A blind spot object alert determination device

Abstract

A blind spot object alert determination device (100) for a vehicle that comprises an event camera 302; a blind spot region detection module (208), which processes event camera data to identify blind s

Description

TECHNICAL FIELD The present disclosure relates broadly to a blind spot object alert determination device, a system for blind spot object alert determination and a computer-implemented method of determining a blind spot object alert. BACKGROUND Vehicles in transit, e.g. on roads, may have multiple blind spots, which are areas around a vehicle that an operator / driver cannot easily see e.g. by using mirrors or by direct line of sight. These blind spots can pose a significant risk as other objects such as vehicles, pedestrians and / or obstacles may be hidden in these blind spots. Blind spots for vehicles may include, for example, the rear blind spot which is an area directly behind a vehicle that is not visible in a rearview mirror. It is recognized that larger vehicles like trucks and Sports Utility Vehicles (SUVs) often have larger rear blind spots. Blind spots for vehicles may also include, for example, side mirror blind spots which are areas to the sides of a vehicle that are not visible in side mirrors. The size of these side mirror blind spots can vary depending on the design of the vehicle and the position of the mirrors. Blind spots for vehicles may also include, for example, front windshield pillar blind spots which arise from the pillars that support the front windshield (also known as A-pillars) that can create blind spots at intersections / junctions or when making turns. Such pillars can obstruct the sighting of pedestrians, cyclists, or other vehicles. Blind spots for vehicles may also include, for example, over-the-shoulder blind spots that arise from areas behind an operator / driver’s left and right shoulders that are not easily visible, even with the use of mirrors. Typically, it is recognized that checking over the shoulder can be important before changing lanes on a road. Blind spots for vehicles may also include, for example, blind spots for larger vehicles such as trucks, buses, and other large vehicles that have significantly larger blind spots due to their size and height. Such blind spots can include areas directly behind a larger vehicle and areas alongside an operator / driver’s side of the vehicle and / or a passenger side of the vehicle. Some typical accident zones due to blind spots for a left-hand road system include, for example, overtaking zones whereby the right-hand blind spot for example can be dangerous during overtaking maneuvers. For example, a motorcycle rider should remain cautious as there is a possibility of not being noticed by a larger vehicle if a larger vehicle decides to move to the right, or to switch lanes, unexpectedly. Typical accident zones due to blind spots may also include areas beneath a larger vehicle’s front windshield. For example, the blind spot in front and beneath a large / tall truck’s windshield typically extends over two meters forward, making it sufficient to obscure another object’s presence and therefore, becoming unsighted to the operator / driver. Typical accident zones due to blind spots may also include turning zones, for example a left turning zone for a left-hand road system. As an example, a motorcycle rider may face heightened risk from the left-hand blind spot of a vehicle and this danger is particularly evident when larger vehicles initiate turns. It is recognized that the available space on the inside of a turn diminishes as longer vehicles make tight turns towards a curb, which also reduces the effectiveness of side mirrors. To address blind spots, existing blind spot monitoring and assistance systems may use active sensors (such as a radar sensor, an ultrasonic sensor and / or a Light Detection and Ranging (LiDAR) sensor) and / or a vision-based sensor (such as an active camera). The inventors recognize that there are challenges / problems facing existing active sensor based systems. In radar-based active sensor systems, there may be difficulty in detecting small objects / obstacles such as a pedestrian and / or a bicycle. It is also recognized that false alarms may typically be generated when a vehicle utilizing such a system is passing through guardrails or tunnels. Further, it is recognized that location inaccuracy typically arises due to the low lateral accuracy of radar sensors. In addition, radar-based active sensor systems provide only a limited field of view and radar sensors are typically not able to detect significantly fast approaching vehicles and motorcycles. Furthermore, it is recognized that radar-based active sensor systems have limitations when recognising and classifying objects. In addition, it is recognized that a long time (latency) is typically incurred in actual detection of an object and delivering an alert or a warning sign to an operator / driver. In ultrasonic-based active sensor systems, in addition to above problems of a radar-based active sensor system, ultrasonic sensors have further a limited range which further reduces a field of view. In addition, it is recognized that ultrasonic-based active sensor systems may also not be as efficient as radar-based systems due to noisy reflections of emitted sound waves. For LiDAR-based active sensor systems, it is recognized that such systems may be significantly more expensive as compared to other systems due to the more expensive sensors used. Further, it is recognized that LiDAR-based active sensor systems have a limited vertical Field of View (FOV) that may result in missing some blind spot regions (e.g. for taller vehicles). In addition, it is recognized that LiDAR-based active sensor systems may require higher bandwidth for processing / transmitting data which increases a system latency. Further, it is recognized that LiDAR-based active sensor systems typically have performance that degrades or deteriorates in relatively poor weather conditions (e.g. when in a foggy environment). For systems that use an active camera, it is recognized that such systems may typically not perform well in environments with low illumination and / or overexposed lighting. Further, it is recognized that systems that use an active camera may be sensitive to weather conditions that can affect image quality. In addition, usage of such cameras may give rise to poor detection results due to motion blur scenarios, e.g. for capturing images of moving objects / vehicles. Further, it is recognized that systems that use an active camera may typically require a higher bandwidth for processing / transmitting data and may have high latency due to the data produced. In view of the above, there exists a need for a blind spot object alert determination device for use with a vehicle that seeks to address at least one of the above problems. 5 SUMMARY In accordance with an aspect of the present disclosure, there is provided a blind spot object alert determination device for use with a vehicle, the device comprising an event camera; a blind spot region detection module, the blind spot region detection module arranged to receive event camera data obtained by the event camera and further arranged to process at least the event camera data to identify one or more blind spots around the vehicle; a collaborative perception module coupled to the blind spot region detection module, the collaborative perception module arranged to receive identification data of the one or more blind spots from the blind spot region detection module; a transceiver coupled to the collaborative perception module, the transceiver arranged to request and receive data from at least one external agent, the request being based on the identification data and the data from the at least one external agent comprises perception data of the at least one external agent; an object detection module coupled to the collaborative perception module, the object detection module arranged to perform object detection at the identified one or more blind spots around the vehicle based on processed received data from the at least one external agent processed at the collaborative perception module; a trajectory estimation module coupled to the object detection module, the trajectory estimation module arranged to estimate a trajectory of at least one object detected at the identified one or more blind spots around the vehicle as a trajectory estimate of the at least one object, the trajectory estimate being based on the object detection and based on the received data from the at least one external agent; and an alert determination module coupled to the trajectory estimation module, the alert determination module arranged to process the trajectory estimate transmitted from the trajectory estimation module and to determine whether to output an alert to an operator of the vehicle. The data from the at least one external agent may further comprise behavioural data transmitted from the at least one external agent. The trajectory estimation module may be arranged to further estimate the trajectory of the at least one object detected at the identified one or more blind spots around the vehicle further based on the behavioural data transmitted from the at least one external agent, the further estimate may include an uncertainty score attached to the trajectory estimate and wherein the alert determination module may be arranged to determine whether to output the alert to the operator of the vehicle further based on the uncertainty score attached to the trajectory estimate. The device may further comprise a processing unit, the processing unit comprising one or more of the blind spot region detection module, the collaborative perception module, the object detection module, the trajectory estimation module and the alert determination module. The collaborative perception module may be capable of generating a view around the vehicle based on the received data from the at least one external agent and the identification data of the one or more blind spots around the vehicle. In accordance with another aspect of the present disclosure, there is provided a system for blind spot object alert determination, the system comprising a vehicle; at least one external agent external to the vehicle, the external agent comprising at least a sensor arranged to obtain data to detect an external agent object in a vicinity of the external agent; wherein the vehicle is provided therein a blind spot object alert determination device, the device comprising an event camera; a blind spot region detection module, the blind spot region detection module arranged to receive event camera data obtained by the event camera and further arranged to process at least the event camera data to identify one or more blind spots around the vehicle; a collaborative perception module coupled to the blind spot region detection module, the collaborative perception module arranged to receive identification data of the one or more blind spots from the blind spot region detection module; a transceiver coupled to the collaborative perception module; an object detection module coupled to the collaborative perception module; a trajectory estimation module coupled to the object detection module; and an alert determination module coupled to the trajectory estimation module; wherein the transceiver is arranged to request and receive the data from the at least one external agent, the request being based on the identification data and the data from the at least one external agent comprises perception data of the at least one external agent; wherein the object detection module is arranged to perform object detection at the identified one or more blind spots around the vehicle based on processed received data from the at least one external agent processed at the collaborative perception module; wherein the trajectory estimation module is arranged to estimate a trajectory of at least one object detected at the identified one or more blind spots around the vehicle as a trajectory estimate of the at least one object, the trajectory estimate being based on the object detection and based on the received data from the at least one external agent; and wherein the alert determination module is arranged to process the trajectory estimate transmitted from the trajectory estimation module and to determine whether to output an alert to an operator of the vehicle. The at least a sensor of the external agent may comprise another event camera. The at least one external agent may comprise an infrastructure object. The data from the at least one external agent may further comprise behavioural data transmitted from the at least one external agent. The trajectory estimation module may be arranged to further estimate the trajectory of the at least one object detected at the identified one or more blind spots around the vehicle further based on the behavioural data transmitted from the at least one external agent, the further estimate may include an uncertainty score attached to the trajectory estimate and wherein the alert determination module may be arranged to determine whether to output the alert to the operator of the vehicle further based on the uncertainty score attached to the trajectory estimate. The collaborative perception module may be capable of generating a view around the vehicle based on the received data from the at least one external agent and the identification data of the one or more blind spots around the vehicle. In accordance with another aspect of the present disclosure, there is provided a computer-implemented method of determining a blind spot object alert, the computer-implemented method comprising providing an event camera in a vehicle; using the event camera to obtain event camera data and using a blind spot region detection module to process at least the event camera data to identify one or more blind spots around the vehicle; receiving identification data of the one or more blind spots from the blind spot region detection module and using a collaborative perception module to instruct a transceiver coupled to the collaborative perception