Self-decision making traffic signal controller

Abstract

Systems and methods for adaptive and autonomous traffic control are disclosed. The disclosures teaches a method for processing traffic information, comprising generating data through capture of images, photos, or videos, and determining and predicting movements of vehicles at an intersection. The disclosed system employs a combination of artificial intelligence (AI) processing, machine learning, and deep learning within the realm of neural network technology to recognize various types and states of traffic. The neural network technology is utilized to process the data, determine optimal traffic flow decisions, and memorize these decisions based on experiential information. The described system aims to achieve safe and efficient traffic flow at intersections for all users, comprising pedestrian and animals, as vehicles, through recognition of traffic conditions and implementation of optimal traffic flow decisions.

Description

File No. : P7229PC00SELF-DECISION MAKING TRAFFIC SIGNAL CONTROLLERCROSS-REFERENCE TO RELATED APPLICATION

[0001] This application claims priority from US provisional patent application No. 63 / 681 ,438 filed August 9, 2024, entitled SELF-DECISION MAKING TRAFFIC SIGNAL CONTROLLER, the specification of which is hereby incorporated herein by reference in its entirety.FIELD OF THE INVENTION

[0002] The present innovation relates to advance traffic control systems and methods, and, more specifically, to systems and method involving adaptive and / or autonomous control of signaling components dedicated to traffic control.BACKGROUND

[0003] Transport networks are complex with many interactions between users of the networks. T ransport network operators try to maintain flows of people and goods around the network in the most efficient ways possible.

[0004] Traditionally road networks are closely managed by the public authority using a variety of forms of traffic technology namely: traffic signals, message signs, travel apps, pricing and incident management teams and systems.

[0005] These systems and methods are oriented to the prevention and mitigation of occurrences of incidents that may disrupt vehicle traffic. The public authority acts quickly to clear the road to ensure that traffic resumes. These systems work relatively well when the travel demand is less than the capacity of the networks is adapted to service.

[0006] When the travel demand approaches the limit of the network capacity on a regular basis, forms of adaptive control are introduced. In traffic control, one control type is systems that manage flows of vehicle in an optimum way to minimize congestion and reduce delay. This operation is normally performed in a corridor by corridor basis or in a small geographical region, typically less than 4 km2. These systems perform well as long as the capacity of the network links is not exceeded and there are no incidents occurring that reduce the capacity of particular links for more than a short period of time.

[0007] When incidents occur, the network operators intervene to clear the incident with local incident management teams and inform travelers on the network of potential disruption. Occasionally the control systems are employed to manage traffic flow, enticing vehicles towards alternative roads thereby creating different flows around the incidents. However, for this action to be well planned andFile No. : P7229PC00 put in place, knowledge of similar passed incidents are required for a suitable plan to be prepared and available to deal with such an incident.

[0008] It can be desirable to frequently monitor traffic on roadways and to enable intelligent transportation system controls. For instance, traffic monitoring allows to improve control of traffic signals, speed sensing, detection of incidents (e.g., vehicular accidents) and congestion, collection of vehicles count data, flow monitoring, and numerous other objectives.

[0009] Existing traffic detection systems are available in various forms, utilizing a variety of different sensors to gather traffic data. Inductive loop systems are known to utilize a sensor installed under pavement within a given roadway. Inductive loop sensors are relatively expensive to install, replace and repair because of the associated road work that are required to access sensors located under pavement, not to mention lane closures and traffic disruptions associated therewith. Other types of sensors, such as machine vision and radar sensors can be also used. These different types of sensors each have their own advantages and disadvantages.

[0010] Safety issues are of upmost importance, and must also be taken in consideration. In modern days, with population growth and multiplication of network users and their transportation means, such as trucks, buses and car of all sizes; motorcycles and scooters; bicycles of different types and sizes, electric or gas; wheelchairs; walking aid scooters; skateboards, motorless or electric scooters; one wheel electric scooters; and all walking considerations with baby carriage, sled, toy vehicles etc., the number of accident involving a pedestrian is increasing and must be addressed to increase the security on the roads for everyone.

[0011] Despite their efforts, complexity and costs, none of the known systems has succeeded in improving the flow of road traffic and while maintaining and / or improving safety considerations.BRIEF SUMMARY OF THE INVENTION

[0012] In order to solve several problems related to autonomous traffic control such as and no limited to a) safety for vehicles (of all types) and pedestrians; b) fluidity, related to environmental concerns of limiting extra CO2emissions generated with avoidable vehicle congestions and waiting times, e.g., at red lights, at intersections etc.; c) vehicle speeding to pass on changing lights, the latter raising safety issues; and d) costs of traffic control systems usually requiring control rooms and large teams to manage.

[0013] The present invention aims to use and take advantage of available technologies and combine them with proprietary software, hardware and method of analysis, the latter benefiting from an autonomous training that keeps it at its most up-to-date level of information and analysis technique, in order to generate an instant and complete road traffic analysis for each managed intersection thatFile No. : P7229PC00 solves congestion and safety issues. This system may be implemented at a plurality of intersections. This system, as autonomous agents operating independently from each other, free of necessary communication therebetween, to understand, analyze and, directs traffic light configuration that makes smooth and safe traffic over the plurality of intersections and in-between.

[0014] A key element of the system resides in the decision analysis process being independent and automated for each of the systems. The processes are managed by Artificial Intelligence (Al) elements trained to improve decisions towards improving flow of traffic, fluidity and safety at the intersections they are managing, as well as analyzing traffic problems downstream and predicting the needs and stress the decisions yield at upstream intersections. The system is consequently able to consider the level of traffic of an intersection taking into account the fact it is dictated by the intersection downstream.

[0015] Thus, the Al module of the system is programmed to make decisions complementary to the Al module of the other systems downstream, e.g., at the two or more roadways leading to this intersection, whatever the angle(s) of the axes of these roadways. The system allows for decisionmaking efficiency complementary to other intersections without having to establish communication or exchange between intersections upstream or downstream. The different components, camera linked to processor related to databases linked to controller, are working together as an autonomous closed- circuit system able to quickly take decisions.

