JP2022026228A - Track generation device, track generation method, and track generation program - Google Patents

Abstract

To provide a track generation device, etc. that can effectively utilize a plurality of prediction results.SOLUTION: A track generation device 100 comprises a behavior prediction unit 110 that acquires prediction results from a plurality of prediction models as to the behavior of moving objects around a self vehicle. The track generation device 100 comprises a prediction result arbitration unit 120 that arbitrates each prediction result based on at least one of a safety, comfort, and fitness of each prediction model to a driving environment when driving based on each prediction result. The track generation device 100 comprises a track planning unit 130 that executes a travel track plan based on the arbitrated prediction result.SELECTED DRAWING: Figure 2

Description

The disclosure herein relates to techniques for predicting the behavior of moving objects.

Patent Document 1 discloses a system for controlling the traveling of the own vehicle. This system weights multiple decisions regarding the future behavior of the vehicle and outputs the final decision.

JP-A-2019-164729

By the way, it is required to predict the behavior of a moving body around the own vehicle and determine the behavior of the own vehicle based on the prediction result. If there are multiple prediction results for the behavior of a moving object, it is necessary to arbitrate these prediction results. Patent Document 1 does not disclose in detail a method of mediation for a plurality of judgments. Therefore, the technique of Patent Document 1 may not be able to effectively utilize a plurality of prediction results.

It is an object of the disclosure to provide an orbit generator, an orbit generation method, and an orbit generation program capable of effectively utilizing a plurality of prediction results.

The plurality of aspects disclosed herein employ different technical means to achieve their respective objectives. Further, the scope of claims and the reference numerals in parentheses described in this section are examples showing the correspondence with the specific means described in the embodiment described later as one embodiment, and limit the technical scope. is not it.

One of the disclosed track generators is a track generator that generates a travel track planned for the own vehicle (Va).
A behavior prediction unit (110) that acquires prediction results from multiple prediction models for the behavior of moving objects around the own vehicle, and
An arbitration unit (120) that arbitrates each prediction result based on at least one of safety, comfort, and fitness of each prediction model to the driving environment when driving based on each prediction result.
The track planning unit (130), which executes the planning of the running track based on the arbitrated prediction result,
To prepare for.

One of the disclosed track generation methods is a track generation method executed by the processor (102) in order to generate a traveling track scheduled for the own vehicle (Va).
A behavior prediction process (S10) that acquires prediction results from multiple prediction models for the behavior of moving objects around the own vehicle, and
An arbitration process (S20, S30, S40,) that arbitrates each prediction result based on at least one of the safety, comfort, and fitness of each prediction model to the driving environment when driving based on each prediction result. S50, S60) and
The track planning process (S80), which executes the planning of the running track based on the arbitrated prediction result,
including.

One of the disclosed track generation programs is a track generation program that includes instructions stored in a storage medium (101) and executed by a processor (102) in order to generate a travel track scheduled for the own vehicle (Va). There,
The order is
A behavior prediction process (S10) for acquiring prediction results from a plurality of prediction models for the behavior of moving objects around the own vehicle, and
An arbitration process (S20, S30, S40,) that arbitrates each prediction result based on at least one of the safety, comfort, and fitness of each prediction model to the driving environment when driving based on each prediction result. S50, S60) and
The track planning process (S80), which executes the planning of the running track based on the arbitrated prediction result,
including.

According to these disclosures, each prediction result is arbitrated and arbitrated based on at least one of the safety, comfort, and fitness of each prediction model to the driving environment when driving based on each prediction result. An orbital plan is generated based on the prediction results. Therefore, a plurality of prediction results arbitrated based on at least one viewpoint of safety, comfort, and goodness of fit can be utilized for trajectory planning. As described above, an orbit generation device, an orbit generation method, and an orbit generation program capable of effectively utilizing a plurality of prediction results can be provided.

It is a figure which shows the system including the orbit generator. It is a block diagram which shows an example of the function which a trajectory generator has. It is a figure for demonstrating the safety degree evaluation of each prediction result. It is a graph which shows an example of weight setting based on a safety degree. It is a figure for demonstrating the comfort degree evaluation of each prediction result. It is a graph which shows an example of weight setting based on comfort degree. It is a figure for demonstrating the trajectory plan using the prediction result. It is a flowchart which shows an example of the trajectory generation method executed by the trajectory generation apparatus. It is a flowchart which shows the detail of the goodness-of-fit evaluation process in FIG.