module to request and receive data from at least one external agent, the request being based on the identification data and the data from the at least one external agent comprises perception data of the at least one external agent; processing received data from the at least one external agent using the collaborative perception module; performing object detection at the identified one or more blind spots around the vehicle using an object detection module based on the processed received data processed at the collaborative perception module; estimating, using a trajectory estimation module, a trajectory of at least one object detected at the identified one or more blind spots around the vehicle as a trajectory estimate of the at least one object, the trajectory estimate being based on the object detection and based on the received data from the at least one external agent; processing the trajectory estimate using an alert determination module; and determining whether to output an alert to an operator of the vehicle using the alert determination module. The data from the at least one external agent may further comprise behavioural data transmitted from the at least one external agent. The step of estimating may further comprise further estimating the trajectory of the at least one object detected at the identified one or more blind spots around the vehicle further based on the behavioural data transmitted from the at least one external agent, the further estimating may include attaching an uncertainty score to the trajectory estimate; and the step of determining whether to output an alert to an operator of the vehicle using the alert determination module may be further based on the uncertainty score attached to the trajectory estimate. The computer-implemented method may further comprise generating a view around the vehicle using the collaborative perception module based on the received data from the at least one external agent and the identification data of the one or more blind spots around the vehicle. BRIEF DESCRIPTION OF THE DRAWINGS Exemplary embodiments of the present disclosure will be better understood and readily apparent to one of ordinary skill in the art from the following written description, by way of example only, and in conjunction with the drawings, in which: FIG. 1 is a schematic block diagram of a blind spot object alert determination device in an exemplary embodiment. FIG. 2 is a schematic block diagram illustrating a blind spot object alert determination device in another exemplary embodiment. FIG. 3 is a schematic drawing of a generated spatial location map in the exemplary embodiment. FIG. 4 is a schematic drawing of a blind spot object alert determination device in use in traffic at an intersection in an exemplary embodiment. FIG. 5 is a schematic drawing of a blind spot object alert determination device in use in traffic at an intersection in another exemplary embodiment. FIG. 6 is a schematic drawing of a blind spot object alert determination device in use in traffic at an intersection in another exemplary embodiment. FIG. 7 is a schematic drawing of a blind spot object alert determination device in use in traffic at an intersection in another exemplary embodiment. FIG. 8 is a schematic block diagram of a blind spot object alert determination device in another exemplary embodiment. FIG. 9 is a schematic drawing of a system for blind spot object alert determination in an exemplary embodiment. FIG. 10 is a schematic flowchart illustrating a computer-implemented method of determining a blind spot object alert in an exemplary embodiment. FIG. 11A is a schematic illustration to show differences in outputs between a conventional camera and an event camera. FIG. 11B is another schematic illustration to show differences in outputs between a conventional camera and an event camera. FIG. 12 is a schematic illustration of an ensemble approach in one example. DETAILED DESCRIPTION Exemplary embodiments described herein may provide a device that may assist in detecting a blind spot object or an object in a blind spot, e.g. with respect to a vehicle, and the device may assist in determining whether to issue / generate an alert for the blind spot object. Thus, a blind spot object alert determination device for use with a vehicle may be provided. A system for blind spot object alert determination and a computer-implemented method of determining a blind spot object alert may also be provided. FIG. 1 is a schematic block diagram of a blind spot object alert determination device in an exemplary embodiment. In the exemplary embodiment, the blind spot object alert determination device 100 for use with a vehicle comprises an event camera 102. In the exemplary embodiment, the blind spot object alert determination device 100 further comprises a blind spot region detection module 104, the blind spot region detection module 104 arranged to receive event camera data obtained by the event camera 102 and further arranged to process at least the event camera data to identify one or more blind spots around the vehicle. In the exemplary embodiment, the blind spot object alert determination device 100 further comprises a collaborative perception module 106 coupled to the blind spot region detection module 104, the collaborative perception module 106 arranged to receive identification data of the one or more blind spots from the blind spot region detection module 104. In the exemplary embodiment, the blind spot object alert determination device 100 further comprises a transceiver 108 coupled to the collaborative perception module 106, the transceiver 108 arranged to request and receive data from at least one external agent, the request being based on the identification data and the data from the at least one external agent comprises perception data of the at least one external agent. In the exemplary embodiment, the blind spot object alert determination device 100 further comprises an object detection module 110 coupled to the collaborative perception module 106, the object detection module 110 arranged to perform object detection at the identified one or more blind spots around the vehicle based on processed received data from the at least one external agent processed at the collaborative perception module 106. In the exemplary embodiment, the blind spot object alert determination device 100 further comprises a trajectory estimation module 112 coupled to the object detection module 110, the trajectory estimation module 112 arranged to estimate a trajectory of at least one object detected at the identified one or more blind spots around the vehicle as a trajectory estimate of the at least one object, the trajectory estimate being based on the object detection and based on the received data from the at least one external agent. In the exemplary embodiment, the blind spot object alert determination device 100 further comprises an alert determination module 114 coupled to the trajectory estimation module 112, the alert determination module 114 arranged to process the trajectory estimate transmitted from the trajectory estimation module 112 and to determine whether to output an alert to an operator of the vehicle. In the exemplary embodiment, the blind spot object alert determination device 100 may further comprise a display unit (not shown). The display unit may be configured to output an alert, or an alert output, to the operator of the vehicle. In addition or alternatively, the display unit may be configured to display a view around the vehicle, e.g. a 360-degree view around the vehicle. Furthermore, the blind spot object alert determination device 100 may be configured to output an alert, or an alert output, to the operator of the vehicle using other forms, e.g. using an audio alert via one or more audio output components, using a visual lighting alert via one or more operable lighting components etc. FIG. 2 is a schematic block diagram illustrating a blind spot object alert determination device in another exemplary embodiment. In the exemplary embodiment, the blind spot object alert determination device is substantially similar to the blind spot object alert determination device 100 described with reference to FIG. 1. In the exemplary embodiment, an event camera 202 (compare event camera 102 of FIG. 1) of a blind spot object alert determination device may be mounted on a vehicle 204. In the exemplary embodiment, the blind spot object alert determination device is provided within / therein the vehicle 204. There may be at least one external agent 206 e.g. other infrastructure object such as a traffic light at an intersection 206a and / or one or more other vehicles / objects 206b, situated external to the vehicle 204 or the blind spot object alert determination device. In the exemplary embodiment, one or more other event cameras substantially similar to the event camera 202 may be mounted on or affixed to at least one external agent 206. In the exemplary embodiment, the event camera 202 may be arranged to begin generating event camera data and to send the event camera data to the blind spot region detection module 208. For example, the generation of event camera data for analysis or processing may be triggered based on a number of different ways such as, but not limited to, triggering via activation of a button coupled to the blind spot object alert determination device, e.g. the button being provided on a dashboard of a vehicle and / or triggering when a vehicle reaches a predetermined speed and / or triggering when a vehicle is started up. When the generation of event camera data is triggered or turned on, the event camera 202 may be arranged to continuously capture data from one or more areas around a vehicle, such areas including one or more blind spot regions. The event camera 202 may be arranged to transmit event camera data to one or more other modules, e.g. for detecting presence of any object in one or more blind spot regions and to determine whether to output an alert to an operator of a vehicle accordingly. In the exemplary embodiment, a blind spot region detection module 208 (compare blind spot region detection module 104 of FIG. 1) may receive a signal e.g. from the event camera 202 mounted on the vehicle 204. See arrow 210. In the exemplary embodiment, the blind spot region detection module 208 may receive event camera data obtained by the event camera 202. The acquired signal may comprise event camera data and other data such as vehicle information of the vehicle 204 such as the motion state (position, velocity etc.), steering direction data and braking data etc. The blind spot region detection module 208 may process the received signal 210 comprising the event camera data from the event camera 202 to identify one or more blind spots or blind spot regions around the vehicle 204 e.g. using a blind spot detection algorithm. In the exemplary embodiment, a collaborative perception module 212 (compare collaborative perception module 106 of FIG. 1) may receive a signal from the blind spot region detection module 208 comprising identification data of one or more blind spots. See arrow 213. The collaborative perception module 212 may also request and receive a signal from the at least one external agent 206. See arrow 216. For example, the collaborative perception module 212 may selectively transmit a request for data to the at least one external agent 206 based on the identification data of one or more blind spots and / or the collaborative perception module 212 may selectively receive data from the at least one external agent 206 based on the identification data of one or more blind spots. Such data comprises perception or perception range data from the at least one external agent 206 and can include vehicle motion or motion state information of the at least one external agent 206. Vehicle motion information may include vehicle heading, speed etc. The data from the at least one external agent 206 may be obtained by an event camera of the at least one external agent 206. The data from the at least one external agent 206 may be requested for and received via a transceiver (not shown) coupled to the collaborative perception module 212. In the exemplary embodiment, the blind spot object alert determination device may compensate view information by receiving data / information from the at least one external agent 206 through different kinds of communication techniques such as vehicle-to-everything (V2X) communication, vehicle-to-cloud (V2C) communication etc. In the exemplary embodiment, e.g. using a V2X approach, perception range information / data for blind spot and view compensation (for occluded regions / objects) may be received at the blind spot object alert determination device. The inventors recognize that a plurality of challenges due to limited bandwidth, such as network congestion, latency, and scaling to more external agents may be encountered. Such challenges may be addressed and usefully mitigated by optimizing the communication to receive perception data based on a spatial location map of the exemplary embodiment. FIG. 3 is a schematic drawing of a generated spatial location map in the exemplary embodiment. In the exemplary embodiment, for generating the spatial location map 300, a vehicle 302 requests data of blind spot regions 306 from at least one external agent 304 e.g. other vehicles / objects and / or infrastructure objects. See e.g. traffic light at an intersection having an event camera 304a and other vehicles 304b, 304c, 304d. The possible blind spots may also be shown schematically as the shaded regions. Each external agent e.g. 304a to 304d may have a spatial map which is used to map their own perception data or perception range data to the spatial locations in the spatial location map 300 via the grids as shown in FIG. 3. In the exemplary embodiment, for example, each external agent 304a to 304d may detect objects such as other vehicles in their own respective vicinities. For example, each external agent 304a to 304d may perform techniques such as image processing and / or use one or more sensors (e.g. an active sensor or an event camera) to collect data and form its own