonic sensor. This has the advantage of offering a long range (typically up to 4m), however typical ultrasonic sensors only operate 10 times a second, and have a risk of interference with each other, or with sensors on other bicycles. These problems can be reduced by alternating the sensors (if there is more than one) and/or varying the timing of firing randomly.

Alternatively (or in addition) the vehicle proximity detector comprises an infra-red sensor. This has the advantage of operating more frequently than a typical ultrasonic sensor, although it may limit the range to about 1-1.5m. It is also believed that infra-red sensors are less prone to interference than ultrasonic sensors because they include a lens to focus the beam more tightly.

Generally the device is adapted to determine both the distance of on overtaking vehicle, and it's direction relative to the device. Or more generally it determines both distance in a first direction, and relative motion in a second direction perpendicular to the first direction. The two main ways to determine the relative motion are the use of two spaced proximity sensors, and the use of video based movement detection using the data from the camera. The detection of the relative motion is important even if the device does not measure the speed of the overtaking vehicle because it allows the device to discriminate between overtaking objects and objects being overtaken/passed. Merely using distance measurement will cause an excess of false detections as the objects being passed. This is because pedestrians and parked cars, will be detected in close proximity far more often than dangerously driven overtaking vehicles.

Optionally the distance detector comprises both a first distance sensor and additionally a stereoscopic image distance measurement device. The first distance sensor can be an Ultrasonic distance sensor (or an Infra-red sensor for example), whilst the stereoscopic image distance measurement device can be provided by either two image recording devices or by a single image recording device arranged proximal to one or more mirrors, such that different parts of the field of view of the image recording device provide a stereoscopic view. Note that the image recording device used to record footage of the overtaking event may also provide the stereoscopic image distance measurement device (E.g. by means of one or more mirrors or a second image recording device). The advantage of a distance sensor and a stereoscopic distance measurement device is that the accuracy and validity of the distance sensor's measurement can be double-checked. Optionally the device is arranged to perform this double-check. Alternatively this can be left to a later step which may be undertaken by for example police upon receipt of the footage - for example if the validity of the data is challenged.

A preferred option has the main image recording device oriented with a main field of view oriented to capture footage of an approaching and/or overtaking vehicle, whilst one or two mirrors are arranged proximal to the image recording device to provide it with stereoscopic view of the vehicle at it's closest approach, and generally the stereoscopic field of view is smaller than the main field of view. Preferably the stereoscopic field of view has the two viewpoints arranged vertically rather than for example horisontally, since motion of the vehicle will be horisontal potentially causing blurring in that direction. Preferably if the image recording device uses a rolling shutter, the orientation (horisontal vs vertical) is chosen so that the two images of the stereoscopic field of view are taken substantially simultaneously.

According to a third aspect of the present invention there is provided a device (optionally a bicycle accessory) for encouraging safer driving, comprising:

- an image data recorder, arranged to record image data;

- a first vehicle proximity detector,

- a second road vehicle proximity detector, arranged to generate road vehicle proximity data, wherein the road vehicle proximity detector comprises a stereoscopic image recording device (optionally the stereoscopic image recording device comprises the image data recorder);

- a data storage device; and

- a computer processor arranged to record the image data and road vehicle proximity data onto the data storage device.

This has the advantage of both determining, and enabling verification or providing verifiable evidence as to, the proximity of an overtaking vehicle during an overtaking manouvre, thus serving as a deterrent to dangerous driving.

Overtaking speed detector

Preferably the device comprises a vehicle overtaking speed detector, arranged to generate vehicle overtaking speed data; and the data transmission module is arranged to transmit to the remote server the vehicle overtaking speed data, or vehicle overtaking data derived therefrom. This has the effect of providing an improved measure of overtaking safety which avoids the effect on drivers being such that they may conclude that they can overtake at an unsafe speed as long as they overtake further than a predetermined distance.

Typically the vehicle proximity detector and the overtaking speed measurement device are provided by a pair of proximity sensors arranged to provide two measurement beams and to measure the distance and timing of vehicles passing through the two measurement beams, and the computer processor is arranged to calculate the speed of an overtaking vehicle by identifying two beam cutting events and determining the time between them and the distance that the vehicle moved between the two beam cutting events occurring.

Alternatively the overtaking speed measurement is performed by measuring relative angular speed of the overtaking vehicle from the video data recorded by the video camera. The proximity detector is arranged to measure the distance to the overtaking vehicle, and based on the distance and relative angular speed the overtaking speed can be determined. Optionally the proximity detector is an ultrasonic sensor and/or a stereoscopic image distance detector. Angular speed requires measuring horisontal movement in the video data corresponding tot the overtaking vehicle (which can either be extracted from compressed video data, or generated by analysing the video data), and is computed using data on the geometry of the video camera field of view based on at least a region of the field of view. One suitable way of measuring movement in video data is the optical flow data processing technique.

Optionally, means for measuring the speed of the device (i.e. of the bicycle etc) is provided. This can be achieved by various methods. One way is to determine horisontal motion in one or more regions of the video data, before and/or after when vehicle overtake occurs (or indeed at the same time for example if there is road surface in view and not obstructed by the overtaking vehicle). To compensate for any variation in height or angle placement of the device (i.e. on the bicycle etc), this can be calibrated by reference to data from other parts of a journey, using GNSS (e.g. GPS) location data from a GNSS device.

