Travel trajectory planning system

Abstract

To determine a planned trajectory, taking into consideration a trajectory traveled by another vehicle traveling on a travel lane at the time of merging.SOLUTION: A travel trajectory planning system 10 includes: a sensing section 200 for sensing a situation around an own vehicle VH; a trajectory predicting section 152 for outputting a plurality of predicted trajectory candidates for each of other vehicles which is included in one or more vehicles 500 other than the own vehicle, detected by the sensing section around the own vehicle; a searching section 160 for searching for a solution to a combination optimization problem formulated for selecting an optimum combination which is a combination related to trajectory candidates obtained by selecting two or more trajectory candidates, for each of the other vehicles, from the plurality of predicted trajectory candidates; and a trajectory planning section 154 for determining a planned trajectory to be traveled by the own vehicle, by using the solution which has been searched for by the searching section.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to a driving trajectory planning system. [Background technology]

[0002] The technology described in Patent Document 1 relates to a technology that allows a vehicle to automatically merge onto a main lane. In the technology described in Patent Document 1, at a predetermined position before the merging point, whether or not automatic merging is possible is determined based on the relative speed, relative distance, etc. of other vehicles that may interfere with the merging operation. If it is determined that automatic merging is not possible, measures such as deceleration and handing over of driving responsibility to the driver are taken. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 2019-149144 Summary of the Invention [Problem to be solved by the invention]

[0004] In the technology described in Patent Document 1, the relative speed, relative distance, etc. of other vehicles at the time of determining whether automatic merging is possible are used to determine whether automatic merging is possible. Although automatic merging may be possible depending on the driving conditions of other vehicles after the time of determination, the technology described in Patent Document 1 is likely to miss opportunities to execute automatic merging because it does not take into account the trajectories of other vehicles traveling in the same lane. [Means for solving the problem]

[0005] According to one embodiment of the present disclosure, there is provided a trajectory planning system (10, 10d, 10e) including: a sensing unit (200, 130, 140) that senses a surrounding situation of a host vehicle (VH); a trajectory prediction unit (152) that outputs a plurality of predicted trajectory candidates for each of one or more other vehicles (500) other than the host vehicle that are detected around the host vehicle by the sensing unit; a search unit (160) that searches for a solution to a combinatorial optimization problem formulated to select an optimal combination of trajectory candidates obtained by selecting two or more trajectory candidates for each of the other vehicles from the plurality of predicted trajectory candidates; and a trajectory planning unit (154) that determines a planned trajectory for the host vehicle to travel using the solution searched for by the search unit.

[0006] According to the above-described trajectory planning system, the planned trajectory of the vehicle is determined by taking into account two or more trajectories that other vehicles are expected to travel in the future. Therefore, it is considered possible to execute a trajectory plan that is robust against the uncertainty of the future trajectories of other vehicles. It is also possible to prevent unnecessary deceleration control, etc., from being performed even if the vehicle has been able to automatically merge. [Brief explanation of the drawings]

[0007] [Figure 1] 1 is a block diagram showing a schematic configuration of a travel trajectory planning system according to a first embodiment. [Figure 2] 10 is a flowchart of a process executed by the travel trajectory planning system. [Figure 3] FIG. 10 is an explanatory diagram of generation of trajectory candidates. [Figure 4] FIG. 10 is an explanatory diagram illustrating generation of trajectory candidates for each type of travel. [Figure 5] FIG. 10 is an explanatory diagram of a method for assigning a plurality of trajectory candidates to variables. [Figure 6] 10 is a flowchart of a process executed by the traveling trajectory planning system in the second embodiment. [Figure 7]FIG. 10 is a block diagram illustrating a schematic configuration of a travel trajectory planning system according to a fourth embodiment. [Figure 8] FIG. 1 is an explanatory diagram showing a vehicle traveling on a rampway as seen from above. [Figure 9] FIG. 10 is an explanatory diagram showing a schematic configuration of a travel trajectory planning system according to a fifth embodiment. [Figure 10] FIG. 10 is an explanatory diagram showing an example in which virtual vehicles are arranged around a vehicle. [Figure 11] 10 is a flowchart of a process executed by the travel trajectory planning system. [Figure 12] FIG. 10 is an explanatory diagram showing an example of trajectory candidates generated for each vehicle. [Figure 13] FIG. 13 is an explanatory diagram of a method for arranging virtual pedestrians in the sixth embodiment. [Figure 14] FIG. 13 is an explanatory diagram of a method for arranging virtual vehicles in a seventh embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0008] A. First embodiment: A1. Overview of the driving trajectory planning system: 1, a trajectory planning system 10 is mounted on a vehicle VH. The vehicle VH is an electric vehicle powered by a motor.