spatial map. Thus, each external agent 304a to 304d may possess perception data of itself, and the perception data may comprise perception range data indicating one or more objects detected in its own vicinity and the relevant spatial range / distance of such one or more objects. As such, such information may be used to map to the spatial locations in the spatial location map 300 via the grids as shown in FIG. 3. In FIG. 3, spatial location is divided into multiple grids as denoted by grid number Gxx, where xx is a numeral. See e.g. left top most grid G11. In the exemplary embodiment, data is requested by the vehicle 302 for the grids falling in the one or more blind spots or blind spot regions 306 as well as the grids for view compensation for occluded regions. For example, the identification of the one or more blind spots or blind spot regions 306 is processed by the collaborative perception module 212 (FIG. 2) and the data request is transmitted via from the vehicle 302 using a transceiver (compare transceiver 108 of FIG. 1). In FIG. 3, data is requested by the vehicle 302 from the traffic light at an intersection 304a and other vehicles 304b, 304c, 304d about the grids G12, G13, G22 and G23. Thus, there is spatial location selection by the collaborative perception module 212 (FIG. 2) for optimized perception data sharing. In the exemplary embodiment, the external agents 304a to 304d in the grids G12, G13, G22 and G23 or an object being able to provide information about the grids G12, G13, G22 and G23 (see external agent 304a) broadcast or transmit data corresponding to those grids for which data has been requested by the vehicle 302. See e.g. double-headed dotted arrows 308 to / from the external agents 304a to 304d, indicating e.g. V2X communications. Other external agents such as other vehicles 304e and 304f do not broadcast or transmit data corresponding to those grids to the vehicle 302. In the exemplary embodiment, this selective data transmission arrangement and / or selective data receiving arrangement instructed by the collaborative perception module 212 (FIG. 2), based on identification of the one or more blind spots or blind spot regions 306, may usefully reduce the bandwidth required in V2X communication and the processing time required to process the transmitted data, thus optimising perception data sharing. In the exemplary embodiment based on the generated spatial location map 300, the collaborative perception module 212 (FIG. 2) may generate a view around a vehicle (e.g. a 360-degree bird’s eye view) by compensating for the occluded regions as well through collaborative perception. Thus, based on the data received from the at least one external agent 304 e.g. data received via V2X communication and the identification data of the one or more blind spots around the vehicle 302 received from the blind spot region detection module 208 (FIG. 2), the collaborative perception module 212 (FIG. 2) may be capable of generating a view around the vehicle 204 (e.g. a 360-degree view). In the exemplary embodiment, the external agent data 308 may comprise behavioural data transmitted from the at least one external agent 304, e.g. behavioural data transmitted from the external agents 304a to 304d. Such behavioural data may comprise information like pedestrian / driver intention and behaviour determined by intelligence of the at least one external agent 304, e.g. the external agents 304a to 304d such as a smart vehicle and / or infrastructure. This information could be, for example, information informing of a distracted pedestrian / driver, drunk driving, drowsiness, and fatigue. Returning to FIG. 2, in the exemplary embodiment, an object detection module 214 (compare object detection module 110 of FIG. 1) may receive data from the collaborative perception module 212 (see arrow 218) to perform object detection and localisation in the one or more blind spots around the vehicle 204. For example, the object detection module 214 may be arranged to perform object detection at the identified one or more blind spots around the vehicle 204 based on processed received data from the at least one external agent 206 processed at the collaborative perception module 212. In the exemplary embodiment, a trajectory estimation module 220 (compare trajectory estimation module 112 of FIG. 1) may receive an output from the object detection module 214. See arrow 222. For example, the trajectory estimation module 220 (compare trajectory estimation module 112 of FIG. 1) may receive indication and / or information of one or more objects detected by the object detection module 214 in the one or more blind spots around the vehicle 204. In the exemplary embodiment, if there is no object detected in the one or more blind spots around the vehicle 204 by the object detection module 214, event camera data is still received by the blind spot region detection module 104 to continue the processing of the event camera data to allow the object detection module 214 to continue performing object detection and localisation in the one or more blind spots around the vehicle 204. The trajectory estimation module 220 may also receive a signal from the at least one external agent 206. See arrow 224. The trajectory estimation module 220 may estimate a trajectory of at least one object detected at the identified one or more blind spots around the vehicle 204, based on the output from the object detection module 214, as a trajectory estimate of the at least one object. In the exemplary embodiment, the signal from the at least one external agent 206 may also comprise information or data related to the intention and behaviour of pedestrians and / or the intention and behaviour of drivers determined from intelligence information of smart vehicles and / or smart infrastructure objects. The information or data may be related to, for example, distracted pedestrians and / or drivers, drunk driving, drowsiness, and fatigue. Compare behavourial data of external agent data 308 of FIG. 3. As an example, such behavourial data may be obtained by e.g. image detection techniques, machine learning models, alert level determination models etc. provided in the one or more external agents that may study / analyse objects within or outside the one or more external agents. For example, a distracted pedestrian may be detected by an external agent, the pedestrian external to the external agent. For example, a driver with low alertness may be detected within an external agent, e.g. by image detection of facial expressions. For example, an external agent may derive such behavourial data information based on its own perception stack which may have e.g. a facial expression and gaze estimation module. In the exemplary embodiment, using intelligence information of smart vehicles and / or smart infrastructure objects may usefully improve trajectory estimation and allow the blind spot object alert determination device of the exemplary embodiment to be more robust, safe and reliable. In the exemplary embodiment, the trajectory estimation module 220 may compute an uncertainty score of a predicted potential / future trajectory of an object detected, besides computing the future trajectory. The uncertainty scores are estimated for the predicted future trajectories. For each predicted future trajectory of an object, the output of the trajectory estimation module 220 may comprise a future position and velocity of the object, and an attached / associated uncertainty score, over a time T. For example, T may be about 5 to 10 seconds for a predicted future trajectory of a currently moving object. In the exemplary embodiment, the trajectory estimation module 220 may compute aleatoric uncertainty (data uncertainty) and epistemic uncertainty (model uncertainty). Computing data uncertainty can capture or indicate uncertainty due to, for example, noise in the environment and / or noise in the sensors (e.g. the event camera 202). Computing epistemic uncertainty may usefully capture or indicate uncertainty due to limited knowledge of a trajectory estimation model that may be used in the exemplary embodiment. In the exemplary embodiment, the trajectory estimation module 220 may be implemented using a machine learning model. For example, various past behavourial data together with output(s) obtained from the object detection module may be used for training models. As an example, for computing data uncertainty (or aleatoric uncertainty), the trajectory estimation model may be modelled to predict an output trajectory as a distribution, for example but not limited to, a Gaussian distribution. For example, the mean of the distribution may be taken as the trajectory output and the variance may be taken as the data uncertainty. As an example, for computing model uncertainty (or epistemic uncertainty), the parameters of the model may be modelled as a distribution and sampling is performed from such a distribution to create N models. For example, such parameters may be the weights for a deep neural network. For example, one way to implement this is via using a model ensemble that has N different models. Thus, each of the N different models may output a mean and a variance, where the mean may be taken as the trajectory predicted and the variance may be taken as the data uncertainty. Thereafter, the variance among these N predicted means (or predicted trajectories) may be taken as the model uncertainty. For example, model uncertainty = variance ([mean 1,......mean N]). In a case of the above-discussed method of ensemble, there may be obtained multiple variances associated with the predicted trajectory output (mean) and thus, the final data uncertainty may be taken as the mean of these variances. For example, data uncertainty = mean ([var 1,......var N]). FIG. 12 is a schematic illustration of an ensemble approach in one example. The ensemble approach is as described above and being illustrated with an exemplary N = 3 models (M1, M2, M3). In the illustration 1200, three models M1 1202, M2 1204 and M3 1206 are used. Each M1 1202, M2 1204 and M3 1206 provides a set e.g. 1208 of a mean and a variance, and an output trajectory e.g. 1210. For example, there is shown meanl, var1; mean2, var2; and mean3, var3 respectively to the models M1 1202, M2 1204 and M3 1206. The mean of a model output from each model M1 1202, M2 1204 and M3 1206 provides an output trajectory e.g. 1210 from each model M1 1202, M2 1204 and M3 1206. For example, there is shown Trajectory 1; Trajectory 2; and Trajectory 3 respectively to the models M1 1202, M2 1204 and M3 1206. The respective variances var1, var2, var 3 may provide the associated uncertainty for each model. In the illustration, the trajectory estimation model’s data uncertainty may be taken as the mean of the respective variances of the models M1 1202, M2 1204 and M3 1206. For example, data uncertainty = mean ([var1, var2, var3]). The trajectory estimation model’s model uncertainty may be taken as the variance of the respective mean of the models M1 1202, M2 1204 and M3 1206. For example, model uncertainty = variance ([meanl, mean2, mean3]). The mean 1212 of the respective three output trajectories from each model M1 1202, M2 1204 and M3 1206 may be taken as the predicted trajectory 1214. In the example, the respective variances of the models M1 1202, M2 1204 and M3 1206 may be taken to compute the associated uncertainty score 1216 of the predicted trajectory 1214. In the exemplary embodiment, for example, the associated / attached uncertainty score 1216 of the predicted trajectory 1214 may be a function of the computed data uncertainty and the computed model uncertainty. As an example, the function may be an addition of the computed data uncertainty to the computed model uncertainty. In some exemplary embodiments, the computed data uncertainty may be used as the associated / attached uncertainty score 1216 of the predicted trajectory 1214. In some other exemplary embodiments, the computed model uncertainty may be used as the associated / attached uncertainty score 1216 of the predicted trajectory 1214. It will be appreciated that the above is an exemplary method and there may be several other ways to estimate uncertainty scores. In the exemplary embodiment, an uncertainty or uncertainty score which is measured via a model's variance is a relative value. A threshold value for low and high uncertainty may be decided empirically based on a numerical value predicted by the trajectory estimation model after experimentations. As an example, a threshold of + / - 2 degree of critical variance / deviation may be used. Compare uncertainty score 1216. For example, if there is a predicted trajectory and the variance / deviation is less than + / - 2 degree, it may be indicated that the uncertainty score is low. For example, if there is a predicted trajectory and the variance / deviation is greater than +1- 2 degree, it may be indicated that the uncertainty score is high. In the exemplary embodiment, the trajectory estimation module 220 may thus also estimate the trajectory of the at least one object detected at the identified one or more blind spots around the vehicle 204 further based on the behavioural data transmitted from the at least one external agent 206, the further estimate including an uncertainty score attached to the trajectory estimate. The uncertainty score computed by the trajectory estimation module 220 provides an indication on the uncertainty of the predicted trajectory of the at least one object detected at the identified one or more blind spots around the vehicle 204. Computing the uncertainty of the predicted trajectory may usefully ensure better safety for the operator of the vehicle 204. In the exemplary embodiment, trajectory estimation can be used to usefully reduce false alarms sent to an operator of the vehicle 204 and alert the operator of the vehicle 204 when there may be a potential collision based on the estimated trajectories of the vehicle 204 and the at least one object detected at the identified one or more blind spots around the vehicle 204. In the exemplary embodiment, based on the trajectory estimate of the at least one object detected at the identified