(First Embodiment)
The trajectory generator 100 of the first embodiment will be described with reference to FIGS. 1 to 9. The track generator 100 is an electronic control device mounted on the own vehicle Va, which has at least one of an automatic driving function and an advanced driving support function. The track generating device 100 generates a traveling track of the own vehicle Va traveling by the above-mentioned function. The track generator 100 is connected to the peripheral monitoring ECU 10, the locator 20, the vehicle speed sensor 30, and the vehicle control ECU 40 via a communication bus or the like.

The peripheral monitoring ECU 10 is mainly composed of a microcomputer including a processor, a memory, an I / O, and a bus connecting these, and executes various processes by executing a control program stored in the memory. The peripheral monitoring ECU 10 acquires a detection result from the peripheral monitoring sensor 11 and recognizes the traveling environment of the own vehicle based on the detection result. The peripheral monitoring sensor 11 is an autonomous sensor that monitors the surrounding environment of the vehicle A, and has LiDAR (Light Detection and Ranging / Laser Imaging Detection and Ranging) that detects a point cloud of feature points of a feature, and a front of the vehicle A. Includes peripheral surveillance cameras and the like that capture images of the included predetermined range. Further, the peripheral monitoring sensor 11 may include a millimeter wave radar, sonar, and the like. The peripheral monitoring ECU 10 recognizes the presence / absence of moving objects such as other vehicles Vb and pedestrians, their relative positions, relative speeds, and the like as peripheral information based on the detection information of the peripheral monitoring sensor 11. Further, the peripheral monitoring ECU 10 determines whether or not the peripheral monitoring sensor 11 has a failure. The peripheral monitoring ECU sequentially provides peripheral information and the presence or absence of a failure to the trajectory generator.

The locator 20 generates own vehicle position information and the like by compound positioning that combines a plurality of acquired information. The locator 20 includes a GNSS (Global Navigation Satellite System) receiver 21, an inertial sensor 22, a map DB 23, and a locator ECU 24. The GNSS receiver 21 receives positioning signals from a plurality of positioning satellites. The inertial sensor 22 is a sensor that detects the inertial force acting on the vehicle A. The inertial sensor 22 includes, for example, a gyro sensor and an acceleration sensor.

The map DB 23 is a non-volatile memory and stores map data such as link data, node data, road shape, and structures. In addition, the map DB 23 stores information such as the position of the traffic light, the type and position of the road sign, and the type and position of the road marking. The map data may be a three-dimensional map composed of point clouds of road shapes and feature points of structures. The three-dimensional map may be generated based on the captured image by REM (Road Experience Management). Further, the map data may include traffic regulation information, road construction information, meteorological information, and the like. The map data stored in the map DB 23 is updated regularly or at any time based on the latest information received by the in-vehicle communication device.

The locator ECU 24 mainly includes a microcomputer including a processor, RAM, a storage unit, an input / output interface, and a bus connecting them. The locator ECU sequentially positions the position of the vehicle A (hereinafter referred to as the own vehicle position) by combining the positioning signal received by the GNSS receiver, the map data of the map DB, and the measurement result of the inertial sensor. The position of the own vehicle may be, for example, configured to be represented by the coordinates of latitude and longitude. For the positioning of the own vehicle position, the mileage obtained from the signals sequentially output from the vehicle speed sensor 30 mounted on the vehicle A may be used. When a three-dimensional map consisting of a point cloud of road shapes and feature points of a structure is used as map data, the locator uses the three-dimensional map and the detection result of the peripheral monitoring sensor without using a GNSS receiver. It may be configured to specify the position of the own vehicle by using it. The locator ECU 24 sequentially provides the vehicle position information, map data, detection information of the inertial sensor 22, and the like to the track generator.

The vehicle control ECU 40 is an electronic control device that controls acceleration / deceleration and steering of the vehicle A. Vehicle control ECUs include steering ECUs that perform steering control, power unit control ECUs that perform acceleration / deceleration control, brake ECUs, and the like. The vehicle control ECU acquires detection signals output from each sensor such as a steering angle sensor and a vehicle speed sensor mounted on the vehicle A, and controls each running of an electronically controlled throttle, a brake actuator, an EPS (Electric Power Steering) motor, and the like. Output the control signal to the device. By acquiring the control instruction of the vehicle A from the travel support device, the vehicle control ECU controls each travel control device so as to realize autonomous driving or driving support according to the control instruction.

The orbit generation device 100 mainly includes a computer including a memory 101, a processor 102, an input / output interface, and a bus connecting them. The processor 102 is hardware for arithmetic processing. The processor 102 includes, for example, at least one of a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a RISC (Reduced Instruction Set Computer) -CPU, and the like as a core.