[0009] The vehicle VH performs autonomous driving using the detection results of the sensor group 200. The sensor group 200 senses the surrounding conditions and state of the vehicle VH. The sensor group 200 includes a camera, a LiDAR (Light Detection and Ranging) device, and a millimeter-wave radar installed in the vehicle VH. The camera captures a predetermined area outside the vehicle VH at a predetermined frame rate. The LiDAR device detects objects around the vehicle VH by emitting laser light to a predetermined area outside the vehicle VH and receiving the reflected waves. The millimeter-wave radar detects objects around the vehicle VH by emitting radio waves to a predetermined area outside the vehicle VH and receiving the reflected waves. The sensor group 200 also includes a GNSS receiver, a gyro sensor, and a vehicle speed sensor, which will be described later. The sensor group 200 is also referred to as the "sensing unit."

[0010] For example, the vehicle VH may be capable of automated driving at levels 3 to 5 as defined by the Society of Automotive Engineers (SAE). The trajectory planning system 10 plans a trajectory along which the autonomously driven vehicle VH will travel. The vehicle VH travels along the trajectory planned by the trajectory planning system 10. In this embodiment, a scenario is assumed in which the autonomously driven vehicle VH merges into a driving lane. The automatic merging of the vehicle VH into a driving lane is also referred to as automatic merging. The driving lane is also referred to as a main lane.

[0011] The vehicle VH includes a vehicle control device (not shown) for controlling each part of the vehicle VH, and an actuator group (not shown) including one or more actuators that are driven under the control of the vehicle control device. The actuator group includes a drive device actuator for accelerating the vehicle VH, a steering device actuator for changing the direction of travel of the vehicle VH, and a braking device actuator for decelerating the vehicle VH. The drive device includes a battery, a traction motor powered by the battery power, and drive wheels rotated by the traction motor. The vehicle control device included in the vehicle VH controls each part of the vehicle VH so that the vehicle VH travels along the trajectory output by the travel trajectory planning system 10.

[0012] 1, the trajectory planning system 10 includes a trajectory planning device 100 and a sensor group 200. The trajectory planning device 100 includes a memory 110, an input / output interface 120, a processor 150, and a quantum-inspired machine 160.

[0013] The memory 110 and the input / output interface 120 are connected to the processor 150 via a bus 191. The quantum-inspired machine 160 is connected to the processor 150 via a bus 192. A sensor group 200 is connected to the input / output interface 120.

[0014] The memory 110 stores programs and data used for various processes executed by the processor 150. In this embodiment, the memory 110 stores a program PG and map data DT.

[0015] The processor 150 includes at least one of a central processing unit (CPU), a graphics processing unit (GPU), and a neural processing unit (NPU). The processor 150 may be realized by an in-vehicle system on chip (SoC).

[0016] The processor 150 executes the program PG stored in the memory 110 to function as a position detection unit 151, a trajectory prediction unit 152, a search processing unit 153, and a trajectory planning unit 154.

[0017] The position detection unit 151 determines the current position of the vehicle VH based on received signals from positioning satellites of the GNSS (Global Navigation Satellite System), detection results of a gyro sensor and a vehicle speed sensor, etc. A receiver that receives signals from positioning satellites of the GNSS and determines the position based on the received signals, the gyro sensor, and the vehicle speed sensor are included in the sensor group 200. Furthermore, the position detection unit 151 detects the current position of the vehicle VH using the determined position information and map data DT.

[0018] The position detection unit 151 also detects other vehicles present around the vehicle VH based on the detection results of the sensor group 200. One or more other vehicles present around the vehicle VH are referred to as vehicles 500. Furthermore, the position detection unit 151 detects the position and speed of each vehicle 500 based on the detection results of the sensor group 200. The position detection unit 151 outputs information representing the detected position and speed of each vehicle 500 to the trajectory prediction unit 152. The vehicles 500 present around the vehicle VH are also referred to as "other vehicles."

[0019] The trajectory prediction unit 152 predicts multiple trajectory candidates for each of one or more vehicles 500 (other vehicles) present around the vehicle VH. In this embodiment, a situation is assumed in which less than 10 vehicles 500 are detected around the vehicle VH.

[0020] The trajectory candidates for the vehicle 500 predicted by the trajectory prediction unit 152 indicate trajectories of less than 100 meters that the vehicle 500 is expected to travel from the current time onward. For example, the trajectory prediction unit 152 predicts 20 or more trajectory candidates for each vehicle 500.

[0021] The search processing unit 153 causes the quantum-inspired machine 160 to execute processing to minimize an objective function formulated as a combinatorial optimization problem. The quantum-inspired machine 160 searches for a solution, and determines a combination of trajectory candidates obtained by selecting two or more trajectory candidates for each vehicle 500.

[0022] The trajectory planning unit 154 determines a planned trajectory for the vehicle VH to travel using the solution searched for by the quantum-inspired machine 160. The trajectory planning unit 154 outputs trajectory information indicating the determined planned trajectory to the vehicle control device of the vehicle VH.

[0023] The quantum-inspired machine 160 serves as a coprocessor. The quantum-inspired machine 160 is a type of Ising machine that executes algorithms that mimic quantum behavior on a classical computer. The quantum-inspired machine 160 is realized by a quantum-inspired processing unit (QiPU). The quantum-inspired processing unit is realized by an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), a GPU (Graphics Processing Unit), or the like.