[0016] In some aspects, the description herein relates to a system for controlling traffic of users at an intersection, the intersection including a plurality of arriving lanes leading to a crossing area, and a plurality of departing lanes leading away to from crossing area, the system including: at least one sensor adapted to capture images of the crossing area and at least a section of each of the lanes of the intersection outside the crossing area; at least one traffic signal head adapted to provide visual signals directing users regarding behavior to adopt in relation with the intersection; and a controller adapted to receive images from the at least one sensor, to determine velocity and direction of all of the users approaching the crossing area using the lanes, to generate prediction of positions of the users overtime, and to generate and transmit traffic light signals to the at least one traffic signal head to provide visual signals based thereon, whereby the system is adapted to set a flow of the users at the intersection.

[0017] In some aspects, the description herein relates to a method of controlling flow of users at an intersection, the intersection including a plurality of arriving lanes leading to a crossing area, and a plurality of departing lanes leading away to from crossing area, the method including: capturing in a continuous manner images of the crossing area and at least a section of each of the lanes of the intersection outside the crossing are, with the users travelling on the lanes thereof; processing theFile No. : P7229PC00 images to establish user data of each one of the users travelling on the lanes thereof, the user data including a user type, a user speed and a user direction; processing the user data to establish travel predictions over time for each of the users; establishing a traffic light setting for the intersection according to at least one parameter; and transmitting a traffic light signal to at least one traffic signal head based on the traffic light setting, the at least one traffic signal head being adapted to provide visual signals directing users regarding behavior to adopt in relation with the intersection, wherein the at least one traffic signal head is controlling user traffic by lighting up colored lights at a time and for a duration based on the received traffic light signal.

[0018] Features and advantages of the subject matter hereof will become more apparent in light of the following detailed description of selected embodiments, as illustrated in the accompanying figures. As will be realized, the subject matter disclosed and claimed is capable of modifications in various respects, all without departing from the scope of the claims. Accordingly, the drawings and the description are to be regarded as illustrative in nature and not as restrictive and the full scope of the subject matter is set forth in the claims.BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Further features and advantages of the present disclosure will become apparent from the following detailed description, taken in combination with the appended drawings, in which:

[0020] Figure 1 is a schematic that illustrates roads, and intersections, with traffic control systems associated with some of the intersections, thereby depicting a realization in accordance with an embodiment;

[0021] Figure 2 is a flowchart depicting an illustrative process flow for neural network array training in accordance with an embodiment; and

[0022] Figure 3 is a flow chart of steps of a method of controlling flow of users at an intersection in accordance with an embodiment.

[0023] It will be noted that throughout the appended drawings, like features are identified by like reference numerals.DETAILED DESCRIPTION OF THE INVENTION

[0024] The realizations will now be described more fully hereinafter with reference to the accompanying figures, in which realizations are illustrated. The foregoing may, however, be embodied in many different forms and should not be construed as limited to the illustrated realizations set forth herein.File No. : P7229PC00

[0025] With respect to the present description, references to items in the singular should be understood to include items in the plural, and vice versa, unless explicitly stated otherwise or clear from the text. Grammatical conjunctions are intended to express any and all disjunctive and conjunctive combinations of conjoined clauses, sentences, words, and the like, unless otherwise stated or clear from the context. Thus, the term "or" should generally be understood to mean "and / or" and so forth.

[0026] Recitation of ranges of values and of values herein or on the drawings are not intended to be limiting, referring instead individually to any and all values falling within the range, unless otherwise indicated herein, and each separate value within such a range is incorporated into the specification as if it were individually recited herein. The use of any and all examples, or exemplary language ("e.g.," "such as," or the like) provided herein, is intended merely to better illuminate the exemplary realizations and does not pose a limitation on the scope of the realizations. No language in the specification should be construed as indicating any unclaimed element as essential to the practice of the realizations.

[0027] In the following description, it is understood that terms such as "first", "second", and the like, are words of convenience and are not to be construed as limiting terms.

[0028] Referring to Figure 1 , an area featuring a plurality of intersections 186 is depicted. Intersections 186 are interconnected, having potentially users, e.g., vehicles (not depicted), crossing an intersection 186 to reach another intersection, thus having the traffic at one of the intersections influencing the traffic at another one of the intersections 186 based on the routes of each of the vehicles travelling on the four roads 180 depicted on Figure 1 .

[0029] According to an embodiment, a self-configuring traffic signal controller system 190, hereinafter system 190, comprises at least one, but preferably a plurality of trajectory sensors 196 including one or more of the following: a radar, a video camera, or a hybrid radar and video camera, each of the trajectory sensors 196 being installed on a mast, on wire(s), on a pole, on a luminaire, or on a building or other elevated support near the traffic intersection 198, with masts, wires, or luminaires also including traffic signal heads 198 attached thereto. The system 190 further comprises a traffic controller 192 comprising a processor 200, optionally a co-processor 202, memory 194, and communication module 204 embodied as electronic hardware and software adapted to establish either wireless or wired communication with the trajectory sensors 196 and the traffic signal heads 198. The traffic controller 192 is programmed with executable instructions that cause the traffic controller 192 to obtain data from the trajectory sensors 196, and process the data to obtain: vehicle trajectory data regarding vehicles approaching, traversing, and leaving the monitored intersection 186a; and vehicle trajectory data including data regarding position, velocity, and acceleration of vehicles approaching,File No. : P7229PC00 traversing, and leaving the monitored intersection 186a. The traffic controller 192 is programmed with executable instructions to transform the vehicle trajectory data into data relative to a coordinate system derived from geometric information about the intersection 186a stored in a memory 194 at the traffic controller 192. The traffic controller 192 is programmed with executable instructions to compute, from at least the vehicle trajectory data: a delay factor representing delay of the vehicles at the intersection; a stop factor representing a number of vehicles stopped at the intersection; a capacity of the intersection reflecting a number of vehicles per minute passing through green lights in each lane; estimated emissions of the vehicles; and a safety factor. The traffic controller 192 is programmed with executable instructions to compute multiple instances of an objective function with user-defined weights that selectively prioritize one or more of the following factors: the delay factor; the stop factor; the capacity of the intersection; the estimated emissions of the vehicles; and the safety factor. The traffic controller 192 is programmed to use outputs from the computed objective function instances to adjust signal timing within a cycle at the intersection by signaling the traffic signal heads 198; and to direct the traffic signal heads 198 to change signal lights according to the adjusted signal timing.

