SYSTEM AND METHOD FOR PREDICTIVE NAVIGATION CONTROL

Abstract

Method (100) for predictive navigation control for an ego vehicle (20), comprising: Comparing a clue node (34) with each of a plurality of episodic memory nodes (36) in an episodic memory structure (38), wherein the clue node (34) represents a new event (30) associated with a set of respective locations (L), velocities (S), and directions (H) associated with one or more newly observed objects (21), each located within a respective plurality of newly defined attention zones (32) around the ego-vehicle (20), and wherein the episodic memory structure (38) comprises a selectively interconnected and directed network of the episodic memory nodes (36), each episodic memory node (36) representing a respective previously existing event (40), each previously existing event (40) being associated with a respective set of locations (L), velocities (S), and directions (H),which are associated with one or more previously observed objects (43), each located within a plurality of previously defined attention zones (42) around the ego-vehicle (20), and wherein each episodic storage node (36) has a respective node risk (NR) and a respective probability (LL) associated with it; , Determine which of the several episodic storage nodes (36) has the smallest respective difference metric, thereby defining a best-fitting episodic storage node (36), wherein each respective difference metric is determined on the basis of a respective total difference between one or more respective properties (49) of the respective episodic storage node (36) and the hint node (34); Consolidating the hint node (34) with the best-matching episodic storage node (36) if the smallest respective difference metric is less than a predetermined matching tolerance, or adding a new episodic storage node (36) corresponding to the hint node (34) to the episodic storage structure (38) if the smallest respective difference metric is greater than or equal to the predetermined matching tolerance; and Identifying one or both of (i) a most probable nearest episodic storage node (36) among one or more episodic storage nodes (36) immediately downstream of the best-matching or new episodic storage node (36), wherein the most probable nearest episodic storage node (36) has a highest probability (LL) among the immediately downstream episodic storage nodes (36), and (ii) a most risky nearest episodic storage node (36) among the one or more episodic storage nodes (36) immediately downstream of the best-matching or new episodic storage node (36), wherein the most risky nearest episodic storage node (36) has a highest node risk (NR) among the immediately downstream episodic storage nodes (36).

Description

[0001] This description refers generally to predictive navigation control systems and methods for autonomous and semi-autonomous vehicles.

[0002] Autonomous and semi-autonomous vehicles (sometimes called ego vehicles or host vehicles) offer a degree of self-navigation (i.e., navigation of the vehicle without human intervention in real time). Such self-navigation can rely on the vehicle's control system knowing its surroundings and its real-time location within those surroundings. To achieve this, the vehicle's control system can use a combination of stored information (e.g., maps), information received periodically or in real time from outside the vehicle via sensors, and logic programmed into the control system.

[0003] German patent DE 10 2018 101 505 A1 describes a system and a method for data processing in which the environment surrounding the vehicle is encoded into egocentric and geocentric overlapping coordinate systems. These overlapping coordinate systems are then subdivided into adaptively dimensioned grid cells according to the properties of the environments and the vehicle's status. Each grid cell is defined by one of several representative event patterns and risk values ​​for the vehicle. The autonomous driving system is then operational to provide a real-time assessment of the environment in response to the grid cell data, and temporal sequences of the grid cell data are stored in episodic memory and retrieved from it during driving.

[0004] German patent application DE 10 2020 103 754 A1 describes an autonomous vehicle, a system, and a method for operating the autonomous vehicle. The system comprises an episodic memory, a hyper-association module, and a navigation system. The episodic memory stores a multitude of episodes, retrieves a multitude of candidate episodes upon receiving a partial prefix, and retrieves a hypothesis episode upon receiving an intermediate episode. The hyper-association module retrieves the multitude of candidate episodes from the episodic memory and obtains the intermediate episode from the multitude of candidate episodes. The navigation system navigates the autonomous vehicle using the hypothesis episode.

[0005] German patent DE 10 2018 102 027 A1 describes a system and a method for data processing in an autonomous vehicle control system. By using information from the vehicle, the network interface, and the sensors attached to the vehicle, the system can perceive situations with less computational complexity without losing important details and then make navigation and control decisions. The system and the method serve to generate, store, and retrieve situation-aware events in order to predict situations for autonomous driving.

[0006] German patent DE 102019 108 644 A1 describes a method and a device for generating an action for controlling an autonomous vehicle. In particular, the method is functional for receiving an input indicating a training event, dividing the driving episode into a plurality of time steps, generating a syntax tree in response to each time step, and generating a highly probable syntax tree from a combination of the generated syntax trees.

[0007] It can be considered a task to specify an improved method and improved control for predictive navigation control for an ego vehicle.

[0008] The problem is solved by a method for predictive navigation control for an ego vehicle according to claim 1 and a control for predictive navigation control for an ego vehicle according to claim 7.

[0009] A method according to the invention for predictive navigation control for an ego-vehicle comprises: (a) comparing a clue node with each of a plurality of episodic memory nodes in an episodic memory structure, wherein the clue node represents a new event associated with a set of respective locations, velocities, and directions associated with one or more newly observed objects, each located within a respective plurality of newly defined attention zones around the ego-vehicle, and wherein the episodic memory structure contains a selectively interconnected and directed network of episodic memory nodes, each episodic memory node representing a respective previously existing event, each previously existing event being associated with a respective set of locations, velocities, and directions.(a) which are associated with one or more previously observed objects, each located within one of several previously defined attention zones around the ego-vehicle, and wherein each episodic memory node has a respective node risk and probability associated with it; (b) Determine which of the plurality of episodic memory nodes has the smallest respective difference metric, thereby defining a best-matching episodic memory node, wherein each respective difference metric is determined based on a respective total difference between one or more respective properties of the respective episodic memory node and the clue node; (c) Consolidate the clue node with the best-matching episodic memory node if the smallest respective difference metric is smaller than a predetermined match tolerance.or adding a new episodic storage node corresponding to the clue node to the episodic storage structure if the smallest respective difference metric is greater than or equal to the predetermined match tolerance; and (d) identifying one or both of (i) a most probable next episodic storage node among one or more episodic storage nodes immediately downstream of the best-matching or new episodic storage node, wherein the most probable next episodic storage node has the highest probability among the immediately downstream episodic storage nodes, and (ii) a riskiest next episodic storage node among the one or more episodic storage nodes immediately downstream of the best-matching or new episodic storage node,where the riskiest nearest episodic storage node has the highest node risk among the immediately downstream episodic storage nodes.

[0010] In one embodiment, the respective node risk of each episodic storage node is a respective maximum, an average, or an aggregate of the respective object risks for the one or more previously observed objects in the associated previously existing event, and the respective object risk for each previously observed object is determined by a sigmoidal function applied to a respective distance between the previously observed object and the ego vehicle.

[0011] In one embodiment, the method further comprises: calculating a respective individual risk for each of the one or more newly observed objects using the sigmoid function applied to a respective distance between each respective newly observed object and the ego-vehicle; determining a respective overall risk for each of the plurality of newly defined attention zones, based on the respective individual risks of the one or more newly observed objects located within each respective newly defined attention zone; and defining the indicator node as a grouping of the plurality of newly defined attention zones organized according to their respective overall risks.

[0012] In one embodiment, the respective object risk or individual risk for each previously observed or newly observed object is determined by R = 2-({1 - 1 / (1 + e^[(mindist - SAFEDIST) / (SAFEDIST / 2)]} - 0.5) + 0.5, where R is the respective object risk or individual risk, mindist is a distance of the previously observed or newly observed object from the ego vehicle, and where SAFEDIST is a distance that depends on one or more of a road surface type, a road structure type, weather / environmental conditions and a relative lane position, closing speed or closing acceleration between the previously observed or newly observed object and the ego vehicle.

[0013] In one embodiment, each of the respective locations, velocities, and directions of the one or more newly observed objects and of the one or more previously observed objects is defined in relation to the ego-vehicle.Each respective total difference is a respective sum of one or more weighted penalties assigned against each of one or more differences between the respective properties of the respective episodic memory node and the clue node, and the properties include one or more of the following: number and type of attention zones, number of objects in each attention zone, road surface type, road structure type, environment type, weather / environmental conditions, destination, driving mode, vehicle type, powertrain type, and respective locations, speeds, directions, distances from the ego vehicle, and object risks associated with the one or more previously observed and / or newly observed objects.

[0014] In one embodiment, the method further comprises backpropagating a respective node risk associated with a high-risk episodic storage node to one or more episodic storage nodes upstream of the high-risk episodic storage node, wherein the associated node risk is greater than a predetermined risk threshold. The backpropagation of the associated node risk uses a linear or a logistic function.

[0015] In one embodiment, when a new episodic memory node is added to the episodic memory structure, the new episodic memory node is added as a child node to one or more parent nodes, each parent node being a previously existing episodic memory node, and the new episodic memory node is assigned an initial probability value. Each episodic memory node also has a corresponding node reward.

[0016] A predictive navigation control system according to the invention for an ego-vehicle comprises: a data input module, an episodic storage module, and a prediction module. The data input module is configured to receive inputs relating to a clue node, wherein the clue node represents a new event associated with a set of respective locations, velocities, and directions associated with one or more newly observed objects, each located within one of several newly defined attention zones around the ego-vehicle.The data input module is further configured to: (i) calculate a respective individual risk for each of the one or more newly observed objects using a sigmoidal function applied to a respective distance between each respective newly observed object and the ego vehicle; (ii) determine a respective overall risk for each of the plurality of newly defined attention zones, based on the respective individual risks of the one or more newly observed objects located within each respective newly defined attention zone; and (iii) define the clue node as a grouping of the plurality of newly defined attention zones, organized according to their respective overall risks.