one or more blind spots around the vehicle 204, an alert determination module 226 (compare alert determination module 114 of FIG. 1) may determine whether to output an alert to the operator of the vehicle 204. In the exemplary embodiment, the alert determination module 226 may determine whether to output an alert based on whether the trajectory estimate of the at least one object detected at the identified one or more blind spots around the vehicle 204 is predicted to intersect with an estimated trajectory of the vehicle 204 and e.g. based on the uncertainty score attached to the trajectory estimate of the at least one object detected at the identified one or more blind spots around the vehicle 204. As an example, an alert output 228 is schematically shown as an output to an operator of the vehicle 204. In the exemplary embodiment, if the trajectory estimate of the at least one object detected at the identified one or more blind spots around the vehicle 204 is predicted to intersect with the estimated trajectory of the vehicle 204 and the uncertainty score attached to the trajectory estimate of the at least one object is low (i.e. certainty is high), the alert determination module 226 may determine and output an alert to the operator of the vehicle 204. In the exemplary embodiment, if the trajectory estimate of the at least one object detected at the identified one or more blind spots around the vehicle 204 is predicted to not intersect with the estimated trajectory of the vehicle 204 and the uncertainty score attached to the trajectory estimate of the at least one object is low (i.e. certainty is high), the alert determination module 226 may determine to not output an alert to the operator of the vehicle 204. In the exemplary embodiment, if the trajectory estimate of the at least one object detected at the identified one or more blind spots around the vehicle 204 is predicted to intersect with the estimated trajectory of the vehicle 204 and the uncertainty score attached to the trajectory estimate of the at least one object is high (i.e. certainty is low), the alert determination module 226 may determine that the trajectory estimate of the at least one object may not be reliable (e.g. due to possible behaviour observed at the at least one object) but the at least one object is present e.g. in a blind spot. The alert determination module 226 may determine to still output an alert to the operator of the vehicle 204. In the exemplary embodiment, if the trajectory estimate of the at least one object detected at the identified one or more blind spots around the vehicle 204 is predicted to not intersect with the estimated trajectory of the vehicle and the uncertainty score attached to the trajectory estimate of the at least one object is high (i.e. certainty is low), the alert determination module 226 may determine that although the estimated trajectories do not intersect, but since the uncertainty score is high, e.g. indicating that the trajectory estimate of the at least one object may not be reliable (e.g. due to possible behaviour observed at the at least one object), to still output an alert to the operator of the vehicle 204. In the exemplary embodiment, the blind spot object alert determination device comprises usage of an event camera for blind spot monitoring and assistance. The event camera may also be used at an infrastructure object for traffic monitoring. The blind spot object alert determination device of the exemplary embodiment may use collaborative perception based on the event cameras and V2X communication may be usefully optimised to receive perception data based on a spatial location map. The blind spot object alert determination device of the exemplary embodiment may also utilise the intelligence of smart vehicles and smart infrastructure objects for blind spot monitoring to obtain information such as the intention and behaviour of pedestrians and / or drivers. In addition, for even better safety, besides computing the future trajectories of objects detected at identified one or more blind spots around a vehicle or the trajectory estimates, uncertainties of such trajectories may also be computed by estimating aleatoric uncertainty (data uncertainty) and / or epistemic uncertainty (model uncertainty). FIG. 4 is a schematic drawing of a blind spot object alert determination device in use in traffic at an intersection in an exemplary embodiment. In the exemplary embodiment, a vehicle 402 approaching an intersection is provided with a blind spot object alert determination device substantially similar to the blind spot object alert determination device described with reference to FIG. 1 and / or FIG. 2. The vehicle 402 is provided with an event camera 404 mounted thereon. In the exemplary embodiment, the event camera 404 may have a relatively high temporal resolution, with microsecond-level time precision. Thus, fast moving objects may be detected by the event camera 404 and without motion blur. These may be in contrast to RGB cameras that may introduce motion blur and may not capture fast-moving objects effectively. In the exemplary embodiment, a motorcycle 406 (an example of an external agent) moves at a fast speed from time t1 to time t2 to time t3 in one of the blind spots 408 of the vehicle 402, and which may be detected by the event camera 404. In the exemplary embodiment, when the blind spot object alert determination device provided with the vehicle 402 is used, the operator of the vehicle 404 may be usefully alerted to the fast-moving motorcycle 406 in time to avoid a potential collision with the motorcycle 406 at the intersection if the vehicle 402 were to continue along the trajectory estimate 410 of the vehicle 402 and if the motorcycle 406 were to continue along the trajectory estimate 412 of the motorcycle 406. In such an exemplary embodiment, a trajectory estimate with or without an uncertainty score may be sufficient. FIG. 5 is a schematic drawing of a blind spot object alert determination device in use in traffic at an intersection in another exemplary embodiment. In the exemplary embodiment, a vehicle 502 approaching an intersection is provided with a blind spot object alert determination device substantially similar to the blind spot object alert determination device described with reference to FIG. 1 and / or FIG. 2. In the exemplary embodiment, when the blind spot object alert determination device provided with the vehicle 502 is in use, a blind spot region detection module (compare blind spot region detection module 104 of FIG. 1) receives event camera data from an event camera (compare event camera 102 of FIG. 1) and processes at least the event camera data to identify one or more blind spots around the vehicle 502. In the exemplary embodiment, the blind spot region detection module identifies blind spots 504. A collaborative perception module (compare collaborative perception module 106 of FIG. 1) receives identification data of the one or more blind spots i.e. blind spots 504 from the blind spot region detection module. A transceiver (compare transceiver 108 of FIG. 1) requests and receives data from at least one external agent (e.g. a second vehicle 506 and / or a third vehicle 508), the request being based on the identification data of the one or more blind spots i.e. blind spots 504 and the data from the at least one external agent comprising perception data of the at least one external agent. In the exemplary embodiment, an object detection module (compare e.g. object detection module 110 of FIG. 1) performs object detection at the blind spots 504 around the vehicle 502. In the exemplary embodiment, the object detection module determines that there are two objects in the blind spots 504, i.e. the second vehicle 506 and the third vehicle 508. In the exemplary embodiment, a trajectory estimation module (compare trajectory estimation module 112 of FIG. 1) estimates a trajectory of the second vehicle 506 and the third vehicle 508 respectively and generates respective trajectory estimates. In FIG. 5, the trajectory estimate 510 of the second vehicle 506 is shown as a path straight in front of the second vehicle 506 and the trajectory estimate 512 of the third vehicle 508 is shown as a path straight in front of the third vehicle 508. In the exemplary embodiment, the data from the at least one external agent further comprises behavioural data transmitted from the at least one external agent. The trajectory estimation module further computes and attaches an uncertainty score to each of the trajectory estimates 510 and 512. In the exemplary embodiment, the level / amount / score of uncertainty is represented by a dotted box where the width of the box is directly proportional to the uncertainty score. A high uncertainty score 514 computed for the trajectory estimate 510 of the second vehicle 506 is indicated with a larger width of the first generated box ahead of the second vehicle 506 and the uncertainty score 516 corresponding to a low uncertainty score for the trajectory estimate 512 of the third vehicle 508 is indicated with the narrower width of the second generated box ahead of the third vehicle 508. In the exemplary embodiment, the trajectory estimate 518 of the vehicle 502 is shown. The trajectory estimation module may additionally estimate the trajectory estimate 518 of the vehicle 502. In the exemplary embodiment, based on the trajectory estimate 510 of the second vehicle 506 and the trajectory estimate 518 of the vehicle 502, the trajectory estimation module may determine that the trajectory estimate 510 does not intersect with the trajectory estimate 518. However, as the uncertainty score computed for the trajectory estimate 510 of the second vehicle 506 is high (see uncertainty score 514), the alert determination module may determine to output an alert to the operator of the vehicle 502 to alert the operator to the second vehicle 506. In the exemplary embodiment, based on the trajectory estimate 512 of the third vehicle 508 and the trajectory estimate 518 of the vehicle 502, the trajectory estimation module may determine that the trajectory estimate 512 does not intersect with the trajectory estimate 518. Further, the uncertainty score computed for the trajectory estimate 512 of the third vehicle 508 is low (see uncertainty score 516). Based on these determinations, the alert determination module may determine to not output an alert to the operator of the vehicle 502 to alert the operator to the third vehicle 508. FIG. 6 is a schematic drawing of a blind spot object alert determination device in use in traffic at an intersection in another exemplary embodiment. In the exemplary embodiment, a vehicle 602 approaching an intersection is provided with a blind spot object alert determination device substantially similar to the blind spot object alert determination device described with reference to FIG. 1 and / or FIG. 2. In the exemplary embodiment, when the blind spot object alert determination device provided with the vehicle 602 is in use, a blind spot region detection module (compare blind spot region detection module 104 of FIG. 1) receives event camera data from an event camera (compare event camera 102 of FIG. 1) and processes at least the event camera data to identify one or more blind spots around the vehicle 602. In the exemplary embodiment, the blind spot region detection module identifies blind spots 604. A collaborative perception module (compare collaborative perception module 106 of FIG. 1) receives identification data of the one or more blind spots i.e. blind spots 604 from the blind spot region detection module. A transceiver (compare transceiver 108 of FIG. 1) requests and receives data from at least one external agent (e.g. a second vehicle 606), the request being based on the identification data of the one or more blind spots i.e. blind spots 604 and the data from the at least one external agent comprising perception data of the at least one external agent. In the exemplary embodiment, an object detection module (compare e.g. object detection module 110 of FIG. 1) performs object detection at the blind spots 604 around the vehicle 602. In the exemplary embodiment, the object detection module determines that there is one object in the blind spots 604, i.e. the second vehicle 606. In the exemplary embodiment, a trajectory estimation module (compare trajectory estimation module 112 of FIG. 1) estimates a trajectory of the second vehicle 606 and generates a trajectory estimate 608 of the second vehicle 606. In the exemplary embodiment, the data from the at least one external agent further comprises behavioural data transmitted from the at least one external agent. The trajectory estimation module further computes and attaches an uncertainty score to the trajectory estimate 608 of the second vehicle 606. In the exemplary embodiment, the level / amount / score of uncertainty is represented by a dotted box where the width of the box is directly proportional to the uncertainty score. A high uncertainty score 610 computed for the trajectory estimate 608 of the second vehicle 606 is indicated with the relatively large width of the first generated box ahead of the second vehicle 606. In the exemplary embodiment, the trajectory estimate 612 of the vehicle 602 is shown. The trajectory estimation module may additionally estimate the trajectory estimate 612 of the vehicle 602. In the exemplary embodiment, based on the trajectory estimate 608 of the second vehicle 606 and the trajectory estimate 612 of the vehicle 602, the trajectory estimation module may determine that the trajectory estimate 608 of the second vehicle 606 intersects with the trajectory estimate 612 of the vehicle 602. However, the uncertainty score 610 computed for the trajectory estimate 608 of the second vehicle 606 is high. See the width for the uncertainty score 610. Based on the above determinations, the trajectory estimation module may determine to output an alert to the operator of the vehicle 602 to alert the operator to the second vehicle 606 despite the high uncertainty score on the basis that the trajectory estimates 608 and 612 intersect. FIG. 7 is a schematic drawing of