[0024] The quantum-inspired machine 160 searches for a solution to the formulated combinatorial optimization problem to select the optimal combination of trajectory candidates. The solution obtained by the search represents a combination of trajectory candidates obtained by selecting MP (MP is an integer equal to or greater than 2) trajectory candidates for each vehicle 500. The quantum-inspired machine 160 is also referred to as a "search unit." MP is also referred to as a "multipath prediction number."

[0025] A2.Processing flow: FIG. 2 shows a flowchart of the processing executed by the trajectory planning system 10. FIG. 3 shows an explanatory diagram of the generation of trajectory candidates. FIG. 3 shows a top view of the vehicle VH traveling on the ramp way L3 and other surrounding vehicles. For example, when the distance between the vehicle VH and the start point SP of the acceleration lane L4 becomes equal to or less than a predetermined distance, the processing shown in FIG. 2 is started. Based on the current position of the vehicle VH detected by the position detection unit 151 and the map data DT stored in the memory 110, the distance between the vehicle 500 and the start point SP of the acceleration lane L4 is determined. The acceleration lane L4 is also called the "merging lane."

[0026] In step S110, the trajectory prediction unit 152 generates a plurality of trajectory candidates for each of the vehicles 500 surrounding the vehicle VH.

[0027] First, other vehicles 500 present around vehicle VH are detected based on the detection results of the sensor group 200. In the example shown in Fig. 3, vehicles 500a and 500b traveling in the driving lane L1 into which vehicle VH will merge, and vehicle 500c traveling in the passing lane L2 adjacent to driving lane L1 are detected. Hereinafter, vehicles 500a to 500c may be simply referred to as vehicles 500. The vehicles 500 to be processed do not include vehicles traveling in a direction different from the direction in which vehicle VH will be traveling in the future. Specifically, vehicles traveling in an oncoming lane (not shown) adjacent to the passing lane L2 are not processed.

[0028] Furthermore, the position and speed of each vehicle 500 are detected based on the detection results of the sensor group 200. Subsequently, multiple trajectory candidates are generated based on the detected position and speed of each vehicle 500. In FIG. 3, the trajectory candidates generated for each vehicle 500 are represented by dashed arrows. In the illustrated example, five trajectory candidates are generated for each vehicle 500. To facilitate understanding of the technology, the dashed arrows representing the five trajectory candidates are shown side by side, but the trajectory candidates actually generated overlap each other in the vicinity of the vehicle 500. Here, the number of trajectory candidates generated is set to five. However, it is desirable to generate more trajectory candidates. Preferably, 10 or more trajectory candidates, more preferably 20 or more trajectory candidates may be generated.

[0029] The trajectory candidate represents a predicted trajectory of the vehicle 500 from the current time until a predetermined time has elapsed. For example, a predicted trajectory of the vehicle 500 from the current time until 60 seconds have elapsed is shown. The trajectory candidate is formed by arranging, along a time axis, a plurality of points to which the target vehicle 500 is predicted to move from the current time. The plurality of points to which the vehicle 500 is predicted to move represents, for example, a point that the vehicle 500 is predicted to arrive at every second that elapses. Each point is expressed by a two-dimensional coordinate value. The coordinate value representing each point may be associated with time information that represents a predicted time at which the vehicle 500 will arrive at that position. The predicted time is expressed, for example, as the time elapsed from the current time.

[0030] The vehicle 500 is assumed to travel at a constant speed in a straight line, accelerate and decelerate, or change lanes. A vehicle 500 traveling at a constant speed in a straight line travels at a constant speed without changing its direction of travel during a certain period of time during which the vehicle 500 is observed (hereinafter referred to as the observation period). Here, a constant speed means that the increase and decrease in speed are within a predetermined range when the current traveling speed is used as a reference. Not changing the direction of travel means continuing to travel in the lane in which the vehicle is currently traveling without changing lanes. A vehicle 500 traveling at an accelerating and decelerating speed travels by performing at least one of acceleration and deceleration during the observation period. Lane-changing travel is travel in which the vehicle changes lanes at least once.

[0031] FIG. 4 shows an explanatory diagram of a method for predicting the position and speed of a vehicle when traveling at a constant speed in a straight line, accelerating and decelerating, and changing lanes. The point (position) where the vehicle 500 is located at time t is expressed by coordinate values ​​(x(t), y(t)). The x-axis is set along the traveling direction of the vehicle 500. The y-axis is set along the width direction of the vehicle 500. The speed of the vehicle 500 at time t is obtained by time differentiation of the point (position) of the vehicle 500. The position and speed of the vehicle 500 when traveling at a constant speed in a straight line can be calculated using equations (1) and (2). The position and speed of the vehicle 500 when accelerating and decelerating can be calculated using equations (3) and (4). The position and speed of the vehicle 500 when changing lanes can be calculated using equations (5) and (6). x(0) and y(0) represent the position of the vehicle 500 at a reference time (t=0). v x and v y represents the X-axis component and the Y-axis component of the speed of the vehicle 500 at a reference time. x represents the acceleration factor.