[0030] The system 190 can be implemented together with any sub combination of the following optional features: the traffic controller 192 having stored in memory 194 preconfigured geometric intersection data so that the traffic controller 192 maps sensor data to appropriate road positions in the intersection 186a to detect vehicle trajectories within lanes 182 and with respect to road features such as stop lines 184; and the user-defined weights are derived from agency policies. The traffic controller 192 is further configured to adjust vehicle trajectory data based on at least one additional data, including: data from in-ground sensors; connected vehicle output; user device output; and output from one or more other traffic controllers of intersection(s) adjacent to the intersection 186a. The traffic controller 192 is adapted to adjust the signal timing by adjusting green signal timing, yellow clearance timing, and red clearance timing, wherein the adjustment of red clearance timing includes increasing red clearance timing based on the traffic controller 192 determining that a vehicle is or will run a red light. The traffic controller 192 is adapted to adjust the signal timing by prolonging or reducing green timing within a single cycle of signal light phases without attempting to optimize cycle offsets of multiple intersections at once or overall cycle time. The traffic controller 192 may include a coprocessor or separate circuit board that overrides a base functionality of the traffic controller 192. The traffic controller 192 uses outputs from the computed objective function instances to adjust signal timing within a cycle at the intersection 186a by selecting a traffic phase from a plurality of possible traffic phases and selecting a phase termination time from a plurality of possible phase termination times.

[0031] According to embodiments, power and communication management may be managed by having back-up power, e.g., automatically activating battery packs, associated with components ofFile No. : P7229PC00 the system 100, comprising e.g., controller 192 and trajectory sensors 196. Local storage of data and images may be performed, using Azure Blob Storage ™ integration from Microsoft ™.

[0032] In certain embodiments, a self-configuring traffic signal controller 192 process a plurality of inputs decoded from sensor signals received from a plurality of trajectory sensors 196 at an intersection 186. Each trajectory sensor 196 can comprise one or more of the following: an ultrasound sensor; a radar; and a video camera. The controller 192 can also provide a plurality of outputs such control signals to traffic signal heads 198 at the intersection 186 to cause the traffic signal heads 198 to selectively turn on and off traffic signals; a storage device 194 having stored thereon geometric intersection data representing a geometry of the intersection 186 and cycle data representing a signal timing configuration for different phases of a signal cycle of the intersection 186; and a combination of hardware and software adapted to: generate trajectory data from the plurality of inputs and the geometric intersection data, the trajectory data including information representing at least current and predicted future vehicle speeds and positions with respect to the geometric intersection data; automatically reconfigure the signal timing configuration multiple times per day by analyzing the trajectory data according to a balancing of different user-defined factors; and adjust the plurality of outputs based on the reconfigured signal timing configuration to transmit second control signals to the traffic signal heads 198 to cause the traffic signal heads 198 to selectively turn on and off the traffic signals according to the reconfigured signal timing configuration.

[0033] The apparatus of the preceding paragraph can be implemented together with any sub combination of the following optional features: the controller 192 may generate the trajectory data in part by sending the geometric intersection data to the trajectory sensors 196 so that the trajectory sensors 196 are configured to preprocess captured data and send inputs to the traffic controller 192 that are described with respect to a coordinate frame matching a geometry of the intersection 186; the inputs from the trajectory sensors 196 are not described with respect to a coordinate system corresponding to a geometry of the intersection 186, and wherein the traffic controller 192 may generate the trajectory data by transforming the inputs into the coordinate system based on the geometric intersection data; the controller 192 may generate the trajectory data in part from predicted traffic volumes in addition to the inputs received from the trajectory sensors 196; the controller 192 may generate the trajectory data in part from traffic data reported from another traffic controller 192 of an intersection adjacent to the intersection 186 in addition to the inputs received from the trajectory sensors 196; and the controller 192 may generate the trajectory data in part by predicting the future vehicle speeds and positions based on estimated or measured acceleration data. The user-defined factors may comprise parameters established based on one, and preferably two or more of the following: delay, vehicle stops, intersection capacity, emissions, and safety. The traffic controller 192 may comprise a co-processor (not depicted) or separate circuit board that overrides a traffic controllerFile No. : P7229PC00192 in relation to one or more processes. The traffic controller 192 may automatically reconfigure the signal timing configuration by selecting a traffic phase from a plurality of possible traffic phases and select a phase termination time from a plurality of possible phase termination times. The traffic controller 192 may generate the trajectory data multiple times within a single traffic signal cycle until a calculated time to remain in a current phase has been reached, wherein the calculated time is based on an average time difference between initial vehicle trajectory detection; obtained from the trajectory data; and a time at which vehicles are detected from the plurality of inputs as entering a dilemma zone. The traffic controller 192 may automatically reconfigure the signal timing configuration in response to reaching the calculated time.

[0034] In certain embodiments, as depicted in Figure 3, a flowchart of an example method for a method of controlling flow of users at an intersection, the intersection comprising a plurality of arriving lanes leading to a crossing area, and a plurality of departing lanes leading away to from crossing area.

[0035] At step 212, the method comprises capturing in a continuous manner images of the crossing area and at least a section of each of the lanes of the intersection outside the crossing are, with the users travelling on the lanes thereof.

[0036] At step 214, the method comprises processing the images to establish user data of each one of the users travelling on the lanes thereof, the user data comprising a user type, a user speed and a user direction.

[0037] At step 216, the method comprises processing the user data to establish travel predictions over time for each of the users.

[0038] At step 218, the method comprises establishing a traffic light setting for the intersection according to at least one parameter.

[0039] At step 220, the method comprises transmitting a traffic light signal to at least one traffic signal head based on the traffic light setting, the at least one traffic signal head being adapted to provide visual signals directing users regarding behavior to adopt in relation with the intersection, wherein the at least one traffic signal is controlling user traffic by lighting up colored lights at a time and for a duration based on the received traffic light signal.