According to a fourth aspect of the present invention there is provided a device (optionally a bicycle accessory) for encouraging safer driving, comprising:

- a video camera, arranged to record video data;

- data relating to the geometry of the field of view of the video camera;

- a first road vehicle proximity detector, arranged to generate road vehicle proximity data;

- a data storage device;

- a road vehicle speed detector, comprising a computer processor arranged to determine a speed of an overtaking vehicle by measuring relative angular velocity of an overtaking road vehicle from at least a region of the field of view of the video data, and based on the data relating to the geometry of the field of view of the video camera and the road vehicle proximity data; and

- a computer processor (optionally the same computer processor) arranged to record the video data, road vehicle proximity data, and speed onto the data storage device.

This has the advantage of recording video footage whilst also determining the distance and speed of an overtaking vehicle, enabling reporting to law enforcement agencies, thus serving as a deterrent to dangerous driving.

Optionally the computer processor is arranged to:

process the vehicle proximity data, or data derived therefrom, and the vehicle overtaking speed data, or data derived therefrom, to:

identify vehicle overtaking manoeuvres, and for each vehicle overtaking event, generate an assessment of the safety of that vehicle overtaking manoeuvre, and;

determine whether the safety of that vehicle overtaking manoeuvre was below a threshold safety level;

identify image data from the image data recorder, that is relevant to each vehicle overtaking manoeuvre for which the safety thereof was below the threshold; and the data transmission module is arranged to transmit, for at least those manoeuvres for which the safety assessment was below the threshold:

the vehicle overtaking manoeuvre safety assessment; and the image data relevant to that overtaking manoeuvre.

This has the effect that due to this initial safety assessment, potentially less data needs to be uploaded. It is also possible for additional computerised analysis to be performed on the data transmitted to the remote server by applying an overtaking assessment algorithm in a computer processor. For example this may further refine the overtaking safety assessment and/or may reject any data that does not correspond to a vehicle overtaking manoeuvre.

Preferably the computer processor is arranged to compress the image data associated with overtaking events

Typically an additional computerised analysis may be performed on the data transmitted to the remote server which comprises identifying and extracting vehicle number plate registration numbers. This greatly facilitates providing the data to insurance companies (and/or law enforcement agencies and/or the public via a website) to ensure that there is a consequence for unsafe overtaking (E.g. so that unsafe overtaking carries a risk of an increase in a driver's insurance premium)

Light

Typically the device comprises a light, and the light provides a distinctive signal to drivers to provide a warning to the drivers that the device has the aforementioned features. This greatly increases the likely effect on drivers.

Thus, instead of merely being aware (e.g. by means of advertising or news reporting) that some bicycles now have devices that record unsafe overtaking and upload the data (e.g. for use by insurance companies), it is instead the case that the driver can more readily identify that a particular bicycle is fitted with such a device, thus strongly motivating the driver to overtake in a safe manner. It would be typical to provide the device with some kind of distinctive appearance to enable it to be identified by drivers.

Generally the light is in the form of at least one light emitting diode for providing a bicycle light. This has the advantage that the distinctive effect is generally visible to drivers approaching the bicycle well in advance. Especially in the case that it is a rear light, such drivers can plan their overtaking manoeuvres with ample fore-warning that they need to drive safely.

Preferably the device is adapted such that in use on in a predetermined in-use orientation (e.g. attached in a predetermined manner and orientation to a standard part of a bicycle) to direct light (optionally infrared) in the direction of the front numberplate of an overtaking car, and the intensity of light directed in that direction is selected, taking into account the image recording device arranged to record footage of overtaking vehicles (i.e. including it's sensitivity, ISO, effective aperture etc), is such that the retroreflective effect of a conventional numberplate (as defined by national or international standards) enables the image recording device to record a readable image of the numberplate.

Note that this light can comprise (or consist of) infrared light provided that the image recording device is sensitive to that infrared light - for example near-infrared, which many common cameras are sensitive to provided that they do not have infrared-absorptive films on their image sensors (an example being the Raspberry Pi Noir (TM) camera).

Additionally, this light (which may be the main light described above having a wide angle, or may be an additional more directed light, or a combination of both), may have a strobe effect, such that it is only illuminated for a short period, for example less than l/60th of a second. The image recording device is typically a video camera, typically having a frame rate of at least 10Hz, typically around 30Hz, and the short period should be less than the frame time, preferably less than a shutter time (if the image recording device uses a shutter or electronic shutter or rolling shutter). Preferably the strobe illumination time is less than 100^th of a second, optionally less than a 200^th of a second. Use of a short strobe illumination time reduces the effect of blur caused by vibration of the image recording device (E.g. if mounted on a moving bicycle). Optionally the strobe light is a different colour to the bike light.