[0032] For example, the position and speed of the vehicle 500 from time t1 to time t10 are calculated. Also, the timing of transition between the constant speed straight line running, accelerating / decelerating running, and lane change running can be shifted discretely.

[0033] Furthermore, for example, by varying the following factors for the vehicle 500, a plurality of trajectory candidates can be generated. (a) Speed ​​passing through the starting point SP (b) Passing time of the starting point SP (c) Acceleration from the starting point SP to the end point EP (d) Whether or not the vehicle is changing lanes (e) The location of the lane change when changing lanes.

[0034] Using the set of positions and the set of velocities thus obtained, a plurality of trajectory candidates are generated. Hereinafter, the generated plurality of trajectory candidates may be referred to as a set of trajectory candidates.

[0035] As shown in FIG. 2 , in step S120, the search processing unit 153 causes the quantum-inspired machine 160 to execute a process of minimizing an objective function formulated as a combinatorial optimization problem in order to select MP (MP is an integer equal to or greater than 2) trajectory candidates from a set of trajectory candidates for each vehicle 500. Necessary parameters defining the objective function, etc., are passed to the quantum-inspired machine 160. In this embodiment, an objective function is set that takes into consideration ensuring safety, suppressing the occurrence of traffic congestion, etc. As the quantum-inspired machine 160 searches for a solution, MP (MP is an integer equal to or greater than 2) trajectory candidates are selected for each vehicle 500 from the set of trajectory candidates. The selected MP trajectory candidates indicate trajectories from the set of trajectory candidates that are most likely to be traveled by the corresponding vehicle 500. In this manner, desired MP trajectory candidates are determined for each vehicle 500. The formulation of the combinatorial optimization problem will be described in detail later.

[0036] In step S130, the trajectory planning unit 154 determines a planned trajectory for the vehicle VH to travel using MP (MP is an integer equal to or greater than 2) trajectory candidates determined for each vehicle 500. The planned trajectory is determined using a known trajectory planning algorithm. Known trajectory planning algorithms include, for example, the Monte Carlo tree search method, the Markov decision process, and the dynamic programming method.

[0037] Trajectory information indicating the planned trajectory determined in step S140 is output to the vehicle control device of the vehicle VH. The trajectory information includes multiple points that the vehicle VH will pass on the ramp way L3 and the acceleration lane L4 before merging into the driving lane, and the planned time to reach each point. When the information indicating the planned trajectory is received, the vehicle VH travels along the planned trajectory. The process of determining the planned trajectory is repeated until the vehicle VH merges into the driving lane. Therefore, when new trajectory information is output, the vehicle VH travels according to the new trajectory information.

[0038] In step S150, it is determined whether or not the process should be ended. When the vehicle VH merges into the driving lane, it is determined that the process should be ended (step S150; YES), and the process shown in Fig. 2 is ended. On the other hand, if it is determined that the process should not be ended (step S150; NO), the process from step S110 onwards is executed again.

[0039] A3. Formulation of combinatorial optimization problems: The quantum-inspired machine 160 solves combinatorial optimization problems as quadratic unconstrained binary optimization (QUBO) problems. For this reason, it is necessary to express the objective function in QUBO format. In this embodiment, the Hamiltonian, which is the objective function, is expressed as in equation (M1). "No quadratic constraints" does not mean that only problems without constraints can be solved. Expressing the constraint conditions as quadratic expressions means that the constraint conditions are also reduced to a quadratic expression.

[0040]

number

[0041] The Hamiltonian represents energy, and the state where the value of the Hamiltonian is small corresponds to the optimal solution when solving a combinatorial optimization problem with an Ising machine. Also, solving a problem using an Ising machine corresponds to finding the most stable spin state (combination of spin directions) of the Ising model. q i,j corresponds to the spin direction. In equation (M1), the variable q i,j is a binary variable that takes the value "0" or "1".

[0042] Figure 5 shows multiple orbit candidates as a function of the variable q i,j Here, an example is shown in which the trajectory prediction unit 152 generates five trajectory candidates. i,jThe subscript "i" in the above expression represents a value for identifying the vehicle 500. The subscript "j" represents a value for identifying the trajectory candidate. In the illustrated example, values ​​i=1 to 3 are used to identify the vehicles 500a to 500c. Values ​​j=1 to 5 are used to identify each trajectory candidate. Each generated trajectory candidate is assigned to a variable q i,j For example, the second trajectory candidate of the first vehicle, vehicle 500a, is assigned to q 1,2 is assigned to.

[0043] The quantum-inspired machine 160 selects MP (MP is an integer equal to or greater than 2) trajectory candidates for each vehicle 500. For example, MP is set to 2. In this case, for the vehicle 500a shown in FIG. 5, the searched solution is a solution obtained by dividing the two variables q i,j is "1" and the other three variables q i,j becomes "0". As a result of solving, q i,j The two orbital candidates with q = 1 are selected, and i,j = 0 indicates that the three trajectory candidates were not selected. As a result, the two trajectory candidates assigned to the variables with a value of "1" are selected. The same applies to vehicle 500b and vehicle 500c.