[0040] Accordingly, a method may be performed by the self-configuring traffic signal controller 192. The method comprises under control of a traffic controller 192 comprising electronic hardware, receiving sensor data from a trajectory sensor 196 at an intersection 186, the trajectory sensor 186 optionally including a radar or video camera. The method comprises generating trajectory data from the sensor data based on intersection geometric data about the intersection stored in data storage. The method comprises analyzing the trajectory data according to an objective function specified byFile No. : P7229PC00 user-defined policies and automatically adjusting a traffic light signal accordingly. The method comprises outputting control signals to traffic signal lights 198 according to the adjusted signal timing configuration to cause the traffic signal lights 198 to selectively turn on and off the traffic signals according to the adjusted signal timing configuration.

[0041] The method of the preceding paragraph can be implemented together with any subcombination of the following optional features: the sensor data is specified according to a coordinate reference frame related to the geometric intersection data; generating the trajectory data includes using vehicle speeds in the sensor data to predict future vehicle positions with respect to the intersection; automatically adjusting the signal timing configuration of the traffic controller includes adjusting one or more of the time and duration the green-light is lit, the time and duration the yellow light is lit, and time and duration the red light is lit according to predicted future vehicle trajectory; the user-defined policies emphasize some policies over other policies in the objective function; further including providing at least some of the trajectory data to another traffic controller associated with another intersection to enable the other traffic controller to use at least some of the trajectory data to adjust signal timing of at the other intersection; the objective function is user-definable; automatically reconfiguring the signal timing configuration includes selecting a traffic phase from a plurality of possible traffic phases and selecting a phase termination time from a plurality of possible phase termination times.

[0042] In certain embodiments, a self-configuring traffic signal controller 192 comprises electronic hardware that: receives sensor data from a trajectory sensor 196 at an intersection 186, the trajectory sensor 196 optionally including a radar or video camera; generates trajectory data from the sensor data based on intersection geometric data about the intersection 186 stored in memory #, the trajectory data including data about vehicles speeds; automatically adjusts a signal timing configuration of the traffic controller by analyzing the trajectory data according to user-defined policies; and outputs control signals to traffic signal lights according to the adjusted signal timing configuration to cause the traffic signal lights to selectively turn on and off the traffic signals according to the adjusted signal timing configuration.

[0043] The traffic controller 192 of the preceding paragraph can be implemented in a system 190 together with any sub combination of the following optional features: the sensor data is specified according to a coordinate reference frame related to the geometric intersection data; the traffic controller 192 generates the trajectory data by at least using vehicle speeds in the sensor data to predict future vehicle positions with respect to the intersection 186; the traffic controller 192 generates the trajectory data by at least using vehicle speeds in the sensor data to predict future vehicle speeds with respect to the intersection 186; the traffic controller 192 automatically adjusts the signal timing configuration of the traffic controller by at least adjusting one or more of time and duration the greenFile No. : P7229PC00 light is lit, time and duration the yellow light is lit, and time and duration the red light is lit according to predicted future vehicle trajectory; the user-defined policies weight some policies over other policies; the traffic controller 192 also provides at least some of the trajectory data to another traffic controller at another intersection to enable the other traffic controller to use at least some of the trajectory data to adjust signal timing of at the other intersection; the traffic controller 192 may comprise a coprocessor or separate circuit board that overrides at least some processing of the processor of the traffic controller 192; the traffic controller 192 automatically reconfigures the signal timing configuration by selecting a traffic phase from a plurality of possible traffic phases and selecting a phase termination time from a plurality of possible phase termination times; the traffic controller 192 generates the trajectory data multiple times within a single traffic signal cycle until a calculated time to remain in a current phase has been reached; the calculated time is based on an average time difference between initial vehicle trajectory detection, obtained from the trajectory data, and a time at which vehicles are detected from the plurality of inputs as entering a dilemma zone; and the traffic controller automatically reconfigures the signal timing configuration in response to reaching the calculated time.

[0044] Certain aspects, advantages and novel features of the inventions can be described herein. It can be understood that not necessarily all such advantages may be achieved in accordance with any particular embodiment of the inventions disclosed herein. Thus, the inventions disclosed herein may be embodied or carried out in a manner that achieves or selects one advantage or group of advantages as taught herein without necessarily achieving other advantages as may be taught or suggested herein.

[0045] This invention takes advantage of different technologies known in the art and make them work independently, such as:1- Image capture2- Filters and conversion into digital data3- Recognition using database #1• In parallel, and image recovery process is executed by Al to feed and educate database #14- Analysis of object movements by comparison and in relation with all elements based on database #2, followed by statistical calculation estimating future events• Uses Al to analyze object movements• Uses Al to maintain and grow database #2 for moving objects5- Draw conclusion about potentially dangerous situation and fluidity improvement based on learned events from database #36- Decision making and sending signal to lights controller affecting traffic lights and pedestrian signalsFile No. : P7229PC00

[0046] This invention involves the different steps that are:1 . Image Capture: Capturing images with one or more trajectory sensors 196, e.g., cameras at a rate of at least, but not limited to, one per half second for tracking fast moving objects, and may use an adaptative rate based on different situations detected by the camera (such as crowded situation, unusually fast moving vehicles, unusual events, troubled mobility, etc.). Alternatively, or in combination, high-resolution cameras and / or even infrared / thermal cameras may be used for facing low visibility conditions.2. Filters and Conversion to Digital Data: pre-processing images to improve the quality and reduce noise, which is important for accurate object recognition. Techniques like edge detection or color filtering are used. The filtered images are converted into digitalized data for further analysis.3. Recognition by data comparison from database #1 : using machine learning models to identify objects in the images. Trained models through a comprehensive dataset that includes different types of vehicles, pedestrians, and other relevant entities are used therefor.4. Object Motion Analysis: tracking the detected objects across multiple frames to understand their trajectory and speed. Techniques such as optical flow or the use of recurrent neural networks (RNNs) are used to analyze the movement over time and anticipate the subject action.5. Statistical computation for future prediction: predictive modeling is used to estimate the future positions of objects based on their current trajectories. Implementing models that can account for sudden changes in direction or speed may be used to improve the accuracy of the predictions.6. Conclusion and hazard prediction: Integrating a risk assessment that takes into account historical accident data, typical traffic patterns, and real-time data from other sensors (such as weather conditions) helping to more accurately assess potential hazardous situations.7. Decision making and signal control: automating the control of traffic and pedestrian lights based on the assessed risks and traffic conditions.Accordingly, the system 190 is adapted to be robust, adapted to handle typical scenarios as well as exceptions, such as the presence of emergency vehicles.