Optionally the strobe is a different colour than the main light, and is directed towards a region where, in an in-use orientation, the numberplate is expected to pass through during an overtake manouvre. The main light is generally red, and the strobe may for example be green. The red light is produces more light than the different coloured strobe such that to a viewer the strobe is masked by the red light so the overall effect is a red light. However, the image recording device is arranged to process the recorded colours in the image (or image frame) so that the high intensity red light does not prevent detection of the strobe-frozen image in the other colour (e.g. green). This has the advantage that the red bike light can be usefully bright, and does not need to be arranged very far from the camera (which would require a larger device), indeed brightly enough to light up the overtaking vehicle sufficiently for the camera to record footage of that vehicle overtaking, without causing the numberplate to be unreadably saturated and bright. More generally, irrespective of whether a fainter strobe in a different colour is used, preferably the image of the numberplate is based on one or more colour channels other than the red colour channel.

According to a fifth aspect of the present invention there is provided a bicycle accessory for encouraging safer driving, comprising:

- an image data recorder, arranged to record image data;

- a vehicle overtaking distance sensor, arranged to generate distance data;

- a bicycle light;

wherein, the light is adapted such that at least a portion of the light from the light is directed with a predetermined intensity, wavelength range, and in-use direction, and the image data recorder is arranged such as to have an in-use field of view when the bicycle accessory is attached to a bicycle in a predetermined orientation and location, such that: at night, the directed light enables the image data recorder to capture a readable image of a retroreflective numberplate of a car as it approaches to overtake the bicycle at an overtaking distance of 1 meter, based on at least one national or international standard for retroreflection and geometry of road vehicle numberplates;

- a data storage device; and

- computer processor arranged to record the image data and the distance data onto the data storage device.

This has the advantage that the cyclist is provided with data on the numberplates of vehicles overtaking the cyclist and the distances that they did so at, thus enabling the cyclist to report dangerous driving, thus providing an incentive for drivers to drive safely.

Bracket

Generally the device comprises a bracket for connection to a bicycle saddle. Typically the bracket is arranged to lock onto both of a pair of standardised seat-post-connection bars of a bicycle saddle undercarriage, such that the connection between the bracket and the seat-post-connection bars governs the orientation of the device relative to the saddle.

This has the advantage that, because a bicycle saddle is generally very well aligned with the bike frame, the image recorder (and any other sensors) will also have a fixed orientation relative to the bicycle, with the result that the image recorder and any other sensors can be configured so that they will point in appropriate, consistent, directions relative to the bicycle, which enables more accurate detection of overtaking vehicle distance (and speed if measured) and ensures that the image recorder can be arranged so that it will collect image data including the front of the overtaking vehicle and the overtaking event as a whole. Furthermore if additional algorithms are applied to the data received at the remote server (such as to verify the overtaking proximity/assessment from image/video data) this will be made easier because the image recorders (i.e. video cameras) will all be pointing substantially the same direction relative to the bicycle frames (and road surface).

Helpfully, the bracket may have a press-release connection to a device housing, such that upon the press-release connection being pressed, the device housing can be removed from the bracket and from the bicycle. This has the advantage of ensuring that the device can be easily removed either for charging a battery of the device, or to avoid it being stolen when the bicycle is left outside unattended. Optionally the press release connection includes a data connection such that when the housing is fitted to the press-release connection the data connection is connected. This enables connection of a separate second camera or other devices arranged elsewhere on the bicycle (magnetic speed sensor or front light etc).

Bicycle speed sensor

Preferably the device is further provided with a bicycle speed sensor adapted to determine either the speed of a bicycle on which the device is mounted, or the radial speed of a wheel thereof, and to generate bicycle speed data therefrom; and the device is arranged either:

to transmit such bicycle speed data to the remote server to enable the server to calculate the absolute speed of overtaking vehicles; or to calculate the absolute speed of overtaking vehicles by deducting the bicycle speed from the relative overtaking velocity of the vehicles.

This has the advantage of providing for an improved assessment of overtaking safety (whether that assessment is carried out on the device or on the data received at the remote server), and as the motivation of the drivers to drive safely is related to how the safety is assessed, this would (via information becoming available to drivers via press releases or news reports or word-of-mouth) provide an improved motivating effect on drivers to overtake safely.

Second image data recorder

Preferably the device further comprises a second image data recorder. This can be positioned in the housing along with the main image data recorder. The two devices (e.g. video cameras) can point in the same direction or substantially the same direction. This would aid an algorithm operating on the server data to verify the distance and speed of the overtaking vehicles. Alternatively the second image data recorder can be in a different direction (e.g. the main one can point backwards relative to the housing/bicycle, and the second one can point forwards or to the side). The second image data recorder can be arranged separately (typically via a cable) for positioning on a different part of the bicycles. For example the housing with the main image recorder can be at the rear of the bicycle (e.g. under the saddle) and the second image recorder can be on the handlebars facing forward.

Pollution sensor

Optionally the device also comprises a pollution sensor. A pollution sensor (e.g. NOx and/or particulate sensors) enables the device to log the levels of pollution along its route (again using wifi SSIDs along a path, or else satellite navigation). Again this could be used to generate a map that could benefit either users or to benefit the public.

Another benefit of a pollution sensor is achieved if the device issues a warning to the cyclist in the event that levels of pollution increase exceed a threshold amount. While cyclist often are well aware of causes of sudden pollution (E.g. they are behind a bus) they have difficulty determining exactly when it is safe to start breathing again, because this depends on the precise wind conditions. It would be useful to be alerted exactly when the pollution levels have dropped so that the cyclist does not have to hold their breath any longer than necessary and does not risk breathing too early.