The memory 101 non-transiently stores or stores a computer-readable program, data, or the like, for example, at least one type of non-transitional substantive storage medium (non-transitional memory, magnetic medium, optical medium, or the like, etc.). transitory tangible storage medium). The memory 101 stores various programs executed by the processor 102, such as an orbit generation program described later.

The processor 102 executes a plurality of instructions included in the orbit generation program stored in the memory 101. As a result, the track generator 100 constructs a plurality of functional units for generating future tracks of the own vehicle Va. In this way, in the trajectory generator 100, a plurality of functional units are constructed by causing the processor 102 to execute a plurality of instructions by the program stored in the memory 101. Specifically, as shown in FIG. 2, the orbit generation device 100 is constructed with functional units such as a behavior prediction unit 110, a prediction result arbitration unit 120, and an orbit planning unit 130.

The behavior prediction unit 110 predicts the behavior of a moving object around the own vehicle Va. The moving object to be predicted may be another vehicle Vb, a pedestrian, an animal, or the like. The behavior prediction unit 110 inputs input information to a plurality of prediction models, and outputs prediction results from each prediction model. The input information includes, for example, position information and speed information of the moving body, position information of the own vehicle Va, speed information, and acceleration information. Further, the input information includes traffic light information such as the position and lighting color of the traffic light, sign information such as the position and type of the sign, and road sign information such as the position and type of the road sign (stop line or the like). The input information to be input to each prediction model may be all common, or at least a part thereof may be different. Each prediction model outputs, for example, a prediction result of the behavior of a moving object as a probability distribution for each position. The behavior prediction unit 110 sequentially provides the output prediction result to the prediction result arbitration unit 120.

Multiple predictive models are defined by different policies. The plurality of prediction models include, for example, an intention non-estimation model that does not estimate the motion intention of the moving object to be predicted, and a plurality of intention estimation models that estimate the motion intention of the moving object. The unintentional non-estimation model outputs the prediction result by linear prediction based on the current mobile information. That is, the unintentional non-estimation model predicts the behavior of the moving object, assuming that the current kinetic physical quantity of the moving object is simply maintained.

On the other hand, a plurality of intention estimation models predict the behavior of a moving body by assuming that the current kinetic physical quantity of the moving body changes in the future by estimating the motion intention. The intent estimation model is provided by, for example, a trained model trained by machine learning to output a motion prediction for an input of motion information. Multiple intent estimation models include models optimized for predictions on motorways, models optimized for predictions on ordinary roads, models optimized for predictions in specific areas, and the like. That is, each of the plurality of intention estimation models is a trained model specialized for a specific driving environment. The intent estimation model may include a model optimized for an area with many bicycles and motorcycles, an area with many animals, an area with many rough driving, and the like. The plurality of intention estimation models may be machine-learned based on a common model, or may be based on different models. The intention estimation model may be provided by a rule-based model instead of the trained model learned by machine learning.

The prediction result arbitration unit 120 evaluates the prediction result of each prediction model based on a plurality of evaluation criteria. The prediction result arbitration unit 120 includes a arbitration necessity determination unit 121, a safety evaluation unit 122, a comfort evaluation unit 123, a conformity evaluation unit 124, and a confirmation unit 125 as sub-functional units.

The arbitration necessity determination unit 121 determines whether or not it is necessary to arbitrate a plurality of prediction results. When the difference between the plurality of prediction results is larger than the allowable range, the arbitration necessity determination unit 121 determines that arbitration is necessary.

Specifically, the arbitration necessity determination unit 121 calculates the variance of a plurality of prediction results, determines that arbitration is necessary when the variance is larger than the threshold value, and arbitration is unnecessary when the variance is smaller than the threshold value. Is determined. Alternatively, the arbitration necessity determination unit 121 may plan the traveling track of the own vehicle Va based on each prediction result, and may determine that arbitration is unnecessary when the degree of coincidence of each traveling track exceeds a predetermined value. ..

The safety evaluation unit 122 evaluates the safety of a plurality of prediction results. Specifically, the safety evaluation unit 122 first generates an evaluation route Re, which is a provisional route, based on each prediction result, and calculates the safety evaluation value EVs of the evaluation route Re.