[0047] In parallel to these steps, the software is adapted to perform specific processes to improve its capabilities.• Maintain and grow databases: Ensuring the continuous update of the databases with new data from the field to improve the accuracy of each Al models. This involves automatic data labeling techniques or semi-supervised learning approaches.File No. : P7229PC00• Al improvements: Implementing machine learning operations (MLOps) practices streamlining the process of training, monitoring, and deploying models, ensuring that the system remains effective as conditions change, wherein such training may be performed in real time as from recording of real situations, e.g., Youtube ™ videos.

[0048] Such a system operates according to two orders of priority: 1 ) safety, and 2) traffic flow minimizing CO2 emissions. The Al analysis is adapted to consider these two priority elements for a traffic load balance at each intersection (for example analyzing all the possibilities of a classic northsouth, east-west intersection) or according to the particular configuration of the intersection.

[0049] In the example presented in Figure 1 , the traffic controller 192 associated with the intersection 186b educates itself and includes in its analysis the influence of traffic at the intersection 186a and intersection 186c. The traffic controller 192 at the intersection 186c uses a decision-making influenced and educated by the controller 192 of the intersection 186b. Etc.

[0050] An important element of the system 190 is that for each intersection, the Al module of the traffic controller 192 automatically adapts to the road characteristics and special (and often unique) events, which could be brief like an accident, or long such as but not limited to emergency vehicles in a building where the road requiring a bypass or slowing of traffic, or which can be even longer such as road repair.

[0051] In another embodiment of this invention, the Al module of the traffic controller 192 considers other environmental events, such as bad weather to extreme weather that can make driving decision dangerous or event that can make its decision dangerous or erroneous. The Al module is adapted to automatically decide to suspend its decision making and go back to the pre-program traffic lights cycle.

[0052] In another embodiment of this invention, the Al module of the controller 192 collects information and data to sporadically send them to a processing center (not depicted) which function is to process collected data, and permit a regular update of the analysis software of each Al module, and together providing an important source of information about city traffic and human behavior.

[0053] FIG. 2 is a diagram of an illustrative process flow for neural network array function and training. Each street intersection 100 are individual and not in communication with any processing and / or decisional center, nor with the other adjacent or remote controller at other street corners. The process starts from cameras 102 that take images at auto-adjustable image capture rates based on the evolution of the events. The captured and stored visual information is electronically filtered 104 in order to increase the read accuracy, then digitalized for processing. The digitalized images are then enhanced, denoised and filtered 106 to be ready for the object recognition 108 with evaluation by the artificial intelligence based database 110 allowing the formal object identification 112 to be processedFile No. : P7229PC00 for first trajectory and speed 114 estimation in work with the database 115 permitting the required statistical analysis dictating the behavior prediction and moving position in time 116. Further analysis using historical events from database 122 leads to the system drafting risk assessment 120 and hazard prediction allowing an automated decision making 124 managing circulation 126 of vehicles and pedestrians, hence drastically increasing intersection safety level while defining the optimal configuration reducing traffic conflicts, traffic jams, congestions in order to reduce CO2 emissions.

[0054] The training method of the system 190 is using real-time information from e.g., traffic cameras 102, which may first undergo an initial layer training. They are presented to the lowest layer neurons, which may be trained to recognize 113 each defined traffic object, traffic object group, and / or position / distance zone and / or displacement speed including trend from past gathered and constantly evolving data 110. In a recognitions processing step, data that is not recognized or classified may be returned for further processing, while recognized data is passed through for secondary handling such as storage and processing associated with a recognition mode. The required number of neurons may be activated until the complete set of these recognitions is accomplished. It may be sufficient for most traffic flow decisions to have overlapping influence fields of various neurons which can yield a match, since the recognition of any traffic object, especially in the case of a solo traffic object, can still produce the correct traffic flow decision at the highest level. The accuracy of recognition is determined by number of available neurons relative to the number of examples input vectors to be recognized.

[0055] After lowest layer neurons are trained, they are set to normal operational recognition mode and presented with the same training data set they were trained with, while the next higher neuron level is put into training mode, referred to as intermediate layer training. The next level neurons are progressively activated and trained, until the defined set of classifications for that level are learned. This process is iterated for each intermediate layer when there is more than one such layer.

[0056] Three levels of analysis are used for fast and highly accurate predictive model 117. Consequently, 3 complementary databases are used to proceed to a complete analysis of all possible situations, while self-feeding new data in the database 115 are fed as new and unpredicted events occurred. The risk assessment relates to a specific database 122 hence greatly increases the analysis accuracy.

[0057] Amongst the parallel and complementary actions of the system 190, valuable data are collected about traffic events and citizen behavior 115. Such data are collected, anonymized, and stored in the processing center for further uses, including improving the operating system of Al modules of the controllers of each intersection, and collect study data to seek improving life, safety and air quality in our always evolving environment.File No. : P7229PC00

[0058] According to embodiments, intersections may include railroad crossings, with users of intersections including trains. According to such embodiment, monitoring and image captures of railroads follows the same principle and process of monitoring and image capturing of other users of intersections, with differences residing in the nature and geometry of the railroad crossings, and characteristics of trains being substantially different from characteristics of other users such as personal vehicles.

[0059] Applicant stresses that image capture and identification process of types of users provide way to prioritize users over other rather than only considering traffic flow in some conditions, for instance in relation with emergency vehicles such as ambulances, fire trucks, and police vehicles. It also allows to prioritize traffic flow further based on potential number of travelers rather only on e.g., vehicles, allowing for instance to prioritize travelling of city buses over personal vehicles.

[0060] According to embodiments, sensors may be adapted in number, position and definition to capture number of occupants of vehicles in addition to identify type of user (vehicle), allowing based on, e.g., regulations, to prioritize traffic flow furthermore based on number of individuals to flowthrough the intersection in addition to other parameters provided.