One problem with road cycling is that drivers sometimes overtake aggressively. This is typically because due to (perhaps understandable) impatience, they often overtake when it is not an ideal time to do so. Aside from being dangerous it has the consequence that the driver often changes down to a lower gear and accelerates hard, which in the case of diesel powered vehicles leaves the cyclist in a small (sometimes large) cloud of diesel fumes. If the cyclist is cycling hard it will be difficult to pause breathing repeatedly so most cyclists tend to just breathing in the pollution except in the worst cases, accepting the health hazard it presents.

Although this is partly just an accepted hardship for cyclists, a small proportion of vehicles have poorly maintained engines which cause an unreasonable amount of pollution and some of them undoubtedly exceed the statutory thresholds of the relevant country.

Optionally the device records vehicles which overtake in the event that the pollution sensor detects particularly dense pollution within a predetermined period of time (i.e. above a threshold value, or an increase in excess of a threshold increase - because many pollution detectors have relative outputs not absolute ones). Data on such overtaking events (or vehicles in general) is then transmitted to the remote server.

It should be noted that it is often not possible to infer that an increase in pollution is associated with a particular vehicle. However with lots of these devices reporting back to the server, it will be possible to identify vehicles regularly associated with increased pollution, and to notify the list to the relevant civil authorities.

Clearly the authorities will typically not take action regarding buses and lorries. However it could be expected that they may take action regarding cars or vans in specific cases where there is strong evidence that statutory emissions limits have been exceeded (for example prioritising a visit to the premises of the owner of those vehicles to perform a vehicle emissions test).

This would have two benefits for cyclists. Firstly it would motivate drivers of particularly polluting vehicles from changing down a gear and revving their engines hard to quickly overtake a cyclist who has such a device - this, combined with the main effect of the device discussed above contributes to ensuring that drivers treat cyclists respectfully, which would lead to more people being willing to cycle (and an overall greener planet as a result). Secondly, it would mean that the relevant authorities would finally be able to identify those vehicles which produce excessive pollution and to take action, thus leading to healthier cycling conditions for everyone.

According to a sixth aspect of the present invention there is provided a bicycle accessory for encouraging safer driving, comprising:

- an image data recorder, arranged to record image data;

- a road vehicle air pollution sensor;

- a data storage device; and

- computer processor arranged to record the image data and and road vehicle air pollution sensor data to the data storage device.

This has the advantage that the cyclist is provided with image data of vehicles overtaking the cyclist, and of a measure of air pollution, thus enabling the cyclist to report motorists using vehicles that are likely to be in contravention of air pollution safety standards.

Note that the rate that air pollution (e.g. diesel particulates) is produced by motor vehicles is dramatically increased when changing down a gear and accelerating quickly. Especially when a diesel engine suddenly accelerates, this causes the engine to require more air than is being supplied by the turbocharger at that time, resulting in a dramatic reduction of turbo boost pressure. This commonly causes a visible cloud of diesel smoke to be produced whilst passing the cyclist. Diesel particulates are known to be harmful to health, and the cyclist is particularly vulnerable not only because of the concentration and location where the particulates are produced, but also since the cyclist is performing aerobic exercise and thus cannot hold their breath every time a vehicle overtakes.

Changing down a gear and accelerating, is a common behaviour of motorists as they overtake cyclists. This bicycle accessory provides an incentive for motorists to minimise engine speed increases (due to changing down a gear and accelerating), whilst overtaking cyclists. Thus the motorist drives more safely both in terms of aggressive acceleration and in terms of the amount of hazardous air pollution they produce in the immediate vicinity of the cyclist.

Inputting the wifi router details into the device

Preferably the computer processor is arranged to:

receive image data from the image data recorder; process the image data to identify a bar code; and identify in the barcode at least a password;

to control the wireless module to use the password to access a wireless data source.

This has the advantage that the user can easily configure the device to access their home wifi router by generating a barcode, for example on their mobile phone, and positioning the barcode in front of the image data recorder (e.g. the video camera). Typically a 2D barcode (known as a QR code) would be used, as this can easily encode larger amounts of information than a ID barcode and because algorithms to detect and read such QR codes are widely available.

There are various ways that a barcode can be generated. For example websites already exists where you can type in some information and a barcode will be generated. A more secure method (for encoding a wifi password!) would be to provide a dedicated app for the user to download onto their mobile phone.

Generally, the display/light can be used to indicate to the user that a wifi SSID/password is needed before the device will operate, or that the device is waiting to check that it can connect to the wifi hotspot/router before it will enter normal operating mode.

User marking of events of interest

Preferably the device comprises a button; and the computer processor is arranged to respond to the actuation of the button by marking the most recent image data and vehicle proximity data as relating to an event of interest such that this data is not deleted.

Preferably the computer processor is configured to upload to the server any data marked as relating to an event of interest. This has the effect that any weaknesses in the computer processor's ability to identify poor driving based on sensors or image data (or indeed any additional analysis on data received at the remote server), is complemented by the ability of the cyclist to identify poor driving themselves. This places the cyclist in charge and means that any unsafe behaviour can be reported, for example a driver who overtakes at a safe distance and speed but then brakes hard immediately after overtaking.

Further embodiments and aspects

According to another aspect of the invention there is provided a bicycle comprising the device of any of the other aspects.