a blind spot object alert determination device in use in traffic at an intersection in another exemplary embodiment. In the exemplary embodiment, a vehicle 702 approaching an intersection is provided with a blind spot object alert determination device substantially similar to the blind spot object alert determination device described with reference to FIG. 1 and / or FIG. 2. In the exemplary embodiment, when the blind spot object alert determination device provided with the vehicle 702 is in use, a blind spot region detection module (compare blind spot region detection module 104 of FIG. 1) receives event camera data from an event camera (compare event camera 102 of FIG. 1) and processes at least the event camera data to identify one or more blind spots around the vehicle 702. In the exemplary embodiment, the blind spot region detection module identifies blind spots 704. A collaborative perception module (compare collaborative perception module 106 of FIG. 1) receives identification data of the one or more blind spots i.e. blind spots 704 from the blind spot region detection module. A transceiver (compare transceiver 108 of FIG. 1) requests and receives data from at least one external agent (e.g. a second vehicle 706), the request being based on the identification data of the one or more blind spots i.e. blind spots 704 and the data from the at least one external agent comprising perception data of the at least one external agent. In the exemplary embodiment, an object detection module (compare e.g. object detection module 110 of FIG. 1) performs object detection at the blind spots 704 around the vehicle 702. In the exemplary embodiment, the object detection module determines that there is one object in the blind spots 704, i.e. the second vehicle 706. In the exemplary embodiment, a trajectory estimation module (compare trajectory estimation module 112 of FIG. 1) estimates a trajectory of the second vehicle 706 and generates a trajectory estimate 708 of the second vehicle 706. In the exemplary embodiment, the data from the at least one external agent further comprises behavioural data transmitted from the at least one external agent. The trajectory estimation module further computes and attaches an uncertainty score to the trajectory estimate 708 of the second vehicle 706. In the exemplary embodiment, the level / amount / score of uncertainty is represented by a dotted box where the width of the box is directly proportional to the uncertainty score. A low uncertainty score 710 computed for the trajectory estimate 708 of the second vehicle 706 is indicated with the relatively narrow width of the first generated box ahead of the second vehicle 706. In the exemplary embodiment, the trajectory estimate 712 of the vehicle 702 is shown. The trajectory estimation module may additionally estimate the trajectory estimate 712 of the vehicle 702. In the exemplary embodiment, based on the trajectory estimate 708 of the second vehicle 706 and the trajectory estimate 712 of the vehicle 702, the trajectory estimation module may determine that the trajectory estimate 708 of the second vehicle 706 intersects with the trajectory estimate 712 of the vehicle 702. Further, the uncertainty score 710 computed for the trajectory estimate 708 of the second vehicle 706 is low. See the width for the uncertainty score 710. Based on the above determinations, the trajectory estimation module may determine to output an alert to the operator of the vehicle 702 to alert the operator to the second vehicle 706. In the described exemplary embodiments, there are a number of modules described in detail. It will be appreciated that one or more of the modules may be comprised in or with a processing unit or module. FIG. 8 is a schematic block diagram of a blind spot object alert determination device in another exemplary embodiment. In the exemplary embodiment, the blind spot object alert determination device is substantially identical to the blind spot object alert determination device described with reference to FIG. 1. The blind spot object alert determination device 800 comprises an event camera 802 (compare event camera 102 of FIG. 1), a blind spot region detection module 804 (compare blind spot region detection module 104 of FIG. 1), collaborative perception module 806 (compare collaborative perception module 106 of FIG. 1), a transceiver 808 (compare transceiver 108 of FIG. 1), an object detection module 810 (compare object detection module 110 of FIG. 1), a trajectory estimation module 812 (compare trajectory estimation module 112 of FIG. 1) and an alert determination module 814 (compare alert determination module 114 of FIG. 1). In the exemplary embodiment, the blind spot object alert determination device 800 comprises a processing unit 816, the processing unit 816 comprising one or more of the blind spot region detection module 804, the collaborative perception module 806, the object detection module 810, the trajectory estimation module 812 and the alert determination module 814. For ease of illustration, it is shown schematically in FIG. 8 that the processing unit 816 comprises all of the above modules 804, 806, 810, 812, 814 but it is appreciated that FIG. 8 is intended to illustrate that the processing unit 816 comprises one or more of the above modules 804, 806, 810, 812, 814. FIG. 9 is a schematic drawing of a system for blind spot object alert determination in an exemplary embodiment. In the exemplary embodiment, the system 900 for blind spot object alert determination comprises a vehicle 902 and at least one external agent 904 external to the vehicle 902, the external agent 904 comprising at least a sensor 906 arranged to obtain data to detect an external agent object in a vicinity of the external agent 904. In the exemplary embodiment, the external agent object in a vicinity of the external agent 904 refers to an object (e.g. a vehicle or other objects) that the external agent 904 may detect with its own perception data or perception range data. In the exemplary embodiment, the system 900 further comprises the vehicle 902 having provided therein a blind spot object alert determination device 908. In the exemplary embodiment, the blind spot object alert determination device 908 is substantially identical to the blind spot object alert determination device 100 described with reference to FIG. 1. For ease of reference, like numerals are used for exemplary implementations of similar components of the device as described in FIG. 1. For ease of description, the like components are not reproduced in FIG. 9. In the exemplary embodiment, the device 908 of the vehicle 902 is arranged to communicate with the at least one external agent 904 using one or more signals 910. In the exemplary embodiment, the device 908 comprises an event camera 102; a blind spot region detection module 104, the blind spot region detection module 104 arranged to receive event camera data obtained by the event camera 102 and further arranged to process at least the event camera data to identify one or more blind spots around the vehicle 902; a collaborative perception module 106 coupled to the blind spot region detection module104, the collaborative perception module 106 arranged to receive identification data of the one or more blind spots from the blind spot region detection module 104; a transceiver 108 coupled to the collaborative perception module 106; an object detection module 110 coupled to the collaborative perception module 106; a trajectory estimation module 112 coupled to the object detection module 110; and an alert determination module 114 coupled to the trajectory estimation module 112. In the exemplary embodiment, the transceiver 108 is arranged to request and receive the data from the at least one external agent 904 via the one or more signals 910, the request being based on the identification data and the data from the at least one external agent 904 comprises perception data of the at least one external agent 904; the object detection module 110 is arranged to perform object detection at the identified one or more blind spots around the vehicle 902 based on processed received data from the at least one external agent 904 processed at the collaborative perception module 106; the trajectory estimation module 112 is arranged to estimate a trajectory of at least one object detected at the identified one or more blind spots around the vehicle 902 as a trajectory estimate of the at least one object, the trajectory estimate being based on the object detection and based on the received data from the at least one external agent 904; and the alert determination module 114 is arranged to process the trajectory estimate transmitted from the trajectory estimation module 112 and to determine whether to output an alert to an operator of the vehicle 902. In the exemplary embodiment, the at least a sensor 906 of the external agent 904 comprises another event camera. In other exemplary embodiments, the sensor 906 may be another active sensor. In the exemplary embodiment, the at least one external agent 904 comprises an infrastructure object. The infrastructure object may be, for example, a traffic light at an intersection. In the exemplary embodiment, the data from the at least one external agent 904 further comprises behavioural data transmitted from the at least one external agent 904. In the exemplary embodiment, the trajectory estimation module 112 is arranged to further estimate the trajectory of the at least one object detected at the identified one or more blind spots around the vehicle 902 further based on the behavioural data transmitted from the at least one external agent 904, the further estimate including an uncertainty score attached to the trajectory estimate and wherein the alert determination module 114 is arranged to determine whether to output the alert to the operator of the vehicle 902 further based on the uncertainty score attached to the trajectory estimate. In the exemplary embodiment, the collaborative perception module 106 is capable of generating a view around the vehicle 902 based on the received data from the at least one external agent 904 and the identification data of the one or more blind spots around the vehicle 902. FIG. 10 is a schematic flowchart 1000 illustrating a computer-implemented method of determining a blind spot object alert in an exemplary embodiment. One or more steps of the method are computer-implemented. Alternatively, the method is a computer-implemented method. At step 1002, an event camera is provided in a vehicle. At step 1004, the event camera is used to obtain event camera data and a blind spot region detection module is used to process at least the event camera data to identify one or more blind spots around the vehicle. At step 1006, identification data of the one or more blind spots is received from the blind spot region detection module and a collaborative perception module is used to instruct a transceiver coupled to the collaborative perception module to request and receive data from at least one external agent, the request being based on the identification data and the data from the at least one external agent comprising perception data of the at least one external agent. At step 1008, received data from the at least one external agent is processed using the collaborative perception module. At step 1010, object detection is performed at the identified one or more blind spots around the vehicle using an object detection module based on the processed received data processed at the collaborative perception module. At step 1012, using a trajectory estimation module, a trajectory of at least one object detected at the identified one or more blind spots around the vehicle is estimated as a trajectory estimate of the at least one object, the trajectory estimate being based on the object detection and based on the received data from the at least one external agent. At step 1014, the trajectory estimate is processed using an alert determination module. At step 1016, a determination is made on whether to output an alert to an operator of the vehicle using the alert determination module. In the exemplary embodiment, the data from the at least one external agent further comprises behavioural data transmitted from the at least one external agent. In the exemplary embodiment, the step of estimating further comprises further estimating the trajectory of the at least one object detected at the identified one or more blind spots around the vehicle further based on the behavioural data transmitted from the at least one external agent, the further estimating including attaching an uncertainty score to the trajectory estimate; and the step of determining whether to output an alert to an operator of the vehicle using the alert determination module is further based on the uncertainty score attached to the trajectory estimate. In the exemplary embodiment, the method further comprises generating a view around the vehicle using the collaborative perception module based on the received data from the at least one external agent and the identification data of the one or more blind spots around the vehicle. In the described exemplary embodiments, there is provided a blind spot object alert determination device for uncertainty aware blind spot monitoring and assistance using an event camera. In exemplary embodiments, the blind spot object alert determination device may improve the efficiency of blind spot monitoring and assistance systems. In the described exemplary embodiments, the use of an event camera may provide a number of advantages. As an example, the event camera can detect objects in low light visibility and / or when there is overexposure due to the high dynamic range that the event camera can provide. The event camera can also detect fast-moving objects due to or owing to the high temporal resolution of the event camera data generated by the event camera. The event camera may be less susceptible to motion blur as compared to RGB cameras because the event camera is configured to capture changes in brightness rather than other cameras’ integration of light over time. The event camera in the exemplary embodiments