[0044] A specific definition of formula (M1) will be explained below: Formula (M2) is set for L(i,j) of the first term (first-order term) on the right side of formula (M1).

[0045]

number

[0046] Formula (M2) is a cost term. Formula (M2) represents the scheduled time at which the ith vehicle is expected to pass the scheduled point of the jth trajectory candidate for the ith vehicle. Hereinafter, the ith vehicle may be referred to as "vehicle i." The jth trajectory candidate may be referred to as "trajectory j." The scheduled point is, for example, the end of the trajectory candidate. The scheduled time is expressed as the elapsed time from the current time. It is desirable to minimize the value represented by formula (M2). By shortening the time until the vehicle 500 passes the end of the trajectory candidate, the objective of suppressing the occurrence of congestion can be achieved. In this embodiment, it is essential to include the cost term represented by formula (M2) in the objective function.

[0047] Equation (M3) is set to Q(i1,j1), (i2,j2) in the second term (quadratic term) on the right side of equation (M1).

[0048]

number

[0049] Equation (M3) is a cost term. Equation (M3) represents the difference between the expected travel speed of the i1th vehicle traveling along the j1th trajectory candidate and the expected travel speed of the i2th vehicle traveling along the j2th trajectory candidate. Note that the j1th trajectory candidate was generated for the i1th vehicle. The j2th trajectory candidate was generated for the i2th vehicle. Equation (M3) assumes that the i1th vehicle and the i2th vehicle are traveling in the same lane or adjacent lanes. Equation (M3) takes into account the relative speed of the vehicles. If the relative speed is high, it is assumed that one vehicle is traveling faster than the other vehicle. Reducing the relative speed can achieve the goal of improving safety. For this reason, it is desirable to minimize the value represented by Equation (M3). In this embodiment, it is essential to include the cost term represented by Equation (M3) in the objective function.

[0050] Equation (M4) is set as the second term (quadratic term) on the right side of equation (M1).

[0051]

number

[0052] Equation (M4) is a constraint term. Equation (M4) represents the constraint condition that MP trajectory candidates are selected. Note that here, it is assumed that the j-th trajectory candidate of the i-th vehicle is the trajectory candidate of the vehicle 500. As mentioned above, the trajectory candidates are defined as the variable q i,j , so the variable q becomes 1 i,j When the sum of these coincides with the number of predicted multipaths MP, the value expressed by equation (M4) takes the minimum value. In this embodiment, it is essential to include the constraint term expressed by equation (M4) in the objective function. Note that in order to set equation (M4) as the second term (quadratic term) on the right side of equation (M1), equation (M4) is set as follows: i,j It is necessary to transform it into a quadratic equation. i,j ≒(q i,j ) 2 This approximation is done by considering the variable q i,j Since is a binary variable that takes "0" or "1", the variable q i,j This is because squaring the equation gives the same value. Also, constant terms that appear when transforming the equation can be ignored in minimizing equation (M1), so such constant terms are sometimes omitted.

[0053] If two or more cost terms are included in the first term (first-order term) on the right side of equation (M1), the first term is expressed as the sum of two or more cost terms. The same applies to the second term. In addition, each cost term and each constraint term may be multiplied by an arbitrary coefficient.

[0054] In this embodiment, by setting the above formulas (M2) and (M3) as cost terms, it is possible to predict the trajectories of multiple vehicles 500 traveling in the driving lane and the passing lane while ensuring safety and suppressing the occurrence of congestion. Furthermore, by setting formula (M4) as a constraint term, it is possible to predict two or more trajectories for each vehicle 500. The planned trajectory to be traveled by vehicle VH is determined taking into account two or more trajectories that vehicle 500 is expected to travel in the future. Therefore, it is believed that it is possible to execute a trajectory plan that is robust against the uncertainty of the future trajectories of other vehicles.

[0055] Furthermore, because the system takes into account the predicted trajectories of other vehicles traveling in the same lane, it is possible to prevent unnecessary deceleration control, etc., from being performed even when the vehicle VH could have merged automatically. As a result, there is a reduced tendency for the vehicle VH to miss an opportunity to merge automatically. This makes it easier to avoid the vehicle VH stopping due to a failed merge, thereby reducing congestion near the merging area.

[0056] The advantages of selecting two or more trajectory candidates for multiple vehicles 500 are as follows: For example, if one trajectory candidate is selected for each vehicle 500, but the vehicle 500 is traveling on a trajectory different from the selected trajectory candidate just before automatic merging, it may be difficult to smoothly execute automatic merging. However, if two or more trajectory candidates are selected, it is possible to reduce the possibility that the vehicle 500 will travel on a trajectory different from the predicted trajectory.

[0057] In the configuration according to this embodiment, other vehicles 500 can be detected without communicating with the traffic infrastructure or other vehicles, and therefore, operation costs can be reduced.

[0058] B. Second embodiment: In the following description, the configurations different from the first embodiment will be mainly described, and a description of the same configurations will be omitted.