Preferably the bicycle safety device is configured to measure both the proximity and speed of an overtaking vehicle, to calculate a driving safety assessment and to upload these to a remote server (e.g. when the cyclist returns home, the device connects to the internet via the user's wifi router). The device has a very distinctive light display/effect (which may double as a rear bike light) so that drivers will recognise the device and respond by only overtaking when they are able to do so safely.

The headings above are included for ease of reading and are not limiting. Further embodiments are set out in the claims. Details of the embodiments relating to any of the aspects are intended to be optionally provided with each of the aspects.

Detailed embodiment of the invention

A detailed embodiment of the invention will now be described, by way of example only, with reference to the figures in which:

Figure 1 is an illustration of a device according to one embodiment of the invention, for encouraging safer driving;

Figure 2 shows an illustration of a device according to an embodiment of the invention arranged on a bicycle;

Figure 3 is an illustration of the device of figure 1 with a front cover removed to show internal components;

Figure 4 shows a side view of a device according to an embodiment of the invention, attached to a bicycle seat via a bracket to the two supporting bars of the bicycle seat that connect the bicycle seat to the seat post;

Figure 5 is a perspective illustration of an in-use scenario of a device according to an embodiment of the invention;

Figure 6 is an illustration of an in-use scenario of the device of figure 3 from a birds eye view; Figure 7 is an illustration of an in-use scenario of the device of figure 3 from in front of a user; Figure 8 is a perspective illustration of an in-use scenario of the device of figure 3 from behind a user;

Figure 9 shows a graph of data from two proximity sensors according to an embodiment of the invention;

Figure 10 shows a graph of weighted averaged sensor data of the sensor data of figure 7; and Figure 11 shows a graph of the rate of change of the weighted averaged sensor data of figure 8.

Referring to figure 2 which sets out components of a device 1 according to an embodiment of the invention and it's interactions with its environment, and figures 1, 3 and 4 which illustrate a device 1 according to a embodiments of the invention, and figures 5 to 8 which illustrate the use of such devices. The device 1, comprises a computer C, powered by a battery 5, which controls a wireless module 15 and lights/LEDs 4, and recieves data from its camera 2 and sensor(s) 3.

Using the sensor(s) and camera the device 1 is able to determine the proximity of overtaking vehicles in a direction perpendicular to the direction of travel, and preferably is able to determine the speed of the overtaking vehicles (preferably in absolute terms, using a sensor to determine the speed of the bicycle, but optionally the speed relative to the bike).

The device 1 is supported by a bracket 14 in use on a bicycle 8, which in turn supports the user U. When in range, the device 1 connects via its wireless module 15 to the user's (or another) wireless route 17, to upload via the internet 18 data on dangerous overtaking manoeuvres to one or more servers 19, so that this data may be conveyed to insurance companies 20 (and potentially, in extreme cases, to law enforcement organisations).

The data sent by the device include the proximity of the vehicle while overtaking, as well as a video or photo showing the vehicle overtaking. Generally, all available information will be transmitted. Extraction of a licence plate number from the video/photo data may be performed by the device, and/or by the server.

As the device is arranged in use to be visible to drivers D, e.g. by means of the LEDs operating in a distinctive pattern and/or shape, and because the driver is likely to be aware that the device 1 is a driving safety reporting device (either by the markings of the device, or else by publicity) the driver is aware that there are likely to be repercussions for driving dangerously, such as video data being used against the driver. Such drivers preferably are made aware that data and video of dangerous driving is automatically made available to insurance companies. This, together, influences (influence) the driver's driving behaviour and the likelihood of the driver controlling (control) the car to perform a dangerous overtaking manouvre is reduced and the safety of the user U is improved (affects safety).

Referring particularly to figures 1, 2, 3 and 4 which show two detailed embodiments, the device 1 has a camera 2 operable to record digital video. This may be at 1080 pixels by 720 pixels resolution at a frame rate of between 5 frames per second and 25 frames per second. The camera is surrounded by a ring of LEDs 4 (red LEDs or variable colour red/green/blue LEDs).

The ring of LEDs 4 is capable of displaying in a first operative mode which provides a distinctive appearance to enable approaching drivers to identify the device as an overtaking recording and uploading device. The ring of LEDs is also capable of displaying a second non-cooperative mode, in which the distinctive appearance is not provided (for example the LEDs may simply provide a conventional blinking display) which indicates to the user that the device needs to be connected to the internet to upload it's data to a remote server. This is made possible by a wireless transciever (see 15, figure 10) such as a wifi module, capable of connecting to the user's home wife router and broadband connection.

The device has a battery (batteries 5) and voltage regulator 6 as shown in the cut-away illustration of figure 2, and has two beam-type sensors 3, which in this embodiment are infra-red sensors. The infra-red sensors each consist of an emitter and detector, and operate by emitting a pulse of infra-red radiation in a beam direction, and monitoring for reflections from nearby objects. Each time the detector detects a reflection the emitter sends another pulse (essentially immediately), and the frequency with which this cycle repeats is recorded and outputted as an analogue voltage, which is converted to digital, and from a calibration table converted to calculate the distance of the nearby object which is in the path of the beam.