is recognised to require less bandwidth which makes the event camera usefully suitable for applications that require fast response time such as for the V2X communications of some exemplary embodiments. The event camera may also provide low latency because the event camera is configured to generate event data in real-time. In the description herein, an event camera is an optical sensor that responds to brightness changes and can asynchronously output events in case of brightness variations at pixel level. FIG. 11A is a schematic illustration to show differences in outputs between a conventional camera and an event camera. Credits to the drawing are given to H. Rebecq et. al., High speed and high dynamic range video with an event camera, IEEE transactions on pattern analysis and machine intelligence, 2019. There is provided a rotating circle 1102 having disposed thereon a black disk 1104. The rotating circle 1102 rotates clockwise as seen in the illustration. A conventional camera output 1106 is based on capturing frames at a fixed rate. For example, refer to a first frame 1108 and a second frame 1110 and the respective positions of the black disk 1104 having been rotated through time. On the other hand, an event camera output 1112 is based on a transmission of the brightness changes continuously in the form of a spiral 1114 of events in space-time. FIG. 11B is another schematic illustration to show differences in outputs between a conventional camera and an event camera. Credits to the drawing are given to Gao et. al., An End-to-End Broad Learning System for Event-Based Object Classification, IEEE Access. PP., 2020. There is provided an object 1116. A conventional camera output 1118 is shown with a series of frames captured over time. An event camera output 1120 is shown. A plurality of events 1122 which are shown as scattered dots over time are captured by the event camera output 1120. The event camera output as an image 1124 from an event camera is synthesized after processing the events 1122. In the illustration, the term ‘event’ refers to an output spike, typically characterized by a specific spatial location (x, y), timestamp(t) and brightness change polarity (p). As such, the output from an event camera is a stream of asynchronous spikes, triggered by brightness changes sensed by individual pixels. In the described exemplary embodiments, communication in collaborative perception may be optimised. In exemplary embodiments, the collaborative perception module may be capable of generating an optimised collaborative perception of a view around a vehicle based on receiving perception data from at least one external agent and that is based on a selective spatial location map. In some described exemplary embodiments, the blind spot object alert determination device may be robust and accurate in that the device may be configured to receive information or behavourial data such as the intention and behaviour or pedestrians and / or drivers by utilising intelligence information obtained from smart vehicles and / or smart infrastructures. As some examples, the intelligence information may relate to distracted pedestrians and / or drivers, drunk driving, drowsiness and fatigue. In some described exemplary embodiments, the trajectory estimation module may be configured to estimate an uncertainty score (e.g. a total uncertainty score) by computing an aleatoric uncertainty score and an epistemic uncertainty score that may be directly predicted by a model at an inference time. In the described exemplary embodiments, the blind spot object alert determination device for uncertainty aware blind spot monitoring and assistance using an event camera may be used in vehicles (e.g. cars, trucks, tractors with trailers), using collaborative perception optimised through artificial intelligence (Al). In the described exemplary embodiments, the event camera may consume less power. The event camera may consume less power because the event camera operates by only producing event data when there is a change in a scene, in contrast to RGB cameras that continuously capture frames even when static. This may make the event camera suitable and / or useful for battery-powered and energy-efficient devices, and such as for the blind spot object alert determination device of the exemplary embodiments. The event camera may also have a high temporal resolution, with microsecond-level time precision. In exemplary embodiments, this allows the event camera to capture rapid motion and changes in a scene with accuracy, in contrast to RGB cameras that use fixed frame rates. The event camera may further have a high dynamic range, which may allow the event camera to deal with scenes with a wide range of lighting conditions, from bright to dark conditions, without having to resort to saturation or losing information. The event camera may thus be suitable and / or useful for applications in challenging lighting environments. The event camera may also provide low latency because the event camera may generate event data in real-time. This feature of the event camera may be significant for applications where quick responses are desired, such as in robotics and autonomous vehicles and such as in the blind spot object alert determination device of the exemplary embodiments. The event camera may further provide sparse data in the form of events, in contrast to RGB cameras that capture dense pixel data for every frame. The sparsity may usefully reduce the amount of data that needs to be processed and transmitted by the event camera, making the event camera suitable and / or useful for applications with limited computational resources or bandwidth constraints and such as for the blind spot object alert determination device of the exemplary embodiments. The event camera may additionally be less susceptible to motion blur because the event camera is configured to capture changes in brightness, in contrast to RGB cameras. This makes the event camera suitable and / or useful for tasks that may involve fast-moving objects or scenes and such as for the blind spot object alert determination device of the exemplary embodiments. The event camera may further be insensitive or at least less sensitive to global illumination changes, and therefore more robust to illumination changes as compared to RGB cameras. This makes the event camera suitable and / or useful for outdoor and dynamic lighting conditions and for the blind spot object alert determination device of the exemplary embodiments. In exemplary embodiments, the high temporal resolution, low latency, and the robustness to motion blur of the event camera may usefully provide significant advantages for the blind spot object alert determination device of the exemplary embodiments. For example, the above useful effects may enable the event camera to be better in applications utilising simultaneous localization and mapping (SLAM) applications e.g. the spatial mapping of exemplary embodiments, and robotics. Further, since the event camera is configured to capture changes in a scene, the event camera may generate less data as compared to other cameras. This may usefully reduce the storage requirements for the purposes of long-term data recording. In exemplary embodiments, event-driven processing owing to the event camera may enable the blind spot object alert determination device to be power-efficient, as the event camera may allow the blind spot object alert determination device to focus computational resources on relevant information when events occur. In the described exemplary embodiments, the use of the event camera may usefully mitigate issues related to low visibility, overexposure, detection in high-speed motion, motion blur or missing an object due to low frame rate, higher bandwidth requirement and high latency, and being unable to alert an operator on time in high-risk situations, which may provide less reaction time to the operator. In the described exemplary embodiments, the blind spot object alert determination device may usefully provide advantages over using one of or a combination of active sensors (such as a radar sensor, an ultrasonic sensor and / or a LiDAR sensor) and / or a vision-based sensor (such as an active camera) for blind spot monitoring and assistance. In the described exemplary embodiments, the blind spot object alert determination device may further usefully provide advantages over using smart infrastructures that utilise active sensors (such as a radar sensor, an ultrasonic sensor and / or a LiDAR sensor) and / or a vision-based sensor (such as an active camera). In the described exemplary embodiments, the blind spot object alert determination device may also usefully provide advantages with respect to using collaborative perception for blind spot monitoring and assistance e.g. mitigates issues associated with limited bandwidth, network congestion, latency and scaling to more external agents. In the described exemplary embodiments, the blind spot object alert determination device may usefully utilise intelligence information obtained from smart vehicles and / or smart infrastructures. Intelligence information may relate to e.g. distracted pedestrians and / or drivers, drunk driving, drowsiness, and fatigue In the described exemplary embodiments, the blind spot object alert determination device may be configured to usefully estimate the uncertainty of a trajectory estimate, thus improving the reliability of the device. In the described exemplary embodiments, the blind spot object alert determination device may be configured to usefully compute a more accurate representation of uncertainty with respect to a trajectory estimate by considering the aleatoric uncertainty (data uncertainty) and epistem ic uncertainty (model uncertainty). Various exemplary embodiments of the blind spot object alert determination device described herein may provide efficient blind spot monitoring and assistance and reduce false alarms that alert a driver erroneously. Various exemplary embodiments of the blind spot object alert determination device described herein may also increase safety for a driver by incorporating the computation of uncertainty scores of a trajectory estimate for an external agent and provide sufficient reaction time for a driver to react in high-risk scenarios. In the description herein, the terms "coupled" or "connected" as used are intended to cover both directly connected or connected through one or more intermediate means, unless otherwise stated. The use of “a”, “an” or “the” is intended to mean “one or more” unless it is described specifically to the contrary. The terms “configured to (perform a task / action)”, “configured for (performing a task / action)” and the like such as “arranged to (perform a task / action)” as used in this description include being programmable, programmed, connectable, wired or otherwise constructed to have the ability to perform the task / action when arranged or installed as described herein. The terms “configured to (perform a task / action)”, “configured for (performing a task / action)” and the like such as “arranged to (perform a task / action)” are intended to cover “when in use, the task / action is performed”, e.g. specifically to and / or specifically configured to and / or specifically arranged to and / or specifically adapted to do or perform a task / action. The term "and / or", e.g., "X and / or Y" is understood to mean either "X and Y" or "X or Y" and should be taken to provide explicit support for both meanings or for either meaning. The use of “or” is intended to mean an “inclusive or,” and not an “exclusive or” unless it is described specifically to the contrary. The terms "associated with", “related to” and the like used herein when referring to two elements refers to a broad relationship between the two elements. The relationship includes, but is not limited to, a physical, a chemical or a biological relationship. For example, when element A is associated with element B, elements A and B may be directly or indirectly attached to each other or element A may contain element B or vice versa. The terms “exemplary embodiment”, “example embodiment”, “exemplary implementation”, “exemplarily” and the like used herein are intended to indicate an example of matters described in the present disclosure. Such an example may relate to one or more features defined in the claims and is not necessarily intended to emphasise a best example or any essentialness of any features. The terms “estimate” and “predict” as used herein may be interchangeably used. The description herein may be, in certain portions, explicitly or implicitly described as algorithms and / or functional operations that operate on data within a computer memory or an electronic circuit. These algorithmic descriptions and / or functional operations are usually used by those skilled in the information / data processing arts for efficient description. An algorithm is generally relating to a self-consistent sequence of steps leading to a desired result. The algorithmic steps can include physical manipulations of physical quantities, such as electrical, magnetic or optical signals capable of being stored, transmitted, transferred, combined, compared, and otherwise manipulated. Further, unless specifically stated otherwise, and would ordinarily be apparent from the following, a person skilled in the art will appreciate that throughout the present specification, discussions utilizing terms such as “scanning”, “calculating”, “determining”, “replacing”, “generating”, “initializing”, “outputting”, and the like, refer to action and processes of an instructing processor / computer system, or similar electronic circuit / device / component, that manipulates / processes and transforms data represented as physical quantities within the described system into other data similarly represented as physical quantities within the system or other information storage, transmission or display devices etc. The description also discloses relevant device / apparatus for performing the steps of the described methods. Such apparatus may be specifically constructed for the purposes of the methods, or may comprise