[0061] This innovative approach combines digital technology, artificial intelligence and fast computing, bringing technological improvement on traffic control processes and thus increasing traffic fluidity, reducing costs, and increasing road safety.

[0062] Elements resulting from analysis of images can be of two types: Overall frame data and Object detection data.

[0063] Examples of Overall frame data are the following:• timestamp: The date and time when the image frame was captured, (e.g., "2025-05- 09T15:59:53.384190")• framejd: A unique identifier for the specific frame captured, (e.g., 596)• detections: A list containing detailed information about each object detected in the frame.• queues: A list that would likely contain information about vehicle queues, though it is empty in this example.

[0064] Examples of Object detection data are the following:• objectjd: A unique identifier for the detected object.• trackjd: An identifier to track the object across multiple frames.• bbox: The bounding box coordinates that locate the object in the image. It consists of:• x1 , y1 : The coordinates of the top-left corner.• x2, y2: The coordinates of the bottom-right corner.File No. : P7229PC00• confidence: The confidence level of the detection model's prediction, from 0 to 1 .• class: The classification of the detected object (e.g., "car").• center: The pixel coordinates of the center of the bounding box.• speed: The calculated speed of the object.• direction: The cardinal direction of the object's movement (e.g., "S" for South, "N" for North).• direction_angle: The angle of the object's direction in degrees.• zonejd: The identifier of the zone where the object is located, which is null in this case.• waiting_time: The duration, in seconds, that the object has been stationary or waiting.• first_seen: The framejd when the object was first detected.• last_seen: The framejd when the object was last seen.• hits: The total number of frames in which the object has been detected.

[0065] According to an embodiment, the traffic controller 192 comprises a Bus Interface Unit (BIU) that translates the command and sends a low-voltage electrical signal to the correct load switches of the traffic signal heads 198, wherein the load switches are adapted to act as relays, sending high-voltage power supply to the appropriate traffic lights, causing them to illuminate in response to low-voltage traffic light signals.

[0066] Hence, embodiments are available according to any one of the following clauses:

[0067] Clause 1. A system for controlling traffic of users at an intersection, the intersection comprising a plurality of arriving lanes leading to a crossing area, and a plurality of departing lanes leading away to from crossing area, the system comprising: at least one sensor adapted to capture images of the crossing area and at least a section of each of the lanes of the intersection outside the crossing area; at least one traffic signal head adapted to provide visual signals directing users regarding behavior to adopt in relation with the intersection; and a controller adapted to receive images from the at least one sensor, to determine velocity and direction of all of the users approaching the crossing area using the lanes, to generate prediction of positions of the users over time, and to generate and transmit traffic light signals to the at least one traffic signal head to provide visual signals based thereon, whereby the system is adapted to set a flow of the users at the intersection.

[0068] Clause 2. The system of clause 1 , wherein the sensors comprise at least one of a camera and a radar.

[0069] Clause 3. The system of any one of clause 1 and clause 2, wherein the controller comprises a memory storing geometry of the intersection.

[0070] Clause 4. The system of any one of clauses 1 to 3, wherein the controller comprises a database associated with machine learning.File No. : P7229PC00

[0071] Clause 5. The system of any one of clauses 1 to 4, wherein the controller comprises a data associated with prediction models.

[0072] Clause 6. The system of any one of clauses 1 to 5, wherein the controller comprises a database associated with events that occurred at the intersection.

[0073] Clause 7. The system of any one of clauses 1 to 6, wherein the controller comprises user parameters, wherein the controller compares users with user parameters to establish a type for each of the users.

[0074] Clause 8. The system of any one of clauses 1 to 7, wherein the each one of the at least one traffic signal head comprises a set of colored lights, wherein the controller is adapted to control time and duration of each of the colored lights are lit.

[0075] Clause 9. The system of any one of clauses 1 to 8, further comprising a weather sensor for detecting weather conditions.

[0076] Clause 10. The system of any one of clauses 1 to 9, wherein lanes are leading to another intersection having another controller associated therewith, wherein the controller and the other controller operate independently from one another.

[0077] Clause 11. A method of controlling flow of users at an intersection, the intersection comprising a plurality of arriving lanes leading to a crossing area, and a plurality of departing lanes leading away to from crossing area, the method comprising: capturing in a continuous manner images of the crossing area and at least a section of each of the lanes of the intersection outside the crossing are, with the users travelling on the lanes thereof; processing the images to establish user data of each one of the users travelling on the lanes thereof, the user data comprising a user type, a user speed and a user direction; processing the user data to establish travel predictions over time for each of the users; establishing a traffic light setting for the intersection according to at least one parameter; and transmitting a traffic light signal to at least one traffic signal head based on the traffic light setting, the at least one traffic signal head being adapted to provide visual signals directing users regarding behavior to adopt in relation with the intersection, wherein the at least one traffic signal is controlling user traffic by lighting up colored lights at a time and for a duration based on the received traffic light signal .

[0078] Clause 12. The method of clause 11 , wherein the step of processing the user data comprises evaluating travel prediction according to intersection geometry data.

[0079] Clause 13. The method of any one of clause 11 and clause 12, wherein the step of processing the user data comprises establishing travel predictions of users having probabilities future users.File No. : P7229PC00

[0080] Clause 14. The method of any one of clauses 11 to 13, wherein the step of processing the user data comprises computing a future safety conflict score for one or more of the users based on user data and safety conflict prediction according to a plurality of available traffic light signals.

[0081] Clause 15. The method of any one of clauses 11 to 14, wherein the step of determining a user type comprises determining whether the user is one of a list comprising at least two of: a regular vehicle, a special vehicle, a human and a quadruped.

[0082] Clause 16. The method of any one of clauses 11 to 15, further comprising automatically adjusting the flow configuration by selecting a traffic light signal from a plurality of stored traffic light signals and selecting a termination time to be applied thereto.

[0083] Clause 17. The method of any one of clauses 11 to 16, wherein the at least one parameter comprises at least one of user safety, travel fluidity at the intersection, CO2emission by the users, number of users in a lane, maximum waiting time of a user, and weather condition.