While the sensor could be just a detector, it is strongly preferred to accurately measure the distance of any vehicles interrupting each beam, for example to the nearest l-2cm. By placing the beams a distance apart, or more typically by arranging them at suitable differing angles from the device, it is possible to determine both the time that each beam is cut, and distance that this occurs at. From these two data points (position of the front of an overtaking vehicle at two particular times), it is possible to calculate the speed of an overtaking vehicle as well as it's proximity (this requires data on the relative arrangement and angles of the two beams). As an alternative (or in addition) the timing when the vehicle ceases to interrupt the beams can be used to measure the distance and relative speed of the vehicle. An example of one suitable sensor is the Sharp ^(TM) model GP2Y0A02YK which provides a narrow detection beam and accurate measurement (about 1cm accuracy) from 20cm to 150cm. Ultrasonic sensors are particularly suitable for measuring proximity, however they operate at a lower frequency (typically 10Hz rather than 100 to 200Hz) and thus will not provide accurate speed measurement where two of these sensors are to be used together to measure vehicle speed, unless the vehicle speed is very low.

Vehicle distance may be calculated from the centre-line of the frame of the bike (see above on how to ensure the device is at a controlled position on the bike), or may be calculated from a predetermined distance away from the centre-line of the frame of the bike (i.e. a value corresponding to the typical distance that a cyclist's body extends from the centre-line of the bike frame) or indeed from the measured extent of the cyclists body (if the device is capable of measuring this).

As an example, a minimum overtaking distance might be defined as 10” (25cm) from the centre line of the bike-frame, plus the overtaking vehicle's absolute speed in miles per hour (mph). So at lOmph the minimum is 20” (51cm) from the bike's centre-line, at 30mph it is 40” (102cm), and at 60mph it is 70” (152cm).

If the device does not (or cannot in a particular journey) calculate the overtaking vehicle's absolute speed, the overtaking vehicle's relative speed can be used instead.

The examples above involve adding one inch per mph because this calculation is simple for drivers to judge, with the advantage that drivers can find out what is expected of them in order to alleviate driver anxiety. The skilled person may use a more sophisticated calculation, perhaps combining both bike speed and overtaking vehicle speed, and perhaps taking into account the width of the cyclist (e.g. as measured by stereoscopic video in real-time).

The device records the video via camera 4 to a data storage device (not shown) such as an SD card ^(TM), and for each overtaking event that matches the overtaking criterion a section of matching video data is marked for upload. For example if an overtaking event is graded as “Unacceptable” then a section of video data corresponding to the 5 seconds (or another predetermined interval) preceding and following the overtaking event is marked for upload.

In a preferred embodiment the device also has a pollution sensor (typically a device which has a resistance that varies in the presence of particular atmospheric compounds and/or of particulates), and the device also records increases in pollution associated with each overtaking event, so that excessively polluting vehicles can also be identified and the data and video evidence can be uploaded. Due to the difficulty in reliably associating a sudden rise in air pollution with a particular nearby vehicle, the server would only list such vehicles if multiple user's devices had recorded the increase in pollution being associated with that vehicle. Vehicle's is identified (from their registration plates) as being for example buses might be ignored and no action taken.

The server 19 is not a required part of the device, however it can usefully play a role in helping to promote driving safety. The server 19 receives and stores the data from multiple devices, and typically will perform additional checks. For example it may apply various computer algorithms in order to double check that the overtaking assessment of the device was not erroneous (E.g. if the user's hand or bag-strap had passed through the sensor beams this could result in an erroneous detection of vehicle speed and proximity, however a computer algorithm could be designed and applied to distinguish such events from genuine overtaking events). Further the server generally will apply a vehicle registration plate detection algorithm to identify the vehicle registration of the overtaking vehicle. Such algorithms may for example be obtained from open source software sources.

The server then presents the overtaking assessment, along with the overtaking vehicle registration plate number and the video evidence, in tabular/database form. This data may be made available to one or more vehicle insurance companies 20 so that the vehicle insurance companies can choose whether (and how) to modify future insurance premiums based on the data. This completes the causal chain of events by influencing drivers to not drive dangerously close to the bicycle 8 and user U.

The device may have multiple cameras, but a wide angle camera is generally desirably (diagonal field of view of 120 degrees or more). In the design shown in figures 1 and 2 the camera is arranged to face horizontally in use, and at 45 degrees away from the plane defined by the bike frame (based on the bracket connection to the bike saddle) so as to face partially sideways and backward in use. Alternatively a wide angle camera may be directed substantially perpendicular to the plane of the bike frame in order to capture events both behind and in front of the bike, as well as particularly capturing the overtaking manoeuvres themselves. The device 1 should have a bracket 14 that connects to the standardised double support bars 13 of the underside of modern bike saddles 12. This bracket connection enables the device to be oriented to record events either to the right of the bike (for use in the UK) or can alternatively be connected the other way around in order to record events to the left of the bike (for use in most other countries of the world).

However the bracket 14 is arranged that once the user has selected between the substantially two possible orientations of the camera 2 and sensors 3, the orientation is then limited such as to be consistent from one bike to another. This has the advantage that it is easier to control the computer C to accurately interpret the sensor data if the possible orientations of the camera and device are constrained.