The safety evaluation values EVs are calculated based on the index parameters that are indicators of the safety of the evaluation path Re. For example, the safety evaluation unit 122 uses the collision margin time (TTC: Time To Collision) between the moving body and the own vehicle Va as an index parameter. The safety evaluation unit 122 evaluates the safety evaluation value EVs smaller as the TTC is smaller. That is, the safety evaluation unit 122 lowers the safety evaluation value EVs as the safety level of the evaluation path Re is lower. As an example, the safety evaluation unit 122 may determine the safety evaluation value EVs according to the graph of the correspondence between the TTC and the safety evaluation value EVs shown in FIG.

Further, the safety evaluation unit 122 sets weights corresponding to the safety evaluation values EVs for the prediction result. Specifically, the safety evaluation unit 122 sets a larger weight as the prediction result has a lower safety evaluation value EVs. In other words, the safety evaluation unit 122 sets the weight so that the contribution of the prediction result having a large influence on the safety of the own vehicle Va (for example, collision risk) to the track plan becomes larger. In the example shown in FIG. 3, the weight w1 of the prediction result of the first prediction model that blocks the right turn of the evaluation path Re is set to be larger than a predetermined specified value, and the weight w1 of the second prediction model that does not block the right turn of the evaluation path Re is set. The weight w2 of the prediction result is not changed from a predetermined specified value. In FIG. 3, the prediction result of each prediction model is represented by a gray scale mosaic shown in the traveling direction of another vehicle Vb. The darker the mosaic, the higher the probability of existence of other vehicle Vb (same for FIGS. 5 and 7). For example, the safety evaluation unit 122 determines the weight of each prediction result based on the relationship between the safety evaluation value EVs defined in the graph shown in FIG. 4 and the weight w _n . More specifically, when the safety evaluation value EVs exceeds a predetermined value, the safety evaluation unit 122 sets a constant weight (base weight w_b) regardless of the size of the safety evaluation value. Then, the safety evaluation unit 122 sets a larger weight with respect to the base weight as the safety evaluation value EVs is smaller than a predetermined value.

The safety evaluation unit 122 may use a value other than TTC as an index parameter. For example, the safety evaluation unit 122 may use the vehicle speed or type of another vehicle Vb as an index parameter. Specifically, the safety evaluation unit 122 may set the safety evaluation value EVs lower as the vehicle speed of the other vehicle Vb is higher or the vehicle class is larger. The safety evaluation unit 122 may appropriately change the index parameters according to the current driving environment.

The comfort level evaluation unit 123 evaluates the comfort level for a plurality of prediction results. Specifically, the comfort evaluation unit 123 generates an evaluation path Re based on each prediction result, and calculates a comfort evaluation value EVc for the evaluation path Re.

The comfort level evaluation value EVc is calculated based on an index parameter that is an index of the comfort level of the evaluation path Re. For example, the comfort evaluation unit 123 uses the traveling time of the evaluation route Re as an index parameter. Specifically, the comfort evaluation unit 123 determines that the comfort evaluation value EVc is large enough to travel on the evaluation route Re in a short time.

More specifically, regarding the behavior of the other vehicle Vb traveling in the lane adjacent to the current lane of the own vehicle Va, when the prediction result that the other vehicle Vb enters the current lane is output by the second prediction model, the evaluation path Re is , It becomes a deceleration route so as to avoid approaching another vehicle Vb (see FIG. 5). In this case, the comfort evaluation unit 123 calculates the comfort evaluation value EVc of the prediction result of the second prediction model smaller than the prediction result of the first prediction model in which the evaluation path Re overtaking the other vehicle Vb is generated. do. As an example, the comfort evaluation unit 123 may determine the comfort evaluation value EVc according to the graph of the correspondence between the traveling time of the evaluation route Re and the comfort evaluation value EVc shown in FIG.

The comfort evaluation unit 123 may use a value other than the traveling time of the evaluation route Re as an index parameter. For example, the comfort evaluation unit 123 may use the lateral acceleration change of the own vehicle Va as an index parameter, and may calculate the comfort evaluation value EVc smaller as the acceleration change is larger. The comfort evaluation unit 123 may appropriately change the index parameters according to the current driving environment.

Further, the comfort evaluation unit 123 sets a weight corresponding to the comfort evaluation value EVc for the prediction result. Specifically, the comfort evaluation unit 123 sets a smaller weight as the prediction result has a lower comfort evaluation value EVc. In other words, the comfort evaluation unit 123 sets the weight so that the contribution of the prediction result having a small comfort to the trajectory plan becomes smaller. For example, the comfort evaluation unit 123 determines the weight of the prediction result based on the relationship between the comfort evaluation value EVc defined in the graph shown in FIG. 6 and the weight. More specifically, when the comfort evaluation value EVc exceeds a predetermined value, the comfort evaluation unit 123 sets a constant weight (base weight w_b) regardless of the magnitude of the comfort evaluation value EVc. Then, the comfort evaluation unit 123 sets a smaller weight with respect to the base weight w_b as the comfort evaluation value EVc is smaller than a predetermined value.