[0017] In one embodiment, the episodic storage module is operationally connected to the data acquisition module and contains an episodic storage structure comprising a selectively interconnected and directed network of episodic storage nodes, wherein each episodic storage node represents a respective previously existing event, wherein each previously existing event is associated with a respective set of locations, velocities, and directions associated with one or more previously observed objects, each located within one of several previously defined attention zones around the ego-vehicle, and wherein each episodic storage node has a respective node risk and a respective associated probability.The episodic storage module is configured to: (x) compare the hint node with each of the episodic storage nodes; (y) determine which of the episodic storage nodes has the smallest respective difference metric, thereby defining a best-matching episodic storage node, with each respective difference metric being determined based on a respective total difference between one or more respective properties of the respective episodic storage node and the hint node; and (z) consolidate the hint node with the best-matching episodic storage node if the smallest respective difference metric is less than a predetermined match tolerance, or add a new episodic storage node matching the hint node to the episodic storage structure if the smallest respective difference metric is greater than or equal to the predetermined match tolerance.

[0018] In one embodiment, the prediction module is operationally connected to the episodic storage module and is configured to identify one or both of the following: a most probable next episodic storage node among one or more episodic storage nodes immediately downstream of the best-matching or new episodic storage node, wherein the most probable next episodic storage node has the highest probability among the immediately downstream episodic storage nodes; and a most risky next episodic storage node among the one or more episodic storage nodes immediately downstream of the best-matching or new episodic storage node, wherein the most risky next episodic storage node has the highest node risk among the immediately downstream episodic storage nodes.

[0019] In one embodiment, the controller further comprises a perception module that is operationally connected to the data input module and configured to: capture as perceived data the set of respective locations, velocities, and directions associated with the one or more newly observed objects; and convert the perceived data into inputs relating to the indicator node to be received by the data input module.

[0020] In one embodiment, the respective node risk of each episodic storage node can be a respective maximum, an average, or an aggregate of the respective object risks for the one or more previously observed objects in the associated previously existing event, and the respective object risk for each previously observed object is determined by the sigmoidal function applied to a respective distance between the previously observed object and the ego vehicle. The episodic storage module can further be configured to forward a respective node risk associated with a high-risk episodic storage node to one or more episodic storage nodes upstream of the high-risk episodic storage node, wherein the associated node risk is greater than a predetermined risk threshold. Fig. 1 is a perspective top view of an ego vehicle and other vehicles on a five-lane highway, representing a new event. Fig. Figure 2 is a schematic top view of the vehicles and the highway of Fig. 1, which shows selected attention zones. Fig. Figure 3 is a diagram showing a hint node compared to an episodic memory node using a common event structure for a predictive navigation control procedure. Fig. Figure 4 is a diagram showing a hint node and an episodic storage structure. Fig. 5 is a perspective top view of an ego vehicle and other vehicles on a three-lane highway, representing a previously existing event. Fig. Figure 6 is a schematic top view of the vehicles and the highway of Fig. 5, which shows selected attention zones. Fig. Figure 7 is a flowchart for a predictive navigation control procedure. Fig. Figure 8 is a block diagram of a system / control for implementing the predictive navigation control method. Fig. Figure 9 is a block diagram of a system / control for implementing a procedure to optimize a decision for a decision-maker. Fig. Figure 10 is a flowchart for the process of optimizing a decision for a decision-maker. Fig. Figure 11 is a diagram with a decision-maker and various stimuli. Fig. 12 is a diagram of the decision-maker and the stimuli of Fig. 11, which shows selected attention zones. Fig. Figure 13 is a diagram showing a clue node compared to an episodic memory node that uses a common event structure, for a procedure to optimize a decision for a decision-maker. Fig. Figures 14-16 are block diagrams of various properties, aspects and characteristics or goals.

[0021] With reference to the figures, in which the same reference numbers denote the same parts in the different views, a method 100 for predictive navigation control for an ego-vehicle 20, a method 300 for optimizing a decision for a decision-maker 50, and systems / controls 230, 430 for implementing the respective methods 100, 300 are shown and described here. It should be noted that certain reference numbers in the drawings are shown as subscripts, such as the three attention zones 32. L , 32 H and 32 U in Fig. 2-3. In the figures and in the present description, subscript numbers are used to refer to individual elements (such as the aforementioned zones of attention), whereas the use of reference numbers without subscript numbers can refer to the collective group of such elements and / or to a single, but generic, element of these elements. Thus, reference number 32 refers to L to a specific attention zone, whereas the reference number 32 (without the subscript) can refer to all attention zones, the group of attention zones, or a single but generic attention zone (i.e., any attention zone).

[0022] It should be noted that, while Fig. Figures 1-2 and 5-6 illustrate exemplary traffic situations involving passenger cars. Procedures 100, 300 and systems / controls 230, 430 of this description also apply to other types of autonomous and semi-autonomous vehicles and environments, such as drones and aircraft flying in airspace, ground vehicles traveling over roads or terrain, mobile robots moving over surfaces and / or through structures, surface watercraft moving over the surface of a body of water, underwater vehicles moving over the surface of a body of water or at various depths below the surface of the water, satellites or spacecraft moving through space, etc.It should also be noted that although the drawings and description use a straight five-lane or three-lane motorway section, procedures 100, 300 and systems / controls 230, 430 also apply to other road and carriageway configurations, including intersections, converging / diverging lanes, curved carriageways, parking lots, open terrain, etc.

[0023] Fig. Figures 1-6 illustrate a process in which a given event 30, such as a traffic situation, involves an ego-vehicle 20 ( Fig. 1) an arrangement of attention zones 32 around the ego vehicle 20 with other adjacent vehicles / objects placed within these attention zones 32 is identified ( Fig. 2), organizing the collection of attention zones 32 and other data according to an event structure 39 ( Fig. 3) and representing this collection of information as a reference node 34, which is linked to individual episodic storage nodes 36 in an episodic storage structure 38 ( Fig. 4) can be compared, in which each episodic storage node 36 represents a respective previously existing event 40 ( Fig. 5-6). (Hereinafter, the abbreviation "EM" is used for the term "episodic memory". Note also that "around" Ego-Vehicle 20 means in front of, behind, beside, and / or next to Ego-Vehicle 20, and "Ego-Vehicle 20" can refer to (i) the current host vehicle experiencing a new event 30 that is converted into a clue node 34, (ii) the current host vehicle that previously experienced events 40 that were converted into EM nodes 36, and / or (iii) other host vehicles that experienced previous events 40 that were converted into EM nodes 36 and imported into the current host vehicle.If the indicator node 34 sufficiently matches one of the EM nodes 36 in the EM structure 38, the indicator node 34 can be merged or consolidated with the matching EM node 36; however, if the indicator node 34 does not sufficiently match any of the EM nodes 36, the indicator node 34 can be added to the EM structure 38 as a new EM node 36. This process is now described in more detail below.

[0024] Fig. Figure 1 shows a perspective top view of an ego-vehicle 20 and other vehicles 21, 22, 23, 24, 25 on a five-lane highway. The ego-vehicle 20 is shown in the fourth lane 27 from the left, and another vehicle 21 (sometimes referred to here as the "neighbor vehicle" 21) is shown in the fifth lane 28 from the left, which is next to the right rear of the ego-vehicle 20. Another vehicle 22 (sometimes referred to here as the "leading vehicle" 22) is shown directly in front of the ego-vehicle 20 in the same fourth lane 27 as the ego-vehicle 20. This grouping of vehicles 21, 22, 23, 24, 25 around the ego-vehicle 20, their respective locations L, speeds S, and directions H relative to the ego-vehicle 20, and the situation of driving along a straight section of highway, as in Figure 1, are shown. Fig. As shown in 1, together they provide a “snapshot” of a particular driving situation or event, which can be referred to as a new event 30.

[0025] Fig. Figure 2 shows a schematic top view of the ego-vehicle 20, the neighboring vehicle 21, and the preceding vehicle 22 within selected attention zones 32. Vehicles / objects located outside the attention zones 32, such as vehicle 23 (in lane 26) and vehicles 24 and 25, are not shown or considered in this step. (Note that the terms "vehicle," "object," and "vehicle / object" can be used interchangeably here. The plural forms of these terms can also be used interchangeably.) An attention zone 32, in the context of an ego-vehicle 20 on a highway, is a road segment with a predefined shape, size, and location relative to the ego-vehicle 20. For example, as in Fig. As shown in Figure 2, each attention zone 32 is generally a rectangular space one lane wide and approximately three car lengths long. The size and shape of the attention zones 32 may be fixed, or the size and shape may vary depending on factors such as the speed of the ego vehicle, the type of road surface or terrain, weather conditions, lane curvature, etc. Note in Fig. 2 also that only some parts of the motorway were designated as attention zones 32; in particular, there is one attention zone 32 e , in which the ego vehicle 20 is centered, an attention zone 32 L , 32 r on each side of the Ego vehicle 20, a series of three attention zones 32 b1 , 32 b2 , 32 b3 immediately behind the Ego vehicle 20 and two rows of three attention zones 32 a1 , 32 a2 , 32 a3 , 32 a4 , 32 a5, 32 a6 immediately in front of the ego-vehicle 20. It should be noted that any number and arrangement of attention zones 32 can be used. It should also be noted that the arrangement of attention zones 32 around the ego-vehicle 20 dynamically moves with the ego-vehicle 20, and that the arrangement of attention zones 32 can optionally change if the ego-vehicle 20 changes speed, changes course, changes lanes, changes road surfaces, changes road conditions, experiences a change in weather conditions, etc.