[0084] Clause 18. The method of any one of clauses 11 to 17, further comprising exchanging intersection usage data between systems controlling user traffic of intersections interconnected by at least one of the lanes thereof.

[0085] Clause 19. The method of any one of clauses 11 to 18, further comprising evaluating user traffic resulting from users crossing the intersections according to the traffic light signal comprising training an Al module with user data and user traffic resulting from a plurality of traffic light signals over time.

[0086] Clause 20. The method of any one of clauses 11 to 19, further comprising processing the images for at least one of anonymizing the users, tokenizing the users, vectorizing the users, and setting a geometry to the users. Provisional application claims

[0087] Clause 21. A self-decision making traffic signal controller method, the method comprising: (a) under control of a traffic controller comprising electronic hardware, (b) receiving sensor data from a trajectory sensor at an intersection, the trajectory sensor optionally including a radar, sensors such as doppler speed detection, underground cables or video camera; (c) generating trajectory data from the sensor data based on intersection geometric data about the intersection stored in data storage, wherein said generating the trajectory data comprises (i) computing predicted future vehicle and pedestrian speeds and positions fora plurality of vehicles for each of a plurality of different future states of a traffic signal, and (ii) computing a future safety conflict score for one or more of the vehicles of the plurality of vehicles and pedestrians based on at least in part on the predicted future vehicle and pedestrian speeds and positions and safety conflicts predicted to occur based on each of a plurality of future state signal timings; (d) automatically adjusting a signal timing configuration of theFile No. : P7229PC00 traffic controller, for vehicles and pedestrians, by analyzing the trajectory data according to an objective function specified by user-defined policies; and (e) outputting control signals to traffic signal lights according to the adjusted signal timing configuration to cause the traffic signal lights to selectively turn on and off the traffic signals according to the adjusted signal timing configuration.

[0088] Clause 22. The method of clause 21 , wherein the sensor data is specified according to a coordinate reference frame related to the intersection geometric data.

[0089] Clause 23. The method of clause 21 , wherein generating the trajectory data comprises using vehicle speeds in the sensor data to predict future vehicle positions with respect to the intersection.

[0090] Clause 24. The method of clause 21 , wherein generating the trajectory data comprises using pedestrian speeds in the sensor data to predict future vehicle positions with respect to the intersection.

[0091] Clause 25. The method of clause 21 , wherein generating the trajectory data comprises using non-human but living biped or quadruped's speed in the sensor data to predict future vehicle positions with respect to the intersection.

[0092] Clause 26. The method of clause 21 , wherein automatically adjusting the signal timing configuration of the traffic controller comprises adjusting one or more of green time, yellow time, and red time according to predicted future vehicle trajectory.

[0093] Clause 27. The method of clause 21 , wherein the user-defined policies emphasize some policies over other policies in the objective function.

[0094] Clause 28. The method of clause 21 , further comprising providing at least some of the trajectory data to a second traffic controller at another intersection to enable the second traffic controller to use at least some of the trajectory data to adjust signal timing of at the second traffic controller.

[0095] Clause 29. The method of clause 21 , wherein said automatically adjusting the signal timing configuration comprises selecting a traffic phase from a plurality of possible traffic phases and selecting a phase termination time from a plurality of possible phase termination times.

[0096] Clause 30. A self-decision making traffic signal controller apparatus, the apparatus comprising: (a) a traffic controller comprising electronic hardware that: (b) receives sensor data from a trajectory sensor at an intersection, the trajectory sensor optionally including a radar, sensors such as doppler speed detection, underground cables or video camera; (c) generates trajectory data using different combined artificial intelligence modules using a neural network method to process data from the sensor data based on intersection geometric data about the intersection stored in data storage,File No. : P7229PC00 the trajectory data comprising data about vehicles, pedestrians and animals speeds, wherein said generation of the trajectory data comprises (i) computing predicted future vehicle, pedestrians and animals speeds and positions for a plurality of vehicles, pedestrians and animals for each of a plurality of different future states of a traffic signal, and (ii) computing a future safety conflict score for one or more of the vehicles, pedestrians and animals of the plurality of vehicles, pedestrians and animals based on at least in part on the predicted future vehicle , pedestrian and animal speeds and positions and safety conflicts predicted to occur based on each of a plurality of future state signal timings; (d) automatically adjusts a signal timing configuration of the traffic controller by analyzing the trajectory data according to user-defined policies; and (e) outputs control signals to traffic and pedestrian signal lights according to the adjusted signal timing configuration to cause the traffic signal lights to selectively turn on and off the traffic signals according to the adjusted signal timing configuration.

[0097] Clause 31. The apparatus of clause 30, wherein the decision analysis of the elements are independent and automated while managed by artificial intelligence elements concentrated on decisions towards the level of traffic, fluidity and danger of the intersection for which it is acting, as well as to analyze traffic problems downstream and to predict the needs and stress that it will send to the intersections upstream.

[0098] Clause 32. The apparatus of clause 30, wherein the system considers the level of traffic of an intersection taking in account the fact it is dictated by the intersection downstream.