However, the comfort evaluation unit 123 cancels the setting of reducing the weight for the prediction result in which the safety evaluation value EVs is out of the permissible range, for example, the prediction result below the threshold value in the safety evaluation unit 122. .. In other words, the comfort evaluation unit 123 stops the decrease in the contribution to the trajectory plan according to the comfort evaluation value EVc for the prediction result in which the safety evaluation value EVs is below the threshold value.

The goodness-of-fit evaluation unit 124 sets the weight of each prediction result based on the traveling environment of the own vehicle Va. Specifically, the goodness-of-fit evaluation unit 124 sets the weight according to the goodness of fit between the traveling environment of the own vehicle Va and each prediction model.

The goodness-of-fit evaluation unit 124 sets the weight of the prediction result by the prediction model to be larger as the goodness of fit of the prediction model to the driving environment is higher. In other words, the goodness-of-fit evaluation unit 124 sets a large weight of the prediction result by the prediction model when driving in the driving environment where the prediction model is good, and when driving in the driving environment which is not good. Is set to a small weight of the prediction result.

For example, the goodness-of-fit evaluation unit 124 fits the intention estimation model when the vehicle is traveling in a region where the difficulty level of driving judgment is relatively high, as compared with the case where the vehicle is traveling in a region where the difficulty level is relatively low. Judge that the degree is high. The difficulty level of the driving determination may be determined based on, for example, the estimated frequency of appearance of moving objects that can hinder the progress of the own vehicle Va. The goodness-of-fit evaluation unit 124 determines whether the difficulty level of the driving judgment is high or low depending on whether or not the current traveling area is a predetermined specific area. The specific area includes an area with a relatively large number of crossing pedestrians, an area with a branch or a confluence, an area with a roundabout, and the like. When it is determined that the conformity evaluation unit 124 is traveling in these specific regions, the weight of the prediction result based on the intention estimation model is made larger than when it is determined that the vehicle is not traveling in the specific region.

Further, the goodness-of-fit evaluation unit 124 determines whether or not at least one of the specialized driving environments of each prediction model, that is, the assumed environment of each prediction model, matches the current driving environment. When it is determined that they match, the goodness-of-fit evaluation unit 124 increases the weight of the prediction result by the matching prediction model more than when it is determined that they do not match. The specialized driving environment of the predictive model is, for example, an area with many bicycles and motorcycles, an area with many animals, an area with many rough driving, and the like.

In addition, the goodness-of-fit evaluation unit 124 determines that the goodness of fit of the intention estimation model is so low that the detection error of the peripheral monitoring sensor 11 is estimated to be large. For example, the goodness-of-fit evaluation unit estimates that the farther the prediction target is, the larger the detection error. Further, the goodness-of-fit evaluation unit 124 estimates that the larger the track curvature, the larger the detection error. Further, the goodness-of-fit evaluation unit 124 estimates that the larger the rainfall, the larger the detection error. Further, the goodness-of-fit evaluation unit 124 estimates that the smaller the amount of light, the larger the detection error. In addition, the goodness-of-fit evaluation unit 124 estimates that the detection error will be larger when traveling inside the tunnel than when traveling outside the tunnel.

Further, the goodness-of-fit evaluation unit 124 determines that the goodness of fit of the intention estimation model is lower when the current driving scene is a non-target scene that is not the prediction target of the intention estimation than when it is not the non-target scene. The non-target scenes include, for example, an accident site, a construction site, a sensor limit, and the like.

Further, the goodness-of-fit evaluation unit 124 determines that the goodness of fit of the intention estimation model is lower when the specific function of the own vehicle Va is not available than when it is available. The specific function includes a detection function of the peripheral monitoring sensor 11, an advanced driving support function, and the like.

The determination unit 125 integrates each weighted prediction result and generates an arbitrated prediction result. Assuming that the weights set in the i-th prediction model are wi and the prediction results Pa and i of the i-th prediction model, the arbitrated prediction result Pa is acquired based on the following mathematical formula (1).