[0026] The attention zones 32 can optionally be divided into two or more types, such as low-risk zones 32. L , High-risk zones 32 H and urgent risk zones 32 U For example, the attention zone is 32 rimmediately to the right of the Ego vehicle 20, represented by a wide cross-hatching pattern, this attention zone 32 r as high-risk zone 32 H indicates (due to the immediate proximity of the neighboring vehicle 22). The attention zone 32 a5 The area directly in front of the Ego vehicle 20 is depicted with a narrow cross-hatching pattern, which defines this attention zone 32. a5 as Urgent Danger Zone 32 U indicates (because of the possibility that the vehicle ahead 22 will brake and thus impede the movement of the ego vehicle 20). Attention zone 32 e of the Ego vehicle and the other nine attention zones 32 L , 32 b1 , 32 b2 , 32 b3 , 32 a1 , 32 a2 , 32 a3 , 32 a4 , 32 a6 are shown hatched in light grey, which indicates that this is a low-risk zone 32 L characterizes.

[0027] An attention zone 32 can be considered a low-risk zone 32 L , High-risk zone 32 H or urgent risk zone 32 U can be characterized based on various risk factors, such as the respective locations L, speeds S, and directions H of the one or more vehicles / objects located in the attention zone 32. These properties of location L, speed S, and direction H can be considered as influencing the proximity of each vehicle / object to the ego-vehicle 20 and the potential of each vehicle / object to disrupt or impair the forward movement of the ego-vehicle 20. For example, the distance d 21 between the ego vehicle 20 and the adjacent vehicle 21 and the position of the adjacent vehicle 21 to the side and behind the ego vehicle 20 result in the attention zone 32 b3 as a high-risk zone 32 His characterized, while the distance d 22 between the ego vehicle 20 and the leading vehicle 22, and the position of the leading vehicle 22 directly in front of the ego vehicle 20, can lead to the attention zone 32 a5 as a zone with urgent risk 32 U is characterized.

[0028] Once the attention zones 32 around the ego vehicle 20 have been identified, and each attention zone 32 optionally designated either as a low-risk area 32 L , as a high-risk area 32 H or as an area of ​​urgent risk 32 UOnce classified, the respective locations L, velocities S, and directions H of the vehicles / objects within each attention zone 32 can be determined relative to the location, velocity, and direction of the ego-vehicle 20, and these relative locations L, relative velocities S, and relative directions H can be stored in an event structure 39, as in Fig. 3 shown. The event structure 39 can be a data structure, an ordered record, a table, an organized grouping, a format, etc., in which the properties 49 of location L, speed S and direction H (and other optional data such as distance d and individual risk IR) are stored for each vehicle / object.

[0029] As in Fig. As shown in Figure 3, event 30 (e.g., the traffic situation of) Fig. 1-2) connected with three attention zones 32 - a low-risk area 32 L, a high-risk area 32 H and an area with urgent risk 32 U (Note that in Fig. 8 For the sake of simplicity, only one low-risk zone 32 L As shown, there could also be nine low-risk zones. 32 L depicted as in Fig. 2.) The elements that make up the low-risk zone 32 L Areas assigned to these zones are shown with dashed lines to indicate that none of the low-risk zones contain 32 L Vehicles / objects are located in high-risk zone 32. H is represented by a vehicle 21, which has a location L 21 , a speed S21 and a direction of travel H 21 has, and the urgent risk zone 32 U is also shown with a vehicle 22, which has a location L 22 , a speed S22 and a direction of travel H 22It should be noted that, while the adjacent vehicle 21 and the leading vehicle 22 are in Fig. 3 are shown (each represented as a hexagon), the event structure 39 can optionally exclude the identification of these individual vehicles 21, 22 and instead the respective locations L 21 , L 22 , speeds S21, S22 and directions H 21 , H 22 These vehicles 21, 22, and other related information such as their respective distances d 21 , d 22 and individual risks IR 21 , IR 22 may contain.

[0030] The event structure 39 can also include a distance d and an individual risk IR for each vehicle / object, where the distance d represents the distance between the vehicle / object and the ego-vehicle 20, and the individual risk IR for each vehicle / object can be determined by a sigmoidal function applied to the distance d (e.g., as in Fig. 2 shown, is d 21 the distance between the adjacent vehicle 21 and the Ego vehicle 20, and d 22 is the distance between the vehicle 22 ahead and the Ego vehicle 20, and the respective individual risks IR 21 , IR 22 can be done using these distances d 21 , d 22 and determined by a suitable sigmoidal equation, an example of which is given below).

[0031] As in Fig. As shown in Figure 14, the properties stored in the event structure 39 can also contain additional information AI, such as: the number and type of attention zones 32 (Al AZ ) (e.g., a grid of twelve zones 42, which as in Fig. 2 and Fig. 5 are arranged and use three zone types 42, which correspond to the low, high and urgency zones 32 discussed above. L 42 L , 32 H , 42H, 32 U 42U are similar); the number of vehicles / objects in each attention zone 32 (AI NO ); the road surface type (AI RS ) (e.g. paved, gravel, dirt, etc.); the road structure type (AI) RT ) (e.g., the number of lanes, the orientation of the lanes, etc.); the environment type (AI) E ) (e.g., city, residential area, highway, parking lot, etc.); the weather / environmental conditions (AI WC ) (e.g., clear, light rain, heavy rain, snow, etc.); the destination (Al DG ) (e.g., fastest route, leisure / scene route, etc.); Driving mode (AI) DM ) (e.g., cruising, sport, fuel saving, etc.); vehicle type (Al VT ) (e.g. passenger car, van, permissible total weight, horsepower, etc.); and powertrain type (AI PT ) (e.g., automatic, manual, electric, hybrid, number of gears, etc.). As in Fig. As shown in 3, for example the reference number AI can be 21AI30 can represent one or more additional pieces of information AI relating to the neighboring vehicle 21, and AI30 can represent one or more additional pieces of information AI relating to the event 30. This organized collection of information, stored in the event structure 39 format, can serve as a numerical representation of the event 30 and can be represented as a note node 34, as shown in Fig. 3 shown. Note that Fig. Figure 3 also shows that the event structure 39 of the hint node 34 and its associated new event 30 is also applied to an EM node 36 and its associated, previously existing event 40, and that the additional information AI is also applied to both events 30, 40 (i.e., Al 30 , AI 40 ).

[0032] A system or process using the approach described here can begin by building an EM structure 38 from scratch or by receiving or importing a previously created EM structure 38, which may have been created by another ego-vehicle that has similarities to the ego-vehicle 20 at hand (i.e., similar vehicle type, similar powertrains, etc.). In either case, the system and approach can employ a process of capturing or receiving information about each new event 30 ( Fig. 1) of identifying an arrangement of attention zones 32 around the ego vehicle 20 with other adjacent vehicles / objects that are placed within these attention zones 32 ( Fig. 2), of organizing information about the attention zones 32 and the vehicles / objects therein in the format of an event structure 39 ( Fig. 3) and the conversion of the organized information into a hint node 34, which can numerically represent the new event 30. The hint node 34 can then be linked to individual EM nodes 36 in the EM structure 38 ( Fig. 4) are compared, where each EM node 36 represents a previously existing event 40. Regardless of whether the system and approach build the EM structure 38 from scratch or start with an imported EM structure 38, the system and approach can further expand the EM structure 38 over time as new events 30 are encountered, translated into indicator nodes 36, and compared with the nodes 36 in the existing EM structure 38.

[0033] This comparison between the hint node 34 and each EM node 36 can be facilitated if each EM node 36 uses the same event structure 39 as the hint node 34, as in Fig. Figure 3 illustrates this. Here, the exemplary EM node 36 represents a corresponding, previously existing event 40 with previously observed vehicles / objects, as shown in Fig. 4 shown. Fig. 4. The ego-vehicle 20 is driving in the middle lane 47 of a three-lane highway. A neighboring vehicle 43 is next to the ego-vehicle 20 in the right lane 48, and two vehicles 44 and 45 are ahead of the ego-vehicle 20 in the middle lane 47. No vehicles / objects are shown in the left lane 46. This traffic situation is similar to that of Fig. 1, regarding the general positions of other neighboring vehicles / objects around the Ego vehicle 20.

[0034] Fig. Figure 5 shows a schematic representation of the in Fig. The event 40 depicted in section 40 shows the attention zones 42. It should be noted that the vehicle 45 is in Fig. 5 is not included because it is not located in any of the defined attention zones 42. The attention zone 42, which contains the adjacent vehicle 43 (i.e., attention zone 42) r ), is shown with a wide cross-hatching to mark it as a high-risk zone 42H, and the attention zone 42, which contains the vehicle ahead 44 (i.e., attention zone 42) a5 ), is shown with a close cross-hatching to indicate it as an urgent risk zone 42 U to mark. In addition to attention zone 42, which contains the Ego vehicle 20 (i.e. attention zone 42). e ), the other nine attention zones are 42 (i.e., 421, 42) b1 42 b2 42 b3 42 a1 42 a2 42 a3 42 a4 and 42 a6 ) shown in light grey hatching to indicate low-risk zones 42 L to mark. Even if the distance d 44from Ego vehicle 20 to the preceding vehicle 44 ( Fig. 5) is slightly longer than the distance d 22 from the ego vehicle 20 to the preceding vehicle 22 ( Fig. 2) and the distance d 43 The distance to the neighboring vehicle 43 is slightly shorter than the distance d 21 Regarding the neighboring vehicle 21, the attention field diagrams of the Fig. 2 and Fig. 5 nevertheless, are generally very similar to each other.