[0099] Clause 33. The apparatus of clause 30, wherein the artificial intelligence based module is programmed to make decisions complementary to the Al module of the corners downstream at the two or more roadways leading to meet ant this intersection, whatever the angle(s) of the axes of these roadways.[000100] Clause 34. The apparatus of clause 30, wherein the traffic controller generates the trajectory data by at least using moving object speeds in the sensor data to predict future vehicle positions with respect to the intersection.[000101] Clause 35. The apparatus of clause 30, wherein the traffic controller generates the trajectory data by at least using moving object speeds in the sensor data to predict future moving object speeds with respect to the intersection.[000102] Clause 36. The apparatus of clause 30, wherein the autonomous and complete system is repeated at each intersection and is able, without communication requirement between these autonomous systems at each intersections, to collect, analyze and proceed to the most adequate decision on the best traffic light configuration to make safe and smooth traffic throughout a city.File No. : P7229PC00[000103] Clause 37. The apparatus of clause 30, wherein the artificial intelligence based module draw conclusion about potentially dangerous situation and fluidity improvement based on learned events from database[000104] Clause 38. The apparatus of clause 30, wherein the traffic controller artificial intelligence based module automatically adjusts the signal timing configuration of the traffic controller by at least adjusting one or more of green time, yellow time, and red time according to predicted future vehicle trajectory.[000105] Clause 39. The apparatus of clause 30, wherein the user-defined policies weight some policies over other policies.[000106] Clause 40. The apparatus of clause 30, wherein the traffic controller artificial intelligence based module has an autonomous training constantly keeping it at its most up-to-date level of information, in order to generate an instant and complete road traffic analysis, for each individual intersection.[000107] Clause 41. The apparatus of clause 30, wherein the traffic controller artificial intelligence based module comprises a co-processor or separate circuit board that overrides a traffic controller.[000108] Clause 42. The apparatus of clause 30, wherein the traffic controller artificial intelligence based module automatically adjusts the signal timing configuration by generating a traffic phase from a plurality of possible traffic phases and selecting a phase termination time from a plurality of possible phase termination times.[000109] Clause 43. The apparatus of clause 30, wherein the traffic controller artificial intelligence based module generates the trajectory data multiple times within a single traffic signal cycle until a calculated time to remain in a current phase has been reached.[000110] Clause 44. The apparatus of clause 43, wherein the calculated time is based on an average time difference between initial vehicle trajectory detection, obtained from the trajectory data, and a time at which vehicles are detected from a plurality of inputs as entering a dilemma zone.[000111] Clause 45. The apparatus of clause 44, wherein the traffic controller automatically adjusts the signal timing configuration in response to reaching the calculated time.[000112] While preferred embodiments have been described above and illustrated in the accompanying drawings, it will be evident to those skilled in the art that modifications may be made without departing from this disclosure. Such modifications are considered as possible variants comprised in the scope of the disclosure.

Claims

File No. : P7229PC00CLAIMS:1 . A system for controlling traffic of users at an intersection, the intersection comprising a plurality of arriving lanes leading to a crossing area, and a plurality of departing lanes leading away to from crossing area, the system comprising: at least one sensor adapted to capture images of the crossing area and at least a section of each of the lanes of the intersection outside the crossing area; at least one traffic signal head adapted to provide visual signals directing users regarding behavior to adopt in relation with the intersection; and a controller adapted to receive images from the at least one sensor, to determine velocity and direction of all of the users approaching the crossing area using the lanes, to generate prediction of positions of the users over time, and to generate and transmit traffic light signal signals to the at least one traffic signal head to provide visual signals based thereon, whereby the system is adapted to set a flow of the users at the intersection.

2. The system as claimed in claim 1 , wherein the at least one sensor comprises at least one of a camera and a radar.

3. The system as claimed in any one of claim 1 and claim 2, wherein the controller comprises a memory storing geometry of the intersection.

4. The system as claimed in any one of claims 1 to 3, wherein the controller comprises a database associated with machine learning.

5. The system as claimed in any one of claims 1 to 4, wherein the controller comprises a data associated with prediction models.

6. The system as claimed in any one of claims 1 to 5, wherein the controller comprises a database associated with events that occurred at the intersection.

7. The system as claimed in any one of claims 1 to 6, wherein the controller comprises user parameters, wherein the controller compares users with user parameters to establish a type for each of the users.File No. : P7229PC008. The system as claimed in any one of claims 1 to 7, wherein the each one of the at least one traffic signal head comprises a set of colored lights, wherein the controller is adapted to control time and duration of each of the colored lights are lit.

9. The system as claimed in any one of claims 1 to 8, further comprising a weather sensor for detecting weather conditions.

10. The system as claimed in any one of claims 1 to 9, wherein lanes are leading to another intersection having another controller associated therewith, wherein the controller and the other controller operate independently from one another.

11. A method of controlling flow of users at an intersection, the intersection comprising a plurality of arriving lanes leading to a crossing area, and a plurality of departing lanes leading away to from a crossing area, the method comprising: capturing in a continuous manner images of the crossing area and at least a section of each of the lanes of the intersection outside the crossing are, with the users travelling on the lanes thereof; processing the images to establish user data of each one of the users travelling on the lanes thereof, the user data comprising a user type, a user speed and a user direction; processing the user data to establish travel predictions over time for each of the users; establishing a traffic light setting for the intersection according to at least one parameter; and transmitting a traffic light signal to at least one traffic signal head based on the traffic light setting, the at least one traffic signal head being adapted to provide visual signals directing users regarding behavior to adopt in relation with the intersection, wherein the at least one traffic signal head is controlling user traffic by lighting up colored lights at a time and for a duration based on the received traffic light signal .

12. The method as claimed in claim 11 , wherein the step of processing the user data comprises evaluating travel prediction according to intersection geometry data.

13. The method as claimed in any one of claim 11 and claim 12, wherein the step of processing the user data comprises establishing travel predictions of users having probabilities future users.

14. The method as claimed in any one of claims 11 to 13, wherein the step of processing the user data comprises computing a future safety conflict score for one or more of the users based on user data and safety conflict prediction according to a plurality of available traffic light signals.File No. : P7229PC0015. The method as claimed in any one of claims 11 to 14, wherein the step of determining a user type comprises determining whether the user is one of a list comprising at least two of: a regular vehicle, a special vehicle, a human and a quadruped.

16. The method as claimed in any one of claims 11 to 15, further comprising automatically adjusting a flow configuration by selecting a traffic light signal from a plurality of stored traffic light signals and selecting a termination time to be applied thereto.

17. The method as claimed in any one of claims 11 to 16, wherein the at least one parameter comprises at least one of user safety, travel fluidity at the intersection, CO2 emission by the users, number of users in a lane, maximum waiting time of a user, and weather condition.

18. The method as claimed in any one of claims 11 to 17, further comprising exchanging intersection usage data between systems controlling user traffic of intersections interconnected by at least one of the lanes thereof.

19. The method as claimed in any one of claims 11 to 18, further comprising evaluating user traffic resulting from users crossing the intersections according to the traffic light signal comprising training an Al module with user data and user traffic resulting from a plurality of traffic light signals over time.

20. The method as claimed in any one of claims 11 to 19, further comprising processing the images for at least one of anonymizing the users, tokenizing the users, vectorizing the users, and setting a geometry to the users.

Images