However, if the determination unit 125 determines that the arbitration of the prediction result is not necessary by the arbitration necessity determination unit 121, the arbitration of the prediction result is stopped. In this case, the determination unit 125 determines the prediction result by the specific prediction model among the plurality of prediction results as the final prediction result of the moving body. The specific prediction model at this time is preset. Alternatively, a specific prediction model may be appropriately changed according to the traveling environment of the own vehicle Va.

The track planning unit 130 generates a future track to be followed by the own vehicle Va based on the prediction result of the moving body determined by the determination unit 125. The future track is a planned travel track that follows future behavior, and defines the travel position of the own vehicle Va according to the progress of the own vehicle Va. In addition, the future track may define the speed of the own vehicle Va at each traveling position.

Specifically, as shown in FIG. 7, the track planning unit 130 defines the travelable area Ap of the own vehicle Va based on the prediction result. For example, the track planning unit 130 may set a region where the existence probability of the moving body is lower than the threshold value as a travelable region Ap. The track planning unit 130 plans a future track so as to be within this travelable area Ap. The track planning unit 130 sequentially provides the generated track plan to the vehicle control ECU 40.

Next, the flow of the orbit generation method executed by the orbit generation device 100 jointly by the functional blocks will be described below according to FIGS. 8 and 9. In the flow described later, "S" means a plurality of steps of the flow executed by a plurality of instructions included in the program.

First, in S10 of FIG. 8, the behavior prediction unit 110 acquires a plurality of prediction results of the moving body based on each prediction model. Next, in S20, the arbitration necessity determination unit 121 determines whether or not arbitration of the prediction result is necessary. If it is determined that arbitration is necessary, the safety evaluation unit 122 evaluates the safety in S30. In the following S40, the comfort level evaluation unit 123 evaluates the comfort level. Further, in S50, the goodness-of-fit evaluation unit 124 evaluates the goodness of fit between the current driving environment and each prediction model. Then, in S60, the determination unit 125 determines the prediction result based on the evaluation results in S30 to S50.

On the other hand, when it is determined in S20 that arbitration of the prediction result is unnecessary, in S70, the determination unit 125 determines the prediction result based on one prediction model. After the processing of S60 or S70, in S80, the track planning unit 130 generates a traveling track of the own vehicle Va based on the prediction result, and then ends a series of processing.

Next, the flow of the goodness-of-fit evaluation method by the goodness-of-fit evaluation unit 124 will be described below with reference to FIG.

First, in S51, the difficulty level of the driving judgment is determined for the current driving environment. Next, in S52, the degree of agreement with the specialized area of the prediction model for the current driving environment is determined. In the following S53, the magnitude of the detection error of the input information to the prediction model is determined. Further, in S54, it is determined whether or not the specific function of the own vehicle Va has failed. Next, in S55, it is determined whether or not the current driving scene is an exception scene. In S56, the weight of each prediction result is determined based on the judgment results of S51 to S55, and the process returns to the flow of FIG.

The above-mentioned S10 is an example of the "behavior prediction process", S20 to S60 are examples of the "arbitration process", and S80 is an example of the "orbit planning process".

According to the above first embodiment, each prediction result is arbitrated based on at least one of the safety level, the comfort level, and the suitability of each prediction model for the driving environment when driving based on each prediction result. , An orbital plan is generated based on the arbitrated forecast results. Therefore, a plurality of prediction results arbitrated based on at least one viewpoint of safety, comfort, and goodness of fit can be utilized for trajectory planning. As a result, it becomes possible to effectively utilize a plurality of prediction results.

Further, according to the first embodiment, the lower the safety level, the more arbitration is performed to increase the contribution of the prediction result to the trajectory plan. Therefore, when the safety level when traveling based on the prediction result is relatively low, the prediction result is more important in the track plan. Therefore, it may be possible to plan an orbit with more emphasis on safety.

Further, according to the first embodiment, the lower the comfort level, the lower the contribution of the prediction result to the activation plan, and the arbitration is executed. In addition, for prediction results where the safety level is out of the permissible range, the decrease in the contribution to the trajectory plan based on the comfort level is stopped. Therefore, as long as the safety level is within the permissible range, the prediction result with a relatively high comfort level is emphasized. This makes it possible to plan a track with higher comfort while ensuring driving safety.

In addition, according to the first embodiment, the goodness of fit is evaluated based on the difficulty level of the driving judgment required in the driving environment, so that the prediction result is arbitrated according to the difficulty level. In particular, in the first embodiment, the higher the difficulty level of the driving judgment, the more important the prediction result by the intention estimation model can be. Therefore, the prediction result can be appropriately arbitrated in a more suitable driving environment of the intention estimation model.