[0035] Each of the two previously observed vehicles 43, 44 within the attention zones 42 has a respective location L 43 , L 44 , a speed S 43 , S 44 and a course H 43 , H 44 , which are stored in event structure 39, as well as a respective distance d 43 , d 44 to the Ego vehicle 20 and a respective object risk OR 43 , OR 44 based on the respective distance d 43 , d 44The EM node 36 can also have a node risk NR and a probability LL assigned to it. The node risk NR represents the degree of risk that the associated, previously existing event 40 poses to the journey of the ego vehicle 20 (e.g., due to the proximity of neighboring vehicles, etc.), and the probability LL represents how likely it is that the event 40 assigned to node 36 will occur (e.g., expressed as a decimal number from 0.00 to 1.00, representing the probability LL as a percentage). Each EM node 36 can optionally also have a corresponding node reward NW, which is displayed in Fig. 3 is shown with a dashed line. The node reward NW can be a numerical representation of one or more rewards or benefits that can be realized when the associated event 40 occurs.

[0036] It should be noted that the example EM node 36 of Fig. 3 has a very similar content to the indicator node 34. That is, the event 30, represented by indicator node 34, involves a vehicle 21 in a high-risk zone 32. H and a vehicle 22 in an urgent risk zone 32 U , and likewise, event 40, represented by the exemplary EM node 36, has a vehicle 43 in a high-risk zone 42 H and a vehicle 44 in an urgent risk zone 42 U For illustrative purposes, it can also be assumed that the location L 21 , the speed S 21 and the direction of travel H 21 of the neighboring vehicle 21 to the location L 43 , the speed S 43 and the direction of travel H 43 is very similar to vehicle 43 (so that the distance d 21 and the individual risk IR 21 the distance d 43 or the object risk OR 43 are very similar); and that the place L22 , the speed S 22 and the course H 22 of the leading vehicle 22 to the location L 44 , the speed S 44 and the course H 44 is very similar to vehicle 44 (thereby the distance d 22 and the individual risk IR 22 the distance d 44 or the object risk OR 44 are very similar). Additionally, and as described below, it can be assumed that the exemplary EM node 36 of Fig. 3 the EM knot 361 from Fig. 4 represents, and that EM node 361 shows the best match with indicator node 34 among all EM nodes 36 within the EM structure 38 of Fig. 4 is. However, it should be noted that while the contents and values ​​of the example node 36 are from Fig. 3 (and of EM node 361 from Fig. 4) closely match the contents and values ​​of the hint node 34, while the contents and values ​​of other EM nodes 36 may not match the hint node 34 as closely. For example, the EM nodes 36 in Fig. 4 with the exception of node 361, each has a different number of vehicles / objects in its respective three attention / risk zones 42 L 42 H 42 U have, with different locations L, speeds S and directions H of them (and thus different distances d and different object risks OR).

[0037] As in Fig. As shown in Figure 6, the EM structure 38 is a selectively interconnected and directed network of EM nodes 36. (Here, each EM node 36 has been given a unique subscript index from 0 to 22; i.e., 360 to 366) 22Each EM node 36 represents a unique, previously existing event 40, and the collection of EM nodes 36 represents an episode or situation that may include various possible sequences of one or more of these events 40. The EM nodes 36 are selectively connected to one another in a directed manner, as indicated by the arrows in Fig. 4 indicated. The direction of an arrow between two EM nodes 36 shows the possible temporal flow direction between these EM nodes 36. For example, each of the arrows that go from EM node 361 to EM nodes 360, 362, 367, 361 indicates the direction of flow between these EM nodes 361 and 361. 10 and 36 11 indicate that, after the event 40 represented by EM node 301 has occurred, a possible next event 40 is represented by one of the EM nodes 360, 362, 367, 36 10 and 36 11The direction of each arrow indicates a possible flow from an "upstream" EM node 36 to a "downstream" EM node 36, as well as from a "parent" EM node 36 to a "child" EM node 36. Thus, EM node 361 is located upstream of EM nodes 360, 362, 367, and 36. 10 and 36 11 , and each of the EM nodes 360, 362, 367, 36 10 and 36 11 lies downstream of 361; likewise, EM node 361 would be a parent node for each of EM nodes 360, 362, 367, 36 10 and 36 11 , and the EM nodes 360, 362, 367, 36 10 and 36 11 These would be child nodes for EM node 361. It should be noted that an upward-pointing arrow in the diagram can still indicate a downstream flow direction (e.g., the arrow from EM node 36). 11(pointing to EM node 360); thus, the directions of the arrows - and not necessarily the position of EM node 36 on the page - determine the downstream / upstream flow direction from node to node.

[0038] As mentioned above, an EM structure 38 represents a collection of possible event sequences 40 for a specific episode or situation. In the context of motor vehicles, each episode could, for example, be a specific driving situation, such as driving, overtaking, turning off the highway at an exit, parking, reversing, etc. Each of these driving situations or episodes can have its own unique EM structure 38, with each EM structure 38 containing a collection of possible event sequences 40. Thus, a system or process using the approach described here can have access to several different stored EM structures 38, which can be used selectively and switched between to match the respective driving situation.

[0039] The solid arrow in Fig. Figure 3 shows that the indicator node 34 is compared to a single EM node 36, and the five solid arrows in Fig. Figure 4 shows that the indicator node 34 is compared with five specific EM nodes 36 in an EM structure 38. (Although in Fig. (4) Only five comparisons are shown; this is done for the sake of simplicity, since in practice the cue node 34 would be compared with each of the EM nodes 36. Each of the EM nodes 36 can use the event structure 39 format to represent its respective attention zones 42 and various event-related information, which can facilitate the comparison of the cue node 34 with each EM node 36.

[0040] As mentioned above, if the indicator node 34 matches sufficiently with one or more of the EM nodes 36, it can be merged or consolidated into the EM node 361 with the best match; however, if the indicator node 34 does not match sufficiently with any of the EM nodes 36, the indicator node 34 of the EM structure 38 can be designated as a new EM node 36. 22to be added. (Note that reference numbers 361 and 36 22 in Fig. Although 4 can be used to represent two specific EM nodes 36, these reference numbers are also used here to generally identify the most suitable node 361 or a new node 36. 22 (to represent). For example, if, as already mentioned, it is found that the contents and values ​​of EM node 361 best match the contents and values ​​of clue node 34, and if this match is close enough (e.g., if, however, the match between node 361 and clue node 34 is not close enough, clue node 34 can be added to EM structure 38 as a new node, e.g., as node 36). 22 , which is at the EM node 36 20 was attached and is shown as a dashed outline. As soon as the reference node 34 is consolidated into a best-fitting node 361 or as a new node 3622 Since the node was added, it is assumed that the best matching or new node is 361, 36. 22 (depending on the case) the node from which the most probable and / or riskiest nodes 362, 36 10 can be determined.

[0041] For each EM node 36, a corresponding difference metric can be determined, based on a corresponding total difference between one or more properties 49 of the EM node 36 and the same one or more properties 49 of the indicator node 34. Once the difference metrics for all EM nodes 36 have been determined, the EM node 36 with the smallest difference metric can be identified as the best-matching EM node 361. If the difference metric of the best-matching EM node 361 is smaller than a predefined matching tolerance, the indicator node 34 is next consolidated into the best-matching node 361; otherwise, the indicator node 34 of the EM structure 38 becomes the new node 36. 22 added.

[0042] If the clue node 34 is merged or consolidated into a best-match EM node 361, the merger or consolidation may optionally have an effect on the best-match EM node 361. For example, the consolidation may increase or boost the probability LL of the best-match EM node 361. On the other hand, if clue node 34 is merged or consolidated into a new EM node 361, the probability LL of the best-match EM node 361 may increase or decrease. 22 is added as a child node to EM node 36, which has the closest match to it, which is in Fig. 4 as EM node 36 20 is assumed. (That is, the new EM node 36 22 can be added as a child node to one or more parent nodes, where each parent node is a previously existing EM node 36. Note that the new node 36 22 Although shown here without child nodes, in some cases a new node 36 22can be inserted between two or more EM nodes 36 (and connected to them), in which case the new node 36 22 would have one or more child knots.

[0043] To describe the above approach in detail, a method 100 for predictive navigation control for an ego-vehicle 20 according to one embodiment (and as shown by the flowchart in Fig. Figure 7 illustrates a comparison step (block 160), a determination step (block 170), a consolidation step (block 200), and an identification step (block 220). In block 160, the comparison step involves comparing a clue node 34 with each of a plurality of EM nodes 36 in an EM structure 38, where the clue node 34 represents a new event 30. The new event 30 is associated with a set of respective locations L, velocities S, and directions H, which in turn are associated with one or more newly observed objects 21. (Note that the reference number 21 in the Fig. (1-2, although used to represent the neighboring vehicle 21, this reference number is also used here to represent one or more newly observed objects 21.) Each newly observed object 21 is located within a corresponding multitude of newly defined attention zones 32 around the ego-vehicle 20. The EM structure 38 comprises a selectively interconnected and directed network of EM nodes 36, each EM node 36 representing a respective previously existing event 40. Each previously existing event 40 is associated with a respective set of locations L, velocities S, and directions H, which are associated with one or more previously observed objects 43. (Note that the reference number 43 in the Fig. Although reference numbers 4-5 are used to represent a specific vehicle / object 43, this reference number is also used here to represent one or more previously observed objects 43. Each previously observed object 43 is located within one of several previously defined attention zones 42 around the ego-vehicle 20. Additionally, each EM node 36 has a corresponding node risk NR and a corresponding probability LL (and optionally a corresponding node reward NW) associated with it.

[0044] In block 170, the determination step involves identifying which of the several EM nodes 36 has the smallest respective difference metric, thereby defining the best-fitting EM node 361. Each respective difference metric is determined based on a respective total difference between one or more respective properties 49 of the respective EM node 36 and the indicator node 34. As in Fig. As shown in Figure 14, these properties can include one or more of the following: number and type of attention zones 32, 42 (AI AZ ), Number of objects in each attention zone 32, 42 (AI NO ), road surface type (AI RS ), street structure type (AI RT ), environment type (AI E ), Weather / Environmental Conditions (AIwc), Destination (AI DG ), Driving mode (AI DM ), Vehicle type (AI VT ), Powertrain type (AI PT ) and the respective locations L, speeds S, directions H, distances d from the ego vehicle 20 and person / object risks IR, OR associated with the one or more previously observed and / or newly observed objects 43, 21.