[0059] In this embodiment, the position detection unit 151 outputs information indicating the detected current position and speed of the vehicle VH to the trajectory prediction unit 152. The trajectory prediction unit 152 predicts multiple trajectory candidates for each of one or more vehicles 500 (other vehicles) other than the vehicle VH that are present around the vehicle VH.

[0060] Furthermore, the trajectory prediction unit 152 predicts multiple trajectory candidates for the vehicle VH. The trajectory candidates predicted for the vehicle VH are also referred to as "host vehicle trajectory candidates." The method for generating trajectory candidates for the vehicle VH is the same as the method for generating trajectory candidates for the vehicle 500, which is another vehicle, as described in the first embodiment. For example, multiple trajectory candidates can be generated by varying the following elements for the vehicle VH: (A) Speed ​​to enter acceleration lane L4 (B) Time to enter acceleration lane L4 (C) Acceleration from the starting point SP to the junction point (D) Location of the confluence (E) Entry angle into driving lane L1

[0061] The search processing unit 153 selects MP (MP is an integer greater than or equal to 2) trajectory candidates for each vehicle 500, and further causes the quantum-inspired machine 160 to execute a process of minimizing a formulated objective function in order to select one trajectory candidate for the vehicle VH.

[0062] 6 shows a flowchart of the processing executed by the trajectory planning system 10 in the second embodiment. In step S110b, the trajectory prediction unit 152 generates multiple trajectory candidates for the vehicle VH and each of the vehicles 500 surrounding the vehicle VH. In this embodiment, multiple trajectory candidates are also generated for the vehicle VH.

[0063] In step S120b, the search processing unit 153 causes the quantum-inspired machine 160 to execute a process of minimizing an objective function formulated as a combinatorial optimization problem. Necessary parameters defining the objective function, etc. are passed to the quantum-inspired machine 160. In this embodiment, an objective function formulated as a combinatorial optimization problem is used to select MP (MP is an integer of 2 or greater) trajectory candidates from multiple trajectory candidates for each vehicle 500 and to select one trajectory candidate from multiple trajectory candidates for the vehicle VH. As in the first embodiment, an objective function is set that takes into consideration ensuring safety, reducing the occurrence of traffic congestion, etc. As the quantum-inspired machine 160 finds a solution, MP (MP is an integer of 2 or greater) trajectory candidates for each vehicle 500 and one trajectory candidate for the vehicle VH are selected. The trajectory candidate selected from the multiple trajectory candidates prepared for the vehicle VH is used as the planned trajectory for the vehicle VH. Details of the formulation of the combinatorial optimization problem will be described later.

[0064] The process of step S140 is the same as that of the first embodiment, except that the planned trajectory output to the vehicle VH is determined by the combination search of step S120b, and the process of step S150 is the same as that of the first embodiment.

[0065] As in the first embodiment, the cost term (M2) is set as the first term (first-order term) on the right side of the formula (M1). The cost term (M3) and the constraint term (M4) are set as the second term (second-order term) on the right side of the formula (M1).

[0066] Furthermore, equation (M5) is set as the second term (quadratic term) on the right side of equation (M1), which is different from the first embodiment.

[0067]

number

[0068] Equation (M5) is a constraint term. Equation (M5) expresses the constraint condition that one trajectory candidate is selected when the jth trajectory candidate of the i-th vehicle is the trajectory candidate of the vehicle VH. As mentioned above, the trajectory candidate is defined as the variable q i,j , so the variable q becomes 1 i,j When the number of variables is 1, the value expressed by formula (M5) is the minimum. In this embodiment, it is essential to include the constraint term expressed by formula (M5) in the objective function. In addition, in order to set formula (M5) as the second term (quadratic term) on the right side of formula (M1), formula (M5) is set to the variable q i,j It is necessary to transform it into a quadratic equation. i,j ≒(q i,j ) 2 This approximation is done by considering the variable q i,j Since is a binary variable that takes "0" or "1", the variable q i,j This is because squaring the equation gives the same value. Also, constant terms that appear when transforming the equation can be ignored in minimizing equation (M1), so such constant terms are sometimes omitted.

[0069] In this embodiment, by setting the above formulas (M2) and (M3) as cost terms, it is possible to predict the trajectories of multiple vehicles 500 traveling in the driving lane and the passing lane in a single search while ensuring safety and suppressing the occurrence of congestion. Furthermore, by setting formula (M4) as a constraint term, it is possible to predict two or more trajectories for each vehicle 500. Furthermore, by setting formula (M5) as a constraint term, it is possible to predict one trajectory for vehicle VH. Unlike the first embodiment, it is not necessary to separately determine the planned trajectory of vehicle VH (see step S130 in FIG. 2).

[0070] In this embodiment as well, by predicting two or more trajectories for a plurality of vehicles 500, it is possible to obtain the same advantages as in the first embodiment.

[0071] In this embodiment, the trajectory planning unit 154 is not required. Alternatively, depending on the situation around the vehicle VH, switching may be performed between a mode in which the trajectory planning unit 154 determines a planned trajectory as described in the first embodiment and a mode in which a trajectory candidate selected by processing by the quantum-inspired machine 160 is adopted as the planned trajectory as described in this embodiment. For example, the situation around the vehicle VH is determined depending on conditions such as the number of other vehicles around the vehicle VH, the time of day the vehicle VH is traveling, and the weather.