Further, according to the first embodiment, since the goodness of fit is evaluated based on the degree of matching between the driving environment and the assumed environment of each prediction model, the prediction result is arbitrated according to the degree of matching. Therefore, the prediction result can be appropriately arbitrated according to the suitable driving environment of each prediction model.

In addition, according to the first embodiment, the goodness of fit is evaluated based on the magnitude of the detection error of the information input to the prediction model, so that the prediction result is arbitrated according to the detection error. .. In particular, in the first embodiment, the larger the detection error, the lower the contribution of the prediction result by the intention estimation model to the trajectory plan. Therefore, the prediction result can be appropriately arbitrated when an error in intention estimation may occur.

Further, according to the first embodiment, since the goodness of fit is evaluated based on the availability of the specific function of the own vehicle, the prediction result is arbitrated according to the availability. In particular, in the first embodiment, when the specific function cannot be used, the contribution of the prediction result by the intention estimation model to the trajectory plan is reduced. Therefore, the prediction result can be appropriately arbitrated even in a driving environment where the intention estimation model is not good.

Further, according to the first embodiment, the goodness of fit is evaluated and the evaluation result is determined based on whether or not the current driving scene is a non-target scene that is not the target of estimation of the motion intention of the own vehicle. The prediction results are arbitrated. In particular, in the first embodiment, when the current driving scene is a non-target scene, the contribution of the prediction result by the intention estimation model to the track plan is reduced. Therefore, the prediction result can be appropriately arbitrated in the driving environment not targeted by the intention estimation model.

Further, according to the first embodiment, it is determined whether or not arbitration of each prediction result is necessary, and if it is determined that it is not necessary, a specific prediction result among a plurality of prediction results is predicted to contribute to the trajectory plan. It is confirmed as a result. Therefore, if mediation of forecast results is not necessary, only specific forecast results are used for orbit planning. Therefore, the computational load required for arbitration of the prediction result can be reduced.

(Other embodiments)
The disclosure herein is not limited to the exemplary embodiments. Disclosures include exemplary embodiments and modifications by those skilled in the art based on them. For example, the disclosure is not limited to the parts and / or combinations of elements shown in the embodiments. Disclosure can be carried out in various combinations. Disclosures can have additional parts that can be added to the embodiments. Disclosures include those in which the parts and / or elements of the embodiment are omitted. Disclosures include the replacement or combination of parts and / or elements between one embodiment and another. The technical scope disclosed is not limited to the description of the embodiments. Some technical scopes disclosed are indicated by the description of the scope of claims and should be understood to include all modifications within the meaning and scope equivalent to the description of the scope of claims. ..

In the above-described embodiment, the trajectory generator 100 arbitrates each prediction result based on all of safety, comfort, and goodness of fit, but it may be configured to arbitrate each prediction result based on at least one. Just do it.

The orbit generator 100 may be a dedicated computer configured to include at least one of a digital circuit and an analog circuit as a processor. Here, particularly digital circuits include, for example, ASIC (Application Specific Integrated Circuit), FPGA (Field Programmable Gate Array), SOC (System on a Chip), PGA (Programmable Gate Array), CPLD (Complex Programmable Logic Device) and the like. Of these, at least one. Further, such a digital circuit may include a memory for storing a program.

The orbit generator 100 may be provided by one computer, or a set of computer resources linked by a data communication device. For example, some of the functions provided by the trajectory generator 100 in the above-described embodiment may be realized by another ECU.

100 Track generator, 101 Memory (storage medium), 102 Processor, 110 Behavior prediction unit, 120 Prediction result arbitration unit (arbitration unit), 130 Track planning unit, Va own vehicle.

Claims (20)