[0045] Each of the respective positions L, velocities S, and directions H of the one or more newly observed objects 21 and of the one or more previously observed objects 43 can be defined with respect to the ego-vehicle 20. (That is, the ego-vehicle 20 can serve as the origin of a reference coordinate axis at which the positions L, velocities S, and directions H can be measured.) The total difference can be a sum, an average, or a maximum of the differences between or among these properties 49. When using the in Fig. 1-2 depicted event 30 and the event structure of Fig. 3. The total difference can, for example, be a sum of the differences between the corresponding locations L of the vehicles / objects, i.e., (L 44 - L 22 ) + (L 43 - L 21), where each of the locations L is decomposed into its x and y components, based on a coordinate system whose origin is at the Ego vehicle 20. Additionally, each respective total difference can be a respective sum of one or more weighted penalties assigned against each of one or more differences between the respective properties 49 of the respective EM node 36 and the clue node 34.

[0046] Block 190 checks whether the smallest respective difference metric (between the pointer node 34 and the best-matching node 361) is smaller than a predefined matching tolerance. If so, block 200 executes the consolidation step, in which pointer node 34 is consolidated with the best-matching EM node 361; however, if this is not the case (i.e., the smallest respective difference metric is greater than or equal to the predefined matching tolerance), a new EM node 36 is created in block 210. 22 , corresponding to indicator node 34, added to EM structure 38. The new EM node 36 22 An initial probability value can be assigned (i.e., an initial probability value LL of the new node); optionally, the new node can be assigned 36 22 An initial node risk NR and an initial node reward NW are also assigned.

[0047] In block 220, the identification step includes identifying the most likely next node 362 and / or the riskiest next node 36. 10 (Note that reference numbers 362 and 36 10 in Fig. Although 3 can be used to represent two specific EM nodes 36, these reference numbers are also used here to generally indicate the most likely and riskiest next EM nodes 36. 22 or 36 10 (to represent). If the indicator node 34 has been consolidated to the best-matching node 361, then the most likely next node 362 is the node 36 with the highest probability LL among the nodes 36 immediately downstream of the best-matching node 361, and the riskiest next node 36 10Node 36 has the highest node risk NR among the nodes 36 immediately downstream of the best-matching node 361. If, on the other hand, the indicator node 34 is considered a new node 36 22 was added and if there is one or more node 36 immediately downstream of the new node 36 22 If there is a , then the most probable next node 362 is the node 36 with the highest probability LL among the nodes 36 immediately downstream of the new node 36. 22 , and the riskiest next node 36 10 Is node 36 with the highest node risk NR among nodes 36 immediately downstream of the new node 36 22 If there is no node 36 immediately downstream of the new node 36 22 there is, for example, in the one in Fig. In the case shown in section 4, then no most probable or riskiest next nodes 362, 36 are identified. 10identified. In such a case, however, it is likely that the ego-vehicle 20 will experience new events 30 that differ significantly from the previously existing events 40, and the ego-vehicle 20 will continue to identify new nodes 36. 22 to add to EM structure 38 as the ego-vehicle 20 continues to progress. In this way, EM structure 38 can be expanded to accommodate all new events 30 that the ego-vehicle 20 experiences. An EM structure 38 can continue to grow and become quite complex over time, so that in some cases the most probable next node 362 and the riskiest next node 36 22 the same node 36 can be.

[0048] The respective node risk NR of each EM node 36 can be a respective maximum, an average, or an aggregate of the respective object risks OR for the one or more previously observed objects 43 in the associated previously existing event 40. The respective object risk OR for each previously observed object 43 can be determined by a sigmoidal function applied to a respective distance d 43 between the previously observed object 43 and the Ego vehicle 20 is applied. (Note that the reference number d 43 in Fig. 5 is used to indicate a specific distance d 43 To represent, this reference number is also used here to generally represent the distance d. 43 to represent between a previously observed object 43 and the ego vehicle 20). Fig. 3 and Fig. Figure 5 shows, for example, two previously observed objects 43, 44 with their respective distances d. 43 , d 44Regarding the Ego vehicle 20. Since sigmoidal functions return a value between 0 and 1, where 0 can mean a low or no approach risk and 1 a high risk, let us assume for illustration that the object risk OR 43 for the more closely adjacent vehicle / object 43 0.7 and the object risk OR 44 for the preceding vehicle / object 44 is 0.3. In this case, the node risk NR for the EM node 36 associated with event 40 can be a maximum (0.7), an average (0.5), or an aggregate of these two object risks OR. 43 , OR 44 be. The aggregate can be a sum of these object risks OR 43 , OR 44 (1.0), a sum of the maximum and the average (1.2), a mean between the maximum and the average (0.6), or another suitable combination.

[0049] The procedure 100 may further comprise the following steps in block 130: (i) calculating a respective individual risk IR for each of the one or more newly observed objects 21 using a sigmoidal function applied to a respective distance d between each respective newly observed object 21 and the ego-vehicle 20; (ii) in block 140, determining a respective overall risk TR for each of the plurality of newly defined attention zones 32, based on the respective individual risks IR of the one or more newly observed objects 21 located within each respective newly defined attention zone 32; and (iii) in block 150, defining the indicator node 34 as a grouping of the plurality of newly defined attention zones 32 organized according to their respective overall risks TR.(For example, the grouping of attention zones 32 can be ordered according to one of the following criteria: Low / High / Urgent-Risk zones 32. L , 32 H , 32 U Urgent / High / Low Risk Zones 32 U , 32 H , 32 L ; and in ascending or descending order of the total risk TR.)

[0050] The overall risk TR for each newly defined attention zone 32 can be determined from the individual risks IR of the newly observed object(s) 21 in that attention zone 32. For example, using the Fig. 1-2 depicted new event 30 the overall / total risk TR for zone 32 U , 32 a5 with urgent risk equal to the individual risk IR of the leading vehicle 22, which is in this zone 32 U , 32 a5 is located. If two other vehicles / objects are in this zone 32 L , 32 a5If they were located, the overall risk TR for this urgency risk zone would be 32. U , 32 a5 from the individual risks IR of all three vehicles / objects in this attention zone 32 U , 32 a5 The overall risk TR for a newly defined attention zone 32 can be a maximum, an average, a sum, or an aggregate of the individual risks IR of the vehicles / objects in this zone 32.

[0051] The object risk OR for each previously observed object 43 and / or the individual risk IR for each newly observed object 21 can be determined by the sigmoidal equation R = 2-({1 - 1 / (1 + e^[(mindist -SAFEDIST) / (SAFEDIST / 2)]} - 0.5) + 0.5, where R is the respective object risk OR or the individual risk IR, and mindist is the distance d. 43 , d 21to the previously observed or newly observed object 43, 21 from the ego vehicle 20, and wherein SAFEDIST is a distance that depends on one or more of a road surface type, a road structure type, weather / environmental conditions and a relative lane position, closing speed or closing acceleration between the previously observed or newly observed object 43, 21 and the ego vehicle 20.

[0052] Procedure 100 can further include in block 180 the step of backpropagating a given node risk NR associated with a high-risk EM node 36 to one or more EM nodes 36 upstream of the high-risk EM node 36, wherein the node risk NR associated with the high-risk EM node 36 is greater than a predetermined risk threshold. This backpropagation of the associated node risk NR can use a linear or a logistic function. For example, if a given EM node 36 has a node risk NR that is above the predetermined risk threshold, then this node can be considered a high-risk node 36, and the node risk NR for this high-risk node 36 can be propagated backproportionally to one or more other nodes 36 immediately upstream of the high-risk node 36.

[0053] The procedure 100 may further include in block 110 the step of acquiring, as perceived data, the set of the respective locations L, velocities S and directions H associated with the one or more newly observed objects 21, and in block 120 the step of converting the perceived data into inputs, signals or data relating to the indicator node 34, for example for reception by a data input module 260 (described in detail below).

[0054] Fig. Figure 8 shows a block diagram of a system / controller 230 for predictive navigation control for an ego vehicle 20, configured to perform some or all of the steps of the procedure 100 mentioned above. The system / controller 230 can perform all of the steps described in Figure 8. Fig. The system may contain the 8 blocks or modules shown, or it may contain a subset of these blocks or modules. Each of these blocks or modules may be implemented in the form of hardware, software, and / or firmware, with two or more of the blocks or modules integrated together (e.g., as part of a single hardware unit) and one or more other blocks or modules implemented separately.

[0055] The system / controller 230 comprises a data input module 260, an EM module 270, and a prediction module 280. The data input module 260 is configured to receive inputs relating to a hint node 34, where the hint node 34 represents a new event 30 associated with a set of respective locations L, velocities S, and directions H associated with one or more newly observed objects 21, each newly observed object 21 being located within one of a plurality of newly defined attention zones 32 around the ego-vehicle 20.The data input module 260 is further configured to: (i) calculate a respective individual risk IR for each of the one or more newly observed objects 21 using a sigmoidal function applied to a respective distance d between each respective newly observed object 21 and the ego vehicle 20; (ii) determine a respective overall risk TR for each of the plurality of newly defined attention zones 32 based on the respective individual risks IR of the one or more newly observed objects 21 located within each respective newly defined attention zone 32; and (iii) define the indicator node 34 as a grouping of the plurality of newly defined attention zones 32 organized according to their respective overall risks TR.