[0072] C. Third embodiment: Cost terms that can be arbitrarily adopted as necessary will be described below. The following formulas are applicable to the configurations according to the first and second embodiments described above, and the fourth embodiment described later.

[0073] The following equation (M6) can be set for Q(i1,j1), (i2,j2), the second term (quadratic term) on the right side of equation (M1). Equation (M6) represents the cost term.

[0074]

number

[0075] Equation (M6) represents the reciprocal of the difference in inter-vehicle distance between the i1th vehicle traveling on the j1th trajectory candidate and the i2th vehicle traveling on the j2th trajectory candidate. Equation (M6) assumes that the i1th and i2th vehicles are traveling in the same lane or adjacent lanes. Increasing the inter-vehicle distance can achieve the goal of improving safety.

[0076] The following equation (M7) can be set for Q(i1,j1), (i2,j2), the second term (quadratic term) on the right side of equation (M1). Equation (M7) represents the cost term.

[0077]

number

[0078] Regarding formula (M7), the ReLU (Rectified Linear Unit) function always outputs 0 when the input value is less than or equal to 0, and outputs the same value as the input value when the input value is greater than 0. Formula (M7) takes a value of 0 when the distance between the vehicle VH to merge into the driving lane and the vehicle 500 traveling in the driving lane exceeds the limit distance at which the human eye can recognize other vehicles. Formula (M7) takes a positive value when the distance between the vehicle VH to merge into the driving lane and the vehicle 500 traveling in the driving lane does not exceed the limit distance at which the human eye can recognize other vehicles. The term represented by formula (M7) is activated only when the distance between the vehicle VH to merge into the driving lane and the vehicle 500 traveling in the driving lane does not exceed the limit distance at which the human eye can recognize other vehicles.

[0079] The following equation (M8) can be set for L(i,j), the first term (first-order term) on the right side of equation (M1). Equation (M8) represents the cost term.

[0080]

number

[0081] Equation (M8) is the merging point (position S LC ) to the end point EP of the acceleration lane L4 (position S ME ) The merging point is the point where the acceleration lane L4 enters the driving lane L1. From a safety standpoint, it is preferable to merge just before the end point EP of the acceleration lane L4 rather than merging right at the end point EP. The goal of improving safety can be achieved by increasing the distance between the merging point and the end point EP.

[0082] The following equation (M9) can be set for L(i,j), the first term (first-order term) on the right side of equation (M1). Equation (M9) represents the cost term.

[0083]

number

[0084] Equation (M9) represents the vehicle's energy loss over a given time interval. Reducing the value of this term can achieve the goal of improving fuel efficiency.

[0085] If two or more terms are included as the first term (first-order term) on the right-hand side of equation (M1), Q((i1,j1)(i2,j2)) is expressed as the sum of two or more terms. The same applies to the second term. Also, each cost term and each constraint term may be multiplied by an arbitrary coefficient.

[0086] D. Fourth embodiment: 7 shows a schematic configuration of a traveling trajectory planning system 10d according to the fourth embodiment. In the following explanation, the configuration different from the first embodiment will be mainly explained, and the explanation of the similar configuration will be omitted. In this embodiment, the trajectory planning device 100d further includes a first communication unit 130 and a second communication unit 140 in addition to the configuration explained in the first embodiment. The first communication unit 130 and the second communication unit 140 are connected to a processor 150 via a bus 191.

[0087] The first communication unit 130 communicates wirelessly with communication devices mounted on other vehicles, etc., through V2X (Vehicle to X) communication. V2X communication is communication that uses frequency bands for ITS (Intelligent Transport Systems). V2X communication includes V2V (Vehicle to Vehicle) communication and V2I (Vehicle to Infrastructure) communication. The first communication unit 130 communicates with communication devices mounted on other vehicles through V2V communication. Communication between the vehicle VH and other vehicles is called vehicle-to-vehicle communication. The first communication unit 130 communicates with communication devices provided in transportation infrastructure through V2I communication. Communication between the vehicle VH and transportation infrastructure is called road-to-vehicle communication. The vehicle VH can acquire detection results from roadside devices (cameras, LiDAR devices) installed on the side of the road through communication with the transportation infrastructure.

[0088] The second communication unit 140 communicates with communication devices mounted on other vehicles by V2N (Vehicle to Network) communication. V2N communication is communication that uses a communication network of a mobile carrier.

[0089] In the first embodiment, an example has been described in which the sensor group 200 is used to detect vehicles 500 (other vehicles) around the vehicle VH.

[0090] 8 shows a top view of vehicle VH traveling on a rampway. In the example shown, vehicles 500a and 500b traveling in driving lane L1 and passing lane L2 are located at a distance from vehicle VH. For this reason, the sensor group 200 of vehicle VH may not be able to detect other vehicle 500.

[0091] In this embodiment, in order to detect the surrounding situation, a first communication unit 130 and a second communication unit 140 are used in addition to the sensor group 200. The first communication unit 130 and the second communication unit 140 are also called "sensing units."