It is a track generator that generates a planned running track for the own vehicle (Va).
A behavior prediction unit (110) that acquires prediction results from a plurality of prediction models for the behavior of moving objects around the own vehicle, and
With the arbitration unit (120) that arbitrates each of the prediction results based on at least one of the safety level, comfort level, and the goodness of fit of each prediction model to the driving environment when traveling based on each of the prediction results. ,
Based on the arbitrated prediction result, the track planning unit (130) that executes the plan of the traveling track and the track planning unit (130)
Orbit generator equipped with. The track generator according to claim 1, wherein the arbitration unit performs arbitration in which the lower the safety level, the greater the contribution of the prediction result to the plan. The arbitration unit performs arbitration in which the lower the comfort level, the lower the contribution of the prediction result to the plan, and the prediction result whose safety level is out of the permissible range is based on the comfort level. The orbit generation device according to claim 2, wherein the decrease in the contribution to the plan is stopped. The behavior prediction unit has an intention estimation model that outputs the prediction result after assuming a future change of the motion physical quantity based on the estimation of the motion intention of the moving body, and a future change of the motion physical quantity based on the estimation of the motion intention. The orbit generation device according to any one of claims 1 to 3, wherein the non-intentional non-estimation model that outputs the prediction result without assuming the above, and the orbit generation device that acquires the prediction result from. The track generating device according to claim 4, wherein the arbitration unit evaluates the goodness of fit based on the difficulty of driving judgment required in the driving environment. The track generating device according to claim 4 or 5, wherein the arbitration unit evaluates the goodness of fit based on the degree of matching between the traveling environment and the assumed environment of each prediction model. The orbit generation device according to any one of claims 4 to 6, wherein the arbitration unit evaluates the goodness of fit based on the magnitude of a detection error of information input to the prediction model. The track generating device according to any one of claims 4 to 7, wherein the arbitration unit evaluates the goodness of fit based on the availability of a specific function of the own vehicle. The arbitration unit has claims 4 to 8 for evaluating the goodness of fit based on whether or not the current driving scene is a non-target scene that is not subject to the estimation of the motion intention of the own vehicle. The orbital generator according to any one of the following items. The arbitration unit determines whether or not arbitration of each of the prediction results is necessary, and if it is determined that the arbitration is not necessary, the specific prediction result among the plurality of prediction results is used as the prediction result contributing to the plan. The orbital generator according to any one of claims 1 to 9 to be determined. A track generation method executed by a processor (102) to generate a planned travel track for the own vehicle (Va).
A behavior prediction process (S10) for acquiring prediction results from a plurality of prediction models for the behavior of moving objects around the own vehicle, and
An arbitration process (S20, S30) that arbitrates each of the prediction results based on at least one of the safety level, the comfort level, and the suitability of each prediction model for the driving environment when driving based on each of the prediction results. , S40, S50, S60),
Based on the arbitrated prediction result, the track planning process (S80) for executing the plan of the traveling track and the track planning process (S80).
Orbit generation method including. The trajectory generation method according to claim 11, wherein in the arbitration process, the lower the safety level, the more the contribution of the prediction result to the plan is arbitrated. In the arbitration process, the lower the comfort level, the lower the contribution of the prediction result to the plan is performed, and the prediction result whose safety level is out of the allowable range is based on the comfort level. The orbital generation method according to claim 12, wherein the decrease in the contribution to the plan is stopped. In the behavior prediction process, an intention estimation model that outputs the prediction result after assuming a future change of the motion physical quantity based on the estimation of the motion intention of the moving body, and a future change of the motion physical quantity based on the estimation of the motion intention. The orbital generation method according to any one of claims 11 to 13, wherein the prediction result is output from the non-intentional non-estimation model without assuming. The track generation method according to claim 14, wherein in the arbitration process, the goodness of fit is evaluated based on the difficulty of driving judgment required in the driving environment. The track generation method according to claim 14 or 15, wherein in the arbitration process, the goodness of fit is evaluated based on the degree of matching between the traveling environment and the assumed environment of each prediction model. The trajectory generation method according to any one of claims 14 to 16, wherein in the arbitration process, the goodness of fit is evaluated based on the magnitude of the detection error of the information input to the prediction model. Claims 14 to 17 evaluate the goodness of fit based on whether or not the current driving scene is a non-target scene that is not subject to the estimation of the motion intention of the own vehicle in the arbitration process. The orbital generation method according to any one of the following items. In the arbitration process, it is determined whether or not arbitration of each of the prediction results is necessary, and if it is determined that the arbitration is not necessary, the specific prediction result among the plurality of prediction results is used as the prediction result contributing to the plan. The orbital generation method according to any one of claims 11 to 18, which is determined. A track generation program that is stored in a storage medium (101) and includes an instruction to be executed by a processor (102) in order to generate a travel track scheduled for the own vehicle (Va).
The command is
A behavior prediction process (S10) for acquiring prediction results from a plurality of prediction models for the behavior of moving objects around the own vehicle, and
An arbitration process (S20, S30) that arbitrates the prediction results based on at least one of the safety level and comfort level when driving based on the prediction results and the suitability of each prediction model for the driving environment. , S40, S50, S60),
Based on the arbitrated prediction result, the track planning process (S80) for executing the plan of the traveling track and the track planning process (S80).
Orbit generation program including.

Images