[0056] The EM module 270 is operationally connected to the data input module 260 and contains an EM structure 38 with a selectively interconnected and directed network of EM nodes 36, wherein each EM node 36 represents a respective previously existing event 40, wherein each previously existing event 40 is associated with a respective set of locations L, velocities S and directions H, which are associated with one or more previously observed objects 43, each of which is located within one of several previously defined attention zones 42 around the ego vehicle 20, and wherein each EM node 36 has a respective node risk NR and a respective probability LL associated with it.The EM module 270 is configured to: (x) compare the antecedent node 34 with each of the EM nodes 36; (y) determine which of the EM nodes 36 has the smallest respective difference metric, thereby defining a best-matching EM node 361, where each respective difference metric is determined based on a respective total difference between one or more respective properties 49 of the respective EM node 36 and the antecedent node 34; and (z) consolidate the antecedent node 34 with the best-matching EM node 361 if the smallest respective difference metric is less than a predetermined matching tolerance, otherwise add a new EM node 3622 corresponding to the antecedent node 34 to the EM structure 38 if the smallest respective difference metric is greater than or equal to the predetermined matching tolerance.

[0057] The prediction module 280 is operationally connected to the EM module 270 and is configured to identify one or both: a most probable next node 362 among one or more EM nodes 36 immediately downstream of the best-matching or new EM node 361, 3622, wherein the most probable next node 362 has a highest probability LL among the immediately downstream EM nodes 36; and a most risky next node 3610 among the one or more EM nodes 36 immediately downstream of the best-matching or new EM node 361, 3622, wherein the most risky next node 3610 has a highest node risk NR among the immediately downstream EM nodes 36.

[0058] The system / controller 230 may further include a perception module 250, which is operationally connected to the data input module 260 and configured to: recognize as perceived data the set of respective locations L, velocities S, and directions H associated with the one or more newly observed objects 21; and convert the perceived data into inputs relating to the indicator node 34 for reception by the data input module 260. The system / controller 230 may additionally include one or more sensors 240 (e.g., radar, lidar, ultrasound, infrared, temperature, etc.) that are operationally connected to the perception module 250 and provide input signals to it. Furthermore, the prediction module 280 may be operationally connected to other modules / devices 290 located outside the system / controller 230.

[0059] The respective node risk NR of each EM node 36 can be a respective maximum, an average, or an aggregate of respective object risks OR for the one or more previously observed objects 43 in the associated previously existing event 40, and the respective object risk OR for each previously observed object 43 can be determined by a sigmoidal function applied to a respective distance d 43 between the previously observed object 43 and the ego vehicle 20. The EM module 270 can further be configured to forward a respective node risk NR, which is assigned to a high-risk EM node 36, to one or more EM nodes 36 upstream of the high-risk EM node 36, where the assigned node risk NR is greater than a predetermined risk threshold.

[0060] While the approach of the present description, as in the Fig. As shown in Figures 1-8, and described in relation to predictive navigation control for an ego-vehicle 20, the approach can also extend to optimizing a decision for a decision-maker 50. For example, Figure 1-8 shows that the approach can also be extended to optimizing a decision for a decision-maker 50. Fig. 10. A flowchart for a procedure 300 for optimizing a decision for a decision-maker 50, wherein the procedure 300 comprises a comparison step (block 360), a determination step (block 370), a consolidation step (block 400), and an identification step (block 420). The procedure 300 can also be described with reference to the Fig. 11-13 are described, whereby Fig. 11 shows a decision-maker 50 and various stimuli 51, 52, 53, 54, 55, 56 that are associated with a new event 30, Fig. 12 the decision-maker 50 and selected stimuli 51, 52, 53 together with selected attention zones 32 UL , 32 UR , 32 LL , 32 LR shows, and Fig. Figure 13 shows a hint node 34, which is compared to an EM node 36 using a common event structure 39. (Note that Fig. 4, which shows a reference node 34, which is compared with the EM nodes 36 in an EM structure 38, also applies to the procedure 300).

[0061] The decision-maker 50 can be a system or device configured to make decisions, such as a process controller, control module, or similar. The decision-maker 50 can use the method 300 to guide, support, or optimize the decision-maker's decision-making process. The stimuli 51-56 can be information relating to the new event 30, and the stimuli 51-56 can be visualized or represented as individual points or bits of information in an information space. The information space can be a two-dimensional, three-dimensional, or higher-dimensional construct with multiple axes or dimensions. As in the Fig. As shown in Figures 11-12, the information space can, for example, be visualized or represented as a two-dimensional area with two orthogonal axes intersecting at an origin where the decision-maker 50 can be visualized or represented. Each of these axes can be used to measure a particular property, quality, or characteristic. For example, the vertical axis can measure the level of difficulty and the horizontal axis the cost, with the origin located at a nominal or predetermined combination of difficulty and cost. The collection of stimuli 51-56 represents various pieces of information about the event 30 as observed in a current state 31, and the event 30 can be any situation that involves or requires an action or a decision.The current state 31 can be a current point in time and can include the state of one or more properties of the decision-maker 50 at the current time. The properties of the decision-maker 50 can include, for example, the contents of memories, registers, and buffers assigned to the decision-maker 50, the current instruction set, and the targets 50 under which the decision-maker 50 operates, etc.

[0062] Each stimulus 51 can have one or more attributes or qualities AT associated with it. (Note that the reference number 51 in the Fig. (11-13, although used to represent a specific newly observed stimulus 51, this reference number is also used here generally for one or more newly observed stimuli 51). These attributes or qualities AT, as in Fig. As shown in Figure 15, one or more of the following properties may be included: Importance (ATI), Urgency (ATU), Risk (ATR), Reward (ATW), Cost (ATC), Duration (ATD), and Difficulty (ATX). The attributes or qualities (AT) may relate to or differ from the properties, qualities, or characteristics measured or displayed by the axes in the information space.

[0063] Procedure 300 involves taking a given event 30 with one or more stimuli 51-56 ( Fig. 11), identifying an arrangement of attention zones 32 defined in relation to the current state 31 of the decision-maker 50, with the associated stimuli 51-53 located within and / or assigned to their respective attention zones 32 ( Fig. 12), organizing the collection of attention zones 32 and other data according to an event structure 39 ( Fig. 13), and representing this collection of information as a reference node 34, which is connected to individual episodic storage nodes 36 in an episodic storage structure 38 ( Fig. 4) can be compared. If the indicator node 34 sufficiently matches one of the EM nodes 36 in the EM structure 38, the indicator node 34 can be merged or consolidated into the matching EM node 361; however, if the indicator node 34 does not sufficiently match any of the EM nodes 36, the indicator node 34 of the EM structure 38 can be considered a new EM node 36. 22 to be added.

[0064] Again to the Fig. 10-13: In block 360, the comparison step involves comparing a hint node 34 with each of a multitude of EM nodes 36 in an EM structure 38. The hint node 34 represents a new event 30, which is associated with a set of respective attributes AT. 51 , AT 52 , AT 53is associated with one or more newly observed stimuli 51, 52, 53, each stimulus 51 being associated with a variety of newly defined attention zones 32 L , 32 H , 32 U is assigned, which are defined in relation to a current state 31 of the decision-maker 50. The EM structure 38 comprises a selectively interconnected and directed network of EM nodes 36, where each EM node 36 represents a respective previously existing event 40. Each previously existing event 40 is associated with a respective set of attributes AT. 61 , AT 62 , AT 63 linked to one or more previously observed stimuli 61, 62, 63, each of which corresponds to a variety of previously defined attention zones 42 L 42 H 42 Uare assigned, which are defined in relation to a respective previous state 41 of the decision-maker 50. (The previous state 41 can be a previous point in time and can include the state of one or more properties of the decision-maker 50 at that previous point in time.) Each EM node 36 is assigned a respective node risk NR and a respective probability LL, and optionally it can be assigned a respective node reward NW. (Note that the reference number 61 in the Fig. (11-13 is used to represent a specific previously observed stimulus 61, but this reference number is also used here to represent one or more previously observed stimuli 61 in general.)

[0065] The objectives G of the decision-maker 50 may include one or more of the following: minimizing the risk G MR Optimization of utility (Gou), optimization of risk versus reward (G) RRLearning to anticipate a user's needs G AN Finding the most helpful information G FI and learning how to recognize when to offer help to a user G OH These objectives G can be useful when the decision-maker 50 is employed in process controls, logic / control modules, personal data assistants, virtual assistants, expert systems, and the like. Each of the respective attributes AT of the one or more newly observed stimuli 51 and the one or more previously observed stimuli 61 can be defined in relation to one or more objectives G of the decision-maker 50. For example, the objectives G can determine the axes used in the information space, and the stimuli 51, 61 can be placed, visualized, and / or represented in the information space in relation to these axes.

[0066] The attention zones 32 are in Fig. Figure 12 is represented as quadrants of an elliptical subset of the information space. It should be noted that other possible shapes and sizes of the subset of the information space can also be used. Stimulus 51 is located in the lower left quadrant or attention zone 32. LL depicted as a zone for urgent risks 32 U is shown (due to the proximity of stimulus 51 to the origin). Stimulus 52 is located in the upper left quadrant or attention zone 32. UL displayed, which is designated as high-risk area 32 H is represented, and stimulus 53 is located in the lower right quadrant or attention zone 32. LR displayed, which is also classified as a high-risk area 32 H is shown. The upper right quadrant or attention zone 32 U R contains no stimuli and is designated as zone 32 LStimuli with low risk are shown in light gray. Note that stimuli 54, 55, and 56 are not shown because they do not fall into any of the attention zones 32. Stimuli 51, 52, and 53 have respective effect potentials ep. 51 , ep 52 and ep 53 , which represent the effect that each stimulus 51, 52, 53 can have on one or more goals G of the decision-maker 50, as described below. The distance between the origin and a given stimulus 51 represents the magnitude of the stimulus's effect on the achievement of the one or more goals G. Since, for example, stimulus 51 is closer to the origin than stimulus 53, it can be assumed that stimulus 51 has a greater effect on the achievement (or non-achievement) of the goals G than stimulus 53.