Preferably the device has a button or other user input device (not shown) arranged such that a cyclist can activate the input device conveniently while cycling. This may include a large button on the main part of the device (e.g. under the saddle), or may be a button on a further part of the device (not shown) for connection to a bike's handlebars or frame, for example connected by a cable.

The button has the effect of marking a segment of video data as relating to objectionable road user behaviour. This might include abusive drivers or dangerous driving, or unspecific criminal activity. A predetermined length of video segment is marked if the button is pressed (the length of the clip that gets marked might be dependent on the number, duration or strength of pressing of the button, or may be fixed - for example one press might indicate to mark 5 seconds of video data before and after the press, two presses might indicate 60 seconds either way, and three presses might indicate 15 minutes either way).

Video data marked by the cyclist in this way should be treated differently from video data identified by the device as relating to an unsafe overtaking event. This is because the former is based on a subjective assessment and the latter based on an objective assessment. The data marked by the user might be made available to the user via a user interface either by smartphone connection to the device, or by browser based access to the server. The user may be permitted to crop the duration of the video data and to add metadata (tags) such as “dangerous” or “interesting” and the data might be made available for the user to make use of in various different ways (E.g. to upload for general interest to youtube.com or similar, or to provide to the police if appropriate). This data typically will not be automatically shared with insurance companies due to the subjective nature of the assessment and data collection.

Referring to figures 5-8, an in-use situation is illustrated, where in figure 5 a user U sits on a bicycle 8 with a device 1 according to an embodiment of the invention. A car approaches from behind oriented to overtake the bicycle.

In figure 6, the car is shown from above overtaking the cyclist. The horizontal distance measured by the device 1 is the minimum distance in the direction substantially perpendicular to the plane of the bicycle frame, or put differently is substantially horizontal and substantially perpendicular to the direction of travel of the bicycle. To achieve this the sensor(s) is arranged at a predetermined angle relative to the device 1, and the device has a bracket which is arranged to limit the possible orientations of the device to the bicycle or bicycle seat, such that when the user connects the device via the bracket, this ensures the sensor is oriented in the aforesaid direction.

Note that the bracket may permit two arrangements, one oriented with the sensor pointing out to the right of the bicycle) for left hand driving/cycling in the UK and Australia, and the other (oriented with the sensor pointing out to the left of the bicycle) for use in most other countries. Of course, it is not a particular problem if the bracket happens to permit some other non-sensical orientations, such as the sensor pointing downwards, as the user would not select these orientations provided that the position of the sensor and/or camera or other useful directional markings are visible to the user.

In figure 7 the cyclist, bicycle and overtaking car are shown from in front of the cyclist. The distance to be measured by the device may be from the centre-line of the bicycle, but preferably is from the actual or (more realistically) the notional extent of the cyclist. Typically a cyclist and bicycle, accounting for their hips, elbows and the handlebars (the latter especially important with mountain bikes) can be considered to be 60cm wide, which means that the cyclist/bike notionally extends 30cm from the centre-line of the bicycle outwards. Thus measurement of the distance of a vehicle should be the distance from such a point (i.e. point 30cm outward from the centreline of the bicycle).

The distance to the vehicle is the closest that any part of the vehicle comes, typically the wingmirror. However as the sensor has a directional beam, the wing mirror may not be detected, and instead it will be the closest part of the body of the vehicle, in the direction of the sensor - i.e. at a height of the sensor (typically Im above the ground for an adult bicycle).

In figure 8 a rear perspective view is shown of an in-use arrangement. Here the LEDs (or other lighting or other markings) and optionally the camera, are clearly visible to the driver as the driver approaches the bicycle in preparation to overtake. This ensures the driver is aware of the device and increases the likelihood that the driver will recognise at least some of the capabilities of the device.

Turning to figures 9-11, three graphs are shown, illustrating how sensor data may be used to detect both the minimum distance of an overtaking vehicle, and it's speed relative to the bike. To determine the absolute speed, determine and add the absolute speed of the bike (E.g. using a magnetic wheel sensor and data on the diameter or circumference of the wheel).

Figure 9 shows 100 data readings from two sensors which are both directed horizontally and near to the perpendicular direction from the plane of the bike frame. Sensor 1 is aimed slightly backward of the perpendicular direction and thus is the first sensor to be affected when a vehicle overtakes. Sensor 2 is aimed slightly forward of the perpendicular direction and thus is affected by an overtaking vehicle slightly later.

Note that data element 1 is the most recent reading, while data element 100 was taken a period of time ago. Generally the sensor can be read 100 to 200 times a second, however typical modern computers are not real-time processors, and thus the time of each sensor reading should be recorded.

Around 26 sensor readings ago, sensor 1 ceased to return it's maximum distance, and instead returned a distance of 44cm which (based on the sensor's actual position 10cm from the centre-line of the bicycle) equates to 24 cm from the cyclist).

A few sensor readings later sensor 2 also exhibits the same pattern. A short period of time Sensor 1 and then Sensor 2 revert to their maximum range.

This indicates that an object overtook the cyclist, rather than being passed by the cyclist, and based on the proximity this suggests it was a potentially dangerous overtaking event.

Whilst this example exhibits relatively clean data and a clear overtaking signature, some events may include more sensor noise, and more variation. For example due to rain drops occasionally interfering with the range estimation, or more typically due to the vehicle having glass windows which may cause the sensor to fail to detect the range of the vehicle at all times.