a general purpose computer / processor or other device selectively activated or reconfigured by a computer program stored in a storage member. The algorithms and displays described herein are not inherently related to any particular computer or other apparatus. It is understood that general purpose devices / machines may be used in accordance with the teachings herein. Alternatively, the construction of a specialized device / apparatus to perform the method steps may be desired. In addition, it is submitted that the description also implicitly covers a computer program, in that it would be clear that the steps of the methods described herein may be put into effect by computer code. It will be appreciated that a large variety of programming languages and coding can be used to implement the teachings of the description herein. Moreover, the computer program if applicable is not limited to any particular control flow and can use different control flows without departing from the scope of the invention. Furthermore, one or more of the steps of the computer program if applicable may be performed in parallel and / or sequentially. Such a computer program if applicable may be stored on any computer readable medium. The computer readable medium may include storage devices such as magnetic or optical disks, memory chips, or other storage devices suitable for interfacing with a suitable reader / general purpose computer. In such instances, the computer readable storage medium is non-transitory. Such storage medium also covers all computer-readable media e.g. medium that stores data only for short periods of time and / or only in the presence of power, such as register memory, processor cache and Random Access Memory (RAM) and the like. The computer readable medium may even include a wired medium such as exemplified in the Internet system, or wireless medium such as exemplified in Bluetooth technology. The computer readable medium may be, for example, cloud storage in the Internet or within an intranet. The computer program when loaded and executed on a suitable reader effectively results in an apparatus that can implement the steps of the described methods, e.g. in a physical embodiment. The computer readable medium is intended to be transferable and is reproducible in that the computer program if applicable is reproducible. The exemplary embodiments may also be implemented as hardware modules. A module is a functional hardware unit designed for use with other components or modules. For example, a module may be implemented using digital or discrete electronic components, or it can form a portion of an entire electronic circuit such as an Application Specific Integrated Circuit (ASIC). A person skilled in the art will understand that the exemplary embodiments can also be implemented as a combination of hardware and software modules. Additionally, when describing some embodiments, the disclosure may have disclosed a method and / or process as a particular sequence of steps. However, unless otherwise required, it will be appreciated that the method or process should not be limited to the particular sequence of steps disclosed. Other sequences of steps may be possible. The particular order of the steps disclosed herein should not be construed as undue limitations. Unless otherwise required, a method and / or process disclosed herein should not be limited to the steps being carried out in the order written. The sequence of steps may be varied and still remain within the scope of the disclosure. Further, in the description herein, the word “substantially” whenever used is understood to include, but not restricted to, "entirely" or “completely” and the like. In addition, terms such as "comprising", "comprise", and the like whenever used, are intended to be non-restricting descriptive language in that they broadly include elements / components recited after such terms, in addition to other components not explicitly recited. For an example, when “comprising” is used, reference to a “one” feature is also intended to be a reference to “at least one” of that feature. Terms such as “consisting”, “consist”, and the like, may, in the appropriate context, be considered as a subset of terms such as "comprising", "comprise", and the like. Therefore, in embodiments disclosed herein using the terms such as "comprising", "comprise", and the like, it will be appreciated that these embodiments provide teaching for corresponding embodiments using terms such as “consisting”, “consist”, and the like. Further, terms such as "about", "approximately" and the like whenever used, typically means a reasonable variation, for example a variation of + / - 5% of the disclosed value, or a variance of 4% of the disclosed value, or a variance of 3% of the disclosed value, a variance of 2% of the disclosed value or a variance of 1 % of the disclosed value. Furthermore, in the description herein, certain values may be disclosed in a range. The values showing the end points of a range are intended to illustrate a preferred range. Whenever a range has been described, it is intended that the range covers and teaches all possible sub-ranges as well as individual numerical values within that range. That is, the end points of a range should not be interpreted as inflexible limitations. For example, a description of a range of 1% to 5% is intended to have specifically disclosed sub-ranges 1 % to 2%, 1 % to 3%, 1 % to 4%, 2% to 3% etc., as well as individually, values within that range such as 1%, 2%, 3%, 4% and 5%. It is to be appreciated that the individual numerical values within the range also include integers, fractions and decimals. Furthermore, whenever a range has been described, it is also intended that the range covers and teaches values of up to 2 additional decimal places or significant figures (where appropriate) from the shown numerical end points. For example, a description of a range of 1 % to 5% is intended to have specifically disclosed the ranges 1.00% to 5.00% and also 1.0% to 5.0% and all their intermediate values (such as 1.01%, 1.02% ... 4.98%, 4.99%, 5.00% and 1.1%, 1.2% ... 4.8%, 4.9%, 5.0% etc.,) spanning the ranges. The intention of the above specific disclosure is applicable to any depth / breadth of a range. In the described exemplary embodiments, although it is described that the blind spot object alert determination device comprises an event camera and usage of the event camera, the exemplary embodiments are not limited as such. In other exemplary embodiments, to further extend the technical solution, the event camera may be combined or operation-fused with other sensors. Such sensors may be a vision-based sensor (such as an active camera) and / or an active sensor (such as a radar sensor and / or a LiDAR sensor). In the described exemplary embodiments, the blind spot object alert determination device may be deployed in autonomous vehicles, where instead of alerting the operator of a vehicle for action, the autonomous vehicle itself may define a course of action to be taken based on an alert determined to be output by the blind spot object alert determination device. In such an instance, an operator of the vehicle may be a processor of the vehicle. In the described exemplary embodiments, although the blind spot object alert determination device has been described to be used in the field of transportation, the exemplary embodiments are not limited as such and may be adapted for use in other fields such as wildlife monitoring, surveillance and security, robotics, aerospace and drones, activity and gesture recognition and sport analysis. In the described exemplary embodiments, while the blind spot object alert determination device has been described to comprise a single event camera, it will be appreciated that the blind spot object alert determination device is not limited as such. That is, depending on the application, the blind spot object alert determination device may be provided with more than one event camera. In the described exemplary embodiments, the at least one external agent and the one or more objects may be described as being in the form of an infrastructure object, another vehicle etc. However, it will be appreciated that the exemplary embodiments are not limited as such. For example, the at least one external agent and the one or more objects may include other matter such as a pedestrian detected as an object, a pedestrian being an external agent, e.g. using an imaging device carried by the pedestrian etc. In the described exemplary embodiments, the blind spot object alert determination device is described for use with a vehicle. The blind spot object alert determination device may be integrally provided in a vehicle or may be removably provided in a vehicle. It will be appreciated by a person skilled in the art that other variations and / or modifications may be made to the specific embodiments without departing from the scope of the claimed invention as broadly described. For example, in the description herein, features of different exemplary embodiments may be mixed, combined, interchanged, incorporated, adopted, modified, included etc. or the like across different exemplary embodiments. For example, exemplary embodiments are not necessarily mutually exclusive as some may be combined with one or more embodiments to form new exemplary embodiments. Furthermore, it will be appreciated that while the present disclosure provides embodiments having one or more of the features / characteristics discussed herein, one or more of these features / characteristics may also be disclaimed in other alternative embodiments and the present disclosure provides support for such disclaimers and these associated alternative embodiments. The present embodiments are, therefore, to be considered in all respects to be illustrative and not restrictive. REFERENCE SIGNS LIST 100 blind spot object alert determination device 102 event camera 104 blind spot region detection module 106 collaborative perception module 108 transceiver 110 object detection module 112 trajectory estimation module 114 alert determination module 202 event camera 204 vehicle 206 external agent 206a traffic light at an intersection 206b one or more other vehicles 208 blind spot region detection module 210 event camera signal 212 collaborative perception module 213 signal comprising identification data of one or more blind spots 214 object detection module 216 signal from at least one external agent 218 data from collaborative perception module 220 trajectory estimation module 222 data from object detection module 224 signal from at least one external agent received at trajectory estimation module 226 alert determination module 228 alert output 300 spatial location map 302 vehicle 304 external agent 304a traffic light at an intersection 304b, 304c, 304d other vehicles 306 blind spot region 308 external agent data 402 vehicle 404 event camera 406 motorcycle 408 blind spots 410 trajectory estimate of vehicle 402 412 trajectory estimate of motorcycle 406 502 vehicle 504 blind spots 506 second vehicle 508 third vehicle 510 trajectory estimate of second vehicle 506 512 trajectory estimate of third vehicle 508 514 uncertainty score attached to trajectory estimate of second vehicle 506 516 uncertainty score attached to trajectory estimate of third vehicle 508 518 trajectory estimate of vehicle 502 602 vehicle 604 blind spots 606 second vehicle 608 trajectory estimate of second vehicle 606 610 uncertainty score attached to trajectory estimate of second vehicle 606 612 trajectory estimate of vehicle 602 702 vehicle 704 blind spots 706 second vehicle 708 trajectory estimate of second vehicle 706 710 uncertainty score attached to trajectory estimate of second vehicle 706 712 trajectory estimate of vehicle 702 800 blind spot object alert determination device 802 event camera 804 blind spot region detection module 806 collaborative perception module 808 transceiver 810 object detection module 812 trajectory estimation module 814 alert determination module 816 processing unit 900 system for blind spot object alert determination 902 vehicle 904 external agent 906 sensor 908 blind spot object alert determination device 910 communication signal 1000 flowchart illustrating a computer-implemented method of determining a blind spot object alert 1002 step of providing an event camera 1004 step including identifying one or more blind spots around the vehicle 1006 step including requesting and receiving data from at least one external agent based on identification data and the data from the at least one external agent comprising perception data of the at least one external agent 1008 step of processing received data using collaborative perception module 1010 step including performing object detection at the identified one or more blind spots around the vehicle based on the processed received data processed at the collaborative perception module 1012 step including estimating a trajectory of at least one object detected at the identified one or more blind spots around the vehicle as a trajectory estimate of the at least one object, the trajectory estimate being based on the object detection and based on the received data from the at least one external agent 1014 step of processing the trajectory estimate using an alert determination module 1016 step of determining whether to output an alert to an operator of the vehicle using the alert determination module 1102 rotating circle 1104 black disk disposed on rotating circle 1102 1106 conventional camera output 1108 1110 1112 1114 5 1116 1118 1120 1122 1124 10 1200 1202 1204 1206 1208 15 1210 1212 1214 1216 first frame second frame event camera output spiral of events in space-time object conventional camera output event camera output plurality of events synthesized event camera image schematic illustration of an ensemble approach in one example machine learning model M1 machine learning model M2 machine learning model M3 a set of a mean and a variance of a machine learning model an output trajectory of a machine learning model mean of outputs of machine learning models predicted trajectory uncertainty score attached to predicted trajectory