[0092] For example, the second communication unit 140 acquires sensing results from a sensor provided in the vehicle 500a. Specifically, the second communication unit 140 receives, from the vehicle 500a, position information and speed information of the vehicle 500a measured by the vehicle 500a. The position detection unit 151 detects the position of the vehicle 500a using the position information received from the vehicle 500a and the map data DT. The trajectory prediction unit 152 generates multiple trajectory candidates using the speed information received from the vehicle 500a and the position of the vehicle 500a output by the position detection unit 151. In this case, the trajectory prediction unit 152 may use a value obtained by varying the value representing the position of the vehicle 500a output by the position detection unit 151 based on the position information received from the vehicle 500a as the initial position (current position) of the vehicle 500a for generating the multiple trajectory candidates. In this case, the initial value of the vehicle 500a used to generate the multiple trajectory candidates includes multiple values ​​representing the position.

[0093] The first communication unit 130 also receives detection results from roadside devices installed on the side of the road from the traffic infrastructure. For example, if the roadside device is a LiDAR device, the detection results include the installation position of the LiDAR device and the detection range of the LiDAR device in addition to the captured image. The position detection unit 151 uses the detection results to detect the position and speed of other vehicles 500.

[0094] E. Fifth embodiment: E1. Trajectory planning system processing: 9 shows a schematic configuration of a trajectory planning system 10e according to the fifth embodiment. In the following description, the configuration different from the first embodiment will be mainly described, and a description of the same configuration will be omitted.

[0095] The sensor group 200 provided in the vehicle VH includes, for example, four LiDAR devices for detecting the front, rear, left, and right sides of the vehicle VH, and four stereo cameras for capturing images of the front, rear, left, and right sides of the vehicle VH. For example, the four LiDAR devices are provided on the front bumper, rear bumper, and left and right front fenders of the vehicle VH. The four stereo cameras are provided on the top of the windshield, left and right pillars, and back door. In this embodiment, the four LiDAR devices are used to detect objects around the vehicle VH. The four stereo cameras are used to detect objects around the vehicle VH, white lines on the road, signs, etc.

[0096] In this embodiment, the processor 150 of the trajectory planning device 100e functions as a position detection unit 151, a trajectory prediction unit 152, a search processing unit 153, a trajectory planning unit 154, and a virtual placement unit 155. The virtual placement unit 155 virtually places one or more virtual moving objects in an area around the vehicle VH where the sensor group 200 cannot detect moving objects. In this specification, a moving object refers to an object that is moving. A moving object includes an object that is currently stationary but can move. Examples of moving objects include vehicles, pedestrians, bicycles, AGVs (Automated Guided Vehicles), and AMRs (Autonomous Mobile Robots). In this embodiment, the virtual placement unit 155 places one or more virtual vehicles in an area around the vehicle VH where the sensor group 200 cannot detect moving objects. In this embodiment, a passenger car will be described as an example of the virtual vehicle.

[0097] FIG. 10 shows an example in which virtual vehicles are arranged around vehicle VH. Vehicle VH is traveling in acceleration lane L4. A circular area centered on vehicle VH represents the range in which vehicle VH can detect surrounding moving objects. Hereinafter, the range in which vehicle VH can detect surrounding moving objects is referred to as a detectable area RD. In FIG. 10, the detectable area RD is represented by a dashed circle. The detectable area RD is the range in which moving objects existing around vehicle VH can be detected by at least some of the sensors included in sensor group 200. In the example shown in FIG. 10, the detectable area RD is determined based on the maximum detection distances of the four LiDAR devices included in sensor group 200 and the maximum detection distances of the four stereo cameras. For example, the detectable area RD is a range with a radius of 200 meters centered on vehicle VH. The area outside the detectable area RD is referred to as an undetectable area RN.

[0098] In the example shown in FIG. 10, vehicle 500a is traveling in driving lane L1, where vehicle VH merges, and vehicles 500b and 500c are traveling in passing lane L2 adjacent to driving lane L1. Hereinafter, vehicles 500a to 500c may be simply referred to as vehicle 500. In FIG. 10, the detected actual vehicle 500 is represented by a hollow pentagon. The acute angle included in this pentagon indicates the direction in which vehicle 500 is traveling. In the example shown, vehicle 500a traveling in driving lane L1 prevents sensor group 200 from detecting a moving object in a portion of passing lane L2. In other words, a blind spot is created by vehicle 500a.

[0099] For example, a vehicle in a blind spot from vehicle VH may interfere with the merging operation of vehicle VH. Also, a vehicle outside the detectable area RD may suddenly accelerate and enter the detectable area RD in a short time. Such a vehicle may interfere with the merging operation of vehicle VH.

[0100] Therefore, in this embodiment, the trajectory planning system 10e virtually places a virtual vehicle in an area around the vehicle VH where the sensor group 200 cannot detect a moving object, and predicts the trajectory of the virtual vehicle.

[0101] In Fig. 10, the virtually placed virtual vehicles 700a to 700