[0067] In block 370, the determination step involves identifying which of the several EM nodes 36 has the smallest respective difference metric, thereby defining the best-fitting EM node 361. Each respective difference metric is determined based on a respective total difference between one or more respective aspects AS of the respective EM node 36 and the indicator node 34. These aspects AS are analogous to the properties 50 of procedure 100 and can be, as in Fig. 15 shown, include one or more of the following elements: User temperament AS T (e.g., mood, energy level, attention level, etc.), user capability level AS U (e.g., based on previous decisions and results compared to goals G), environmental factors AS E (e.g., time of day or night, level of distraction in the environment, etc.) and previous tendencies and habits of the user AS H(based on previous performance) as well as the respective importance AS1, urgency AS L , Risk AS R , Reward ASw, Cost ASc, Duration AS D and difficulty AS X , which are associated with one or more newly observed or previously observed stimuli 51, 61. It should be noted that certain aspects AS can also be attributes AT and vice versa, as indicated by the dashed lines in Fig. 15 indicated; for example, importance can be considered an aspect of AS I and / or as an attribute AT I can be expressed. The same applies to urgency AS. U , AT U , Risk AS R , AT R , Reward AS W , AT W , Costs AS C , AT C , Duration AS D , AT D and difficulty AS X , AT XIn some cases, these respective aspects AS and attributes AT can be used interchangeably. It should be noted that... Fig. 13 only one attribute AT for each newly observed stimulus 51, 52, 53 (i.e. AT 51 , AT 52 , AT 53 ) and for each previously observed stimulus 61, 62, 63 (i.e., AT) 61 , AT 62 , AT 63 ) and only one aspect AS for the new event 30 (i.e. AS 30 ) and for the previously existing event 40 (i.e. AS 40 ) shows that more than one attribute AT or one aspect AS can be associated with each of them.

[0068] The total difference can be a sum, an average, or a maximum of the differences between or under the respective aspects AS of clue node 34 and each EM node 36. Additionally, each respective total difference can be a respective sum of one or more weighted penalties assigned against each of the one or more differences between the respective aspects AS of the respective EM node 36 and clue node 34.

[0069] Block 390 checks whether the smallest respective difference metric (between the pointer node 34 and the best-matching node 361) is smaller than a predefined matching tolerance. If so, block 400 executes the consolidation step, in which pointer node 34 is consolidated with the best-matching EM node 361; however, if this is not the case (i.e., the smallest respective difference metric is greater than or equal to the predefined matching tolerance), a new EM node 36 is created in block 410. 22 , corresponding to indicator node 34, added to EM structure 38. The new EM node 36 22 An initial probability value can be assigned (i.e., an initial probability value LL of the new node); optionally, the new node can be assigned 36 22 An initial node risk NR and an initial node reward NW are also assigned.

[0070] In block 420, the identification step includes identifying the most probable next node 362 and / or the riskiest next node 36. 10 If the indicator node 34 has been consolidated into the best-matching node 361, then the most likely next node 362 is the node 36 with the highest probability LL among the nodes 36 immediately downstream of the best-matching node 361, and the riskiest next node 36 10 The node 36 with the highest node risk NR among the nodes 36 immediately downstream of the best-matching node 361. If, on the other hand, the indicator node 34 is considered a new node 36 22 was added and if there is one or more node 36 immediately downstream of the new node 36 22 If there is a , then the most probable next node 362 will be the node 36 with the highest probability LL among the nodes 36 immediately downstream of the new node 36.22 be, and the riskiest next node 36 10 The node 36 with the highest node risk NR among the nodes 36 will be located immediately downstream of the new node 36. 22 If there is no node 36 immediately downstream of the new node 36 22 there is, for example, in the one in Fig. In the case shown in section 4, then no most probable or riskiest next nodes 362, 36 are identified. 10 identified. In such a case, however, it is likely that the decision-maker will learn of 50 new events 30 that are clearly different from the previously learned, previously existing events 40, and the decision-maker 50 will continue to identify new nodes 36. 22to add to EM structure 38 as the decision-maker 50 continues to progress. In this way, EM structure 38 can be expanded to include all new events 30 that the decision-maker 50 experiences. An EM structure 38 can continue to grow over time and become quite complex, so that in some cases the most probable next node 362 and the riskiest next node 36 22 the same node 36 can be.

[0071] The respective node risk NR of each EM node 36 can be a respective maximum, an average, or an aggregate of the respective stimulus risks SR. 61 , SR 62 , SR 63 for one or more previously observed stimuli 61, 62, 63 in the associated previously existing event 40. The respective stimulus risk SR for each previously observed stimulus 61, 62, 63 can be determined by a sigmoid function that is based on a respective action potential ep 61 , ep 62 , ep 63between the previously observed stimulus 61, 62, 63 and the one or more goals G of the decision-maker 50 is applied. (Note that the reference number ep 61 in Fig. 12, however, to illustrate a specific effect potential ep 61 This reference number is used here, but also generally to represent the effect potential ep. 61 between a previously observed stimulus 61 and the one or more goals G of the decision-maker 50).

[0072] Procedure 300 may further include the following steps in block 330: (i) Calculating an individual risk IR 51 , IR 52 , IR 53 for each of the one or more newly observed stimuli 51, 52, 53 using a sigmoid function that is applied to a respective effect potential ep 51 , ep 52 , ep 53(ii) at Block 340, establishing a respective total risk TR for each of the plurality of newly defined attention zones 32, based on the respective individual risks IR of the one or more newly observed stimuli 51 assigned to each respective newly defined attention zone 32; and (iii) at Block 350, defining the clue node 34 as a grouping of the plurality of newly defined attention zones 32 organized according to their respective total risks TR.

[0073] Procedure 300 can further include, in block 380, the step of backpropagating a given node risk NR associated with a high-risk EM node 36 to one or more EM nodes 36 upstream of the high-risk EM node 36, wherein the node risk NR associated with the high-risk EM node 36 is greater than a predetermined risk threshold. This backpropagation of the associated node risk NR can use a linear or a logistic function. For example, if a given EM node 36 has a node risk NR that is above the predetermined risk threshold, then this node can be considered a high-risk node 36, and the node risk NR for this high-risk node 36 can be propagated backproportionally to one or more other nodes 36 immediately upstream of the high-risk node 36.

[0074] The procedure 300 may further include in block 310 the step of capturing the set of the respective attributes AT associated with the one or more newly observed stimuli 51 as perceived data; and in block 320 the step of converting the perceived data into inputs, signals or data relating to the indicator node 34, e.g. for reception by a data input module 460 (described in detail below).

[0075] Fig. Figure 9 shows a block diagram of a system / controller 430 for optimizing a decision for a decision-maker 50, configured to perform some or all of the steps of the procedure 300 described above. The system / controller 430 can perform all of the steps described in the procedure 300. Fig.The system may contain the 9 blocks or modules shown, or it may contain a subset of these blocks or modules. Each of these blocks or modules may be implemented in the form of hardware, software, and / or firmware, with two or more of the blocks or modules integrated together (e.g., as part of a single hardware unit) and one or more other blocks or modules implemented separately.

[0076] The system / controller 430 comprises a data input module 460, an EM module 470, and a prediction module 480. The data input module 460 is configured to receive inputs relating to a hint node 34, where the hint node 34 represents a new event 30 associated with a set of respective attributes AT, which are associated with one or more newly observed stimuli 51, each newly observed stimulus 51 being located within and / or assigned to one of several newly defined attention zones 32, which are defined in relation to a current state 31 of the decision-maker 50.The data input module 460 is further configured to: (i) calculate a respective individual risk IR for each of the one or more newly observed stimuli 51 using a sigmoidal function applied to a respective action potential ep between each respective newly observed stimulus 51 and the one or more goals G of the decision-maker 50; (ii) determine a respective total risk TR for each of the plurality of newly defined attention zones 32, based on the respective individual risks IR of the one or more newly observed stimuli 51 that are located within and / or assigned to each respective newly defined attention zone 32; and (iii) define the indicator node 34 as a grouping of the plurality of newly defined attention zones 32, organized according to their respective total risks TR.

[0077] The EM module 470 is operationally connected to the data input module 460 and contains an EM structure 38 with a selectively interconnected and directed network of EM nodes 36, wherein each EM node 36 represents a respective previously existing event 40, wherein each previously existing event 40 is associated with a respective set of attributes AT that are associated with one or more previously observed stimuli 61, each of which is located within a respective of several previously defined attention zones 42 and / or is assigned to a respective of several previously defined attention zones 42 that are defined in relation to a respective previous state 41 of the decision-maker 50, and wherein each EM node 36 has a respective node risk NR and a respective probability LL that are associated with it.The EM module 470 is configured to: (x) compare the antecedent node 34 with each of the EM nodes 36; (y) determine which of the EM nodes 36 has the smallest respective difference metric, thereby defining a best-matching EM node 361, where each respective difference metric is determined based on a respective total difference between one or more respective aspects AS of the respective EM node 36 and the antecedent node 34; and (z) consolidate the antecedent node 34 with the best-matching EM node 361 if the smallest respective difference metric is less than a predetermined matching tolerance, otherwise add a new EM node 3622 corresponding to the antecedent node 34 to the EM structure 38 if the smallest respective difference metric is greater than or equal to the predetermined matching tolerance.