The data is smoothed as shown in figure 10. Here a weighted average has been applied over a number of sensor data readings for each of sensor 1 and sensor 2 data. Again the overtaking signature is clear, and the time difference between them gives a clear indication of the time it took the vehicle to cut the two beams. Note that the beams are not parallel and thus the vehicle speed is not simply inversely proportional to the time difference, but rather the position of the front (and/or rear) of the vehicle at each point should be calculated and the speed calculated from the time it took to travel that distance.

In figure 11, the rate of change of the weighted data is shown. This takes into account the data collection rate. When less clean data is recovered from the sensors, comparison of the time difference between the minimum values (-3300 and -4200) was found to give the most reliable measure of the time it took the vehicle to cut the two beams. From this the speed in this instance was calculated to be 30.37mph at a distance of 13.5cm from the cyclist (in this case the sensor was at the centre plane of the bike, 30 further away).

This incident (which was generated with a simple test rig, not an actual vehicle) gives a ratio of “vehicle distance (cm) / vehicle speed (mph)” of 0.44, which is a very low value (E.g. it is well below a threshold value of 1), indicating a very dangerous situation.

However it is also possible here to calculate the approximate length of the vehicle, in this example this calculation wasn't performed, but if it had been performed it would have shown that the passing object was much shorter than a car. As such the incident may be validly discarded, as the vehicles that are most dangerous to cyclists are the larger ones. In particular, when a cyclist is overtaken by another cyclist at close range this is unlikely to be as dangerous as when a car or truck overtakes in this fashion.

That said, if the incident was not disregarded due to the short vehicle length, the video data would be inspected to find licence plates, and preferably an algorithm is implemented (e.g. machine learning or video movement detection) to identify whether the vehicle with the identified numberplate did in fact overtake at the time that the sensor data was recorded. Preferably the algorithm also estimates the vehicle distance in order to make sure that the sensor data could not have been caused by something else, e.g. the user's jacket or bag.

In this case, the licence plate of the overtaking vehicle is recorded, the algorithm determines that it was overtaking at the time the sensor data recorded the sudden proximity, and the speed and distance of the vehicle based on video data are compatible with that calculated from the sensor data. Accordingly the vehicle (in this case a motorbike) is identified as having overtaken dangerously, and the sensor and video data from that incident (e.g. the time duration that the sensor data recorded proximity plus 5 or 10 seconds of video data before and after) are uploaded to the server when the user returns home and the bike enters the vicinity of the user's wireless router.

More generally, a bicycle safety device is provided, adapted to measure how closely vehicles overtake, and preferably sends this data along with relevant portions of video to a server for it to be used to discourage dangerous overtaking (e.g. the server may identify the relevant vehicle's licence plate number and share all of this data with vehicle insurance companies).

The bicycle safety device preferably measures both the proximity and speed of an overtaking vehicle, and calculates a driving safety assessment. This assessment, along with the video evidence, is uploaded to the server at a later time (e.g. when the cyclist returns home, the device connects to the internet via the user's wifi router).

The device may have a very distinctive light display/effect (which may double as a rear bike light) so that drivers will recognise the device and respond by only overtaking when they are able to do so safely. A suitable warning symbol is the triangular 'play' symbol, or an exclamation symbol or a camera symbol.

Claims

1. A portable device for a vulnerable non-motorist road user, for encouraging safer driving by motorists, comprising:

- a video camera arranged to record video data comprising a video;

- a data storage device;

2. Portable device of claim 1, further comprising:

a vehicle speed detector, arranged to determine a speed of a vehicle overtaking the vehicle speed detector substantially at a time of a road vehicle overtaking the portable device;

wherein the computer processor is arranged to record the proximity onto the data storage device.

3. Portable device of claim 1 or 2, and wherein the computer processor is arranged to overlay at least the vehicle proximity data onto the video so as to generate overlaid video data, and to store the overlaid video data onto the data storage device.

4. A bicycle accessory comprising the device of claim 1, 2 or 3.

5. A bicycle accessory of claim 4 comprising a red rear bicycle light as the warning light.

6. A bicycle accessory of claim 5 wherein the red rear bicycle light has a static or moving shape comprising a warning symbol.

7. Bicycle accessory of claim 4, 5 or 6 comprising a bracket arranged to hold the accessory in a predetermined in-use orientation and location on the bicycle.

8. Bicycle accessory of claim 7 wherein the video camera is oriented in the in-use orientation to capture the front numberplate of a car approaching to overtake the bicycle.

9. Bicycle accessory of claim 8 wherein the image data recorder comprises a front lens and wherein the front lens is recessed into a side of the bicycle accessory with respect to a direction of travel of the bicycle, such as to substantially protect the front lens from road water spray from the rear wheel of the bicycle.

10. Bicycle accessory of claim 7, 8 or 9, wherein the bracket is arranged to connect to the seat post or seat rails at the rear of a bicycle.

11. Bicycle accessory of claim 10 wherein the bracket is arranged to be supported at the bracket substantially at it's centre of mass with respect to the direction of travel of the bicycle, and wherein the bicycle accessory comprises a battery arranged forward of the centre of mass, and the image data recorder is arranged rearward of the centre of mass.