Claims

1. A blind spot object alert determination device (100) for use with a vehicle, the device comprisingan event camera (102);characterized in that the device further comprisesa blind spot region detection module (104), the blind spot region detection module arranged to receive event camera data obtained by the event camera and further arranged to process at least the event camera data to identify one or more blind spots (306) around the vehicle;a collaborative perception module (106) coupled to the blind spot region detection module, the collaborative perception module arranged to receive identification data of the one or more blind spots from the blind spot region detection module;a transceiver (108) coupled to the collaborative perception module, the transceiver arranged to request and receive data from at least one external agent (304), the request being based on the identification data and the data from the at least one external agent comprises perception data of the at least one external agent;an object detection module (110) coupled to the collaborative perception module, the object detection module arranged to perform object detection at the identified one or more blind spots around the vehicle based on processed received data from the at least one external agent processed at the collaborative perception module;a trajectory estimation module (112) coupled to the object detection module, the trajectory estimation module arranged to estimate a trajectory of at least one object detected at the identified one or more blind spots around the vehicle as a trajectory estimate of the at least one object, the trajectory estimate being based on the object detection and based on the received data from the at least one external agent; andan alert determination module (114) coupled to the trajectory estimation module, the alert determination module arranged to process the trajectory estimatetransmitted from the trajectory estimation module and to determine whether to output an alert to an operator of the vehicle.

2. The device as claimed in claim 1, wherein the data from the at least one external agent further comprises behavioural data transmitted from the at least one external agent.

3. The device as claimed in claim 2, wherein the trajectory estimation module is arranged to further estimate the trajectory of the at least one object detected at the identified one or more blind spots around the vehicle further based on the behavioural data transmitted from the at least one external agent, the further estimate including an uncertainty score attached to the trajectory estimate and wherein the alert determination module is arranged to determine whether to output the alert to the operator of the vehicle further based on the uncertainty score attached to the trajectory estimate.

4. The device as claimed in any one of claims 1 to 3, further comprising a processing unit (816), the processing unit comprising one or more of the blind spot region detection module, the collaborative perception module, the object detection module, the trajectory estimation module and the alert determination module.

5. The device as claimed in any one of claims 1 to 4, wherein the collaborative perception module is capable of generating a view around the vehicle based on the received data from the at least one external agent and the identification data of the one or more blind spots around the vehicle.

6. A system (900) for blind spot object alert determination, the system comprisinga vehicle;at least one external agent (904) external to the vehicle, the external agent comprising at least a sensor (906) arranged to obtain data to detect an external agent object in a vicinity of the external agent;characterized in that the system further comprisesthe vehicle (902) having provided therein a blind spot object alert determination device (908), the device comprisingan event camera (102);a blind spot region detection module (104), the blind spot region detection module arranged to receive event camera data obtained by the event camera and further arranged to process at least the event camera data to identify one or more blind spots (306) around the vehicle;a collaborative perception module (106) coupled to the blind spot region detection module, the collaborative perception module arranged to receive identification data of the one or more blind spots from the blind spot region detection module;a transceiver (108) coupled to the collaborative perception module;an object detection module (110) coupled to the collaborative perception module;a trajectory estimation module (112) coupled to the object detection module; andan alert determination module (114) coupled to the trajectory estimation module;wherein the transceiver is arranged to request and receive the data from the at least one external agent, the request being based on the identification data and the data from the at least one external agent comprises perception data of the at least one external agent;wherein the object detection module is arranged to perform object detection at the identified one or more blind spots around the vehicle based on processed received data from the at least one external agent processed at the collaborative perception module;wherein the trajectory estimation module is arranged to estimate a trajectory of at least one object detected at the identified one or more blind spots around the vehicle as a trajectory estimate of the at least one object, the trajectory estimate being based on the object detection and based on the received data from the at least one external agent; andwherein the alert determination module is arranged to process the trajectory estimate transmitted from the trajectory estimation module and to determine whether to output an alert to an operator of the vehicle.

7. The system as claimed in claim 6, wherein the at least a sensor of the external agent comprises another event camera.

8. The system as claimed in claims 6 or 7, wherein the at least one external agent comprises an infrastructure object.

9. The system as claimed in any one of claims 6 to 8, wherein the data from the at least one external agent further comprises behavioural data transmitted from the at least one external agent.

10. The system as claimed in claim 9, wherein the trajectory estimation module is arranged to further estimate the trajectory of the at least one object detected at the identified one or more blind spots around the vehicle further based on the behavioural data transmitted from the at least one external agent, the further estimate including an uncertainty score attached to the trajectory estimate and wherein the alert determination module is arranged to determine whether to output the alert to the operator of the vehicle further based on the uncertainty score attached to the trajectory estimate.

11. The system as claimed in any one of claims 6 to 10, wherein the collaborative perception module is capable of generating a view around the vehicle based on the received data from the at least one external agent and the identification data of the one or more blind spots around the vehicle.

12. A computer-implemented method of determining a blind spot object alert, the computer-implemented method comprisingproviding an event camera in a vehicle (1002);characterized in that the computer-implemented method further comprisesusing the event camera to obtain event camera data and using a blind spot region detection module to process at least the event camera data to identify one or more blind spots around the vehicle (1004);receiving identification data of the one or more blind spots from the blind spot region detection module and using a collaborative perception module to instruct a transceiver coupled to the collaborative perception module to request and receive data from at least one external agent, the request being based on the identification data and the data from the at least one external agent comprises perception data of the at least one external agent (1006);processing received data from the at least one external agent using the collaborative perception module (1008);performing object detection at the identified one or more blind spots around the vehicle using an object detection module based on the processed received data processed at the collaborative perception module (1010);estimating, using a trajectory estimation module, a trajectory of at least one object detected at the identified one or more blind spots around the vehicle as a trajectory estimate of the at least one object, the trajectory estimate being based on the object detection and based on the received data from the at least one external agent (1012);processing the trajectory estimate using an alert determination module (1014); anddetermining whether to output an alert to an operator of the vehicle using the alert determination module (1016).

13. The computer-implemented method as claimed in claim 12, wherein the data from the at least one external agent further comprises behavioural data transmitted from the at least one external agent.

14. The computer-implemented method as claimed in claim 13, wherein the step of estimating further comprises further estimating the trajectory of the at least one object detected at the identified one or more blind spots around the vehicle further based on the behavioural data transmitted from the at least one external agent, the further estimating including attaching an uncertainty score to the trajectory estimate;and the step of determining whether to output an alert to an operator of the vehicle using the alert determination module is further based on the uncertainty score attached to the trajectory estimate.5 15. The computer-implemented method as claimed in any one of claims 12 to 14,further comprising generating a view around the vehicle using the collaborative perception module based on the received data from the at least one external agent and the identification data of the one or more blind spots around the vehicle.10A

Images