[0078] The prediction module 480 is operationally linked to the EM module 470 and is configured to identify one or both: a most probable next node 362 among one or more EM nodes 36 immediately downstream of the best-matching or new EM node 361, 3622, wherein the most probable next node 362 has a highest probability LL among the immediately downstream EM nodes 36; and a most risky next node 3610 among the one or more EM nodes 36 immediately downstream of the best-matching or new EM node 361, 3622, wherein the most risky next node 3610 has a highest node risk NR among the immediately downstream EM nodes 36. Furthermore, the prediction module 480 can be operationally connected to other modules / devices 490 that are located outside the system / control unit 430.

[0079] The system / controller 430 may further include a perception module 450, which is operationally connected to the data input module 460 and configured to: capture the set of respective attributes AT associated with the one or more newly observed stimuli 51 as perceived data; and convert the perceived data into inputs relating to the indicator node 34 for reception by the data input module 460. The system / controller 430 may additionally include one or more sensors 440 (e.g., radar, lidar, ultrasound, infrared, temperature, etc.) that are operationally connected to the perception module 450 and provide input signals to it.

[0080] The respective node risk NR of each EM node 36 can be a respective maximum, an average, or an aggregate of respective object risks OR for the one or more previously observed stimuli 61 in the associated previously existing event 40, and the respective object risk OR for each previously observed stimulus 61 can be determined by a sigmoidal function applied to a respective action potential ep between the previously observed object 43 and the one or more targets G of the decision-maker 50. The EM module 470 can further be configured to forward a respective node risk NR assigned to a high-risk EM node 36 to one or more EM nodes 36 upstream of the high-risk EM node 36, where the assigned node risk NR is greater than a predetermined risk threshold.

[0081] Being able to predict the riskiest and most likely next EM nodes 3610, 362 can help the Ego vehicle 20 to make its next move or decision (thus providing a form of predictive navigation), and likewise it can help the Decision Maker 50 to optimize its next move or decision.

Claims

[1] Method (100) for predictive navigation control for an ego vehicle (20), comprising: Comparing a clue node (34) with each of a plurality of episodic memory nodes (36) in an episodic memory structure (38), wherein the clue node (34) represents a new event (30) associated with a set of respective locations (L), velocities (S), and directions (H) associated with one or more newly observed objects (21), each located within a respective plurality of newly defined attention zones (32) around the ego-vehicle (20), and wherein the episodic memory structure (38) comprises a selectively interconnected and directed network of the episodic memory nodes (36), each episodic memory node (36) representing a respective previously existing event (40), each previously existing event (40) being associated with a respective set of locations (L), velocities (S), and directions (H),which are associated with one or more previously observed objects (43) located within a plurality of previously defined attention zones (42) around the ego-vehicle (20), and wherein each episodic storage node (36) has a respective node risk (NR) and a respective probability (LL) associated with it; Determine which of the several episodic storage nodes (36) has the smallest respective difference metric, thereby defining a best-fitting episodic storage node (36), wherein each respective difference metric is determined on the basis of a respective total difference between one or more respective properties (49) of the respective episodic storage node (36) and the hint node (34); Consolidating the hint node (34) with the best-matching episodic storage node (36) if the smallest respective difference metric is less than a predetermined matching tolerance, or adding a new episodic storage node (36) corresponding to the hint node (34) to the episodic storage structure (38) if the smallest respective difference metric is greater than or equal to the predetermined matching tolerance; and Identifying one or both of (i) a most probable nearest episodic storage node (36) among one or more episodic storage nodes (36) immediately downstream of the best-matching or new episodic storage node (36), wherein the most probable nearest episodic storage node (36) has a highest probability (LL) among the immediately downstream episodic storage nodes (36), and (ii) a most risky nearest episodic storage node (36) among the one or more episodic storage nodes (36) immediately downstream of the best-matching or new episodic storage node (36), wherein the most risky nearest episodic storage node (36) has a highest node risk (NR) among the immediately downstream episodic storage nodes (36). [2] Method (100) according to claim 1, wherein the respective node risk (NR) of each episodic storage node (36) is a respective maximum, an average or an aggregate of respective object risks (OR) for the one or more previously observed objects (43) in the associated previously existing event (40), and wherein the respective object risk (OR) for each previously observed object (43) is determined by a sigmoidal function applied to a respective distance (d) between the previously observed object (43) and the ego vehicle (20). [3] Method (100) according to claim 2, further comprising: Calculating a respective individual risk (IR) for each of the one or more newly observed objects (21) using the sigmoidal function applied to a respective distance (d) between each respective newly observed object (21) and the ego vehicle (20); Establishing a respective overall risk (TR) for each of the several newly defined attention zones (32), based on the respective individual risks (IR) of the one or more newly observed objects (21) located within each respective newly defined attention zone (32); and Defining the indicator node (34) as a grouping of the several newly defined attention zones (32) which are organized according to their respective total risks (TR). [4] Method (100) according to claim 3, wherein the respective object risk (OR) or individual risk (IR) for each previously observed or newly observed object (21) is determined by R = 2-({1 - 1 / (1 + e^[(mindist - SAFEDIST) / (SAFEDIST / 2)]} - 0.5) + 0.5, where R is the respective object risk (OR) or individual risk (IR), mindist is a distance (d) to the previously observed or newly observed object (21) from the ego vehicle (20), and where SAFEDIST is a distance (d) that depends on one or more of a road surface type, a road structure type, weather / environmental conditions and a relative lane position, closing speed or closing acceleration between the previously observed or newly observed object (43, 21) and the ego vehicle (20). [5] Method (100) according to claim 1, further comprising: Back transmission of a respective node risk (NR) associated with a high-risk episodic storage node (36) to one or more episodic storage nodes (36) upstream of the high-risk episodic storage node (36), wherein the associated node risk (NR) is greater than a predetermined risk threshold. [6] Method (100) according to claim 1, wherein when a new episodic storage node (36) is added to the episodic storage structure (38), the new episodic storage node (36) is added as a child node to one or more parent nodes, each parent node being a previously existing episodic storage node (36), and wherein an initial probability value is assigned to the new episodic storage node (36). [7] Control (230) for predictive navigation control for an ego vehicle (20), comprising: (a) a data input module (260) configured to receive inputs relating to a clue node (34), wherein the clue node (34) represents a new event (30) associated with a set of respective locations (L), velocities (S), and directions (H) associated with one or more newly observed objects (21), each located within a respective plurality of newly defined attention zones (32) around the ego-vehicle (20), the data input module (260) further configured to: Calculating an individual risk (IR) for each of the one or more newly observed objects (21) using a sigmoidal function applied to a respective distance (d) between each newly observed object (21) and the ego vehicle (20); to determine a respective overall risk (TR) for each of the several newly defined attention zones (32) based on the respective individual risks (IR) of the one or more newly observed objects (21) located within each respective newly defined attention zone (32); and define the indicator node (34) as a grouping of the several newly defined attention zones (32) which are organized according to their respective total risks (TR); (b) an episodic storage module (270) operationally connected to the data input module (260) and containing an episodic storage structure (38) comprising a selectively interconnected and directed network of episodic storage nodes (36), wherein each episodic storage node (36) represents a respective previously existing event (40), wherein each previously existing event (40) is associated with a respective set of locations (L), velocities (S), and directions (H) associated with one or more previously observed objects (43), each located within a respective plurality of previously defined attention zones (42) around the ego-vehicle (20), and wherein each episodic storage node (36) has a respective node risk (NR) and a respective associated probability (LL), wherein the episodic storage module (270) is configured to: Comparing the hint node (34) with each of the episodic storage nodes (36); Determine which of the episodic storage nodes (36) has the smallest respective difference metric, thereby defining a best-fitting episodic storage node (36), wherein each respective difference metric is determined on the basis of a respective total difference between one or more respective properties (49) of the respective episodic storage node (36) and the hint node (34); and Consolidating the hint node (34) with the best-matching episodic storage node (36) if the smallest respective difference metric is less than a predetermined matching tolerance, or adding a new episodic storage node (36) corresponding to the hint node (34) to the episodic storage structure (38) if the smallest respective difference metric is greater than or equal to the predetermined matching tolerance; and (c) a prediction module (280) operationally connected to the episodic storage module (270) and configured to (i) identify a most probable next episodic storage node (36) among one or more episodic storage nodes (36) immediately downstream of the best-matching or new episodic storage node (36), wherein the most probable next episodic storage node (36) has a highest probability (LL) among the immediately downstream episodic storage nodes (36), and (ii) identify a most risky next episodic storage node (36) among the one or more episodic storage nodes (36) immediately downstream of the best-matching or new episodic storage node (36), wherein the most risky next episodic storage node (36) has a highest node risk (NR) among the immediately downstream episodic storage nodes (36). [8] Control (230) according to claim 7, further comprising: (d) a perception module (250) that is operationally connected to and configured with the data input module (260) to: Capturing the set of respective locations (L), velocities (S), and directions (H) associated with the one or more newly observed objects (21) as perceived data; and Converting the perceived data into the inputs related to the indicator node (34) for reception by the data input module (260). [9] Controller (230) according to claim 7, wherein the respective node risk (NR) of each episodic storage node (36) is a respective maximum, an average or an aggregate of respective object risks (OR) for the one or more previously observed objects (43) in the associated previously existing event (40), and wherein the respective object risk (OR) for each previously observed object (43) is determined by the sigmoidal function applied to a respective distance (d) between the previously observed object (43) and the ego vehicle (20). [10] Controller (230) according to claim 7, wherein the episodic storage module (270) is further configured to: Transferring a corresponding node risk (NR) associated with a high-risk episodic storage node (36) back to one or more episodic storage nodes (36) upstream of the high-risk episodic storage node (36), wherein the associated node risk (NR) is greater than a predetermined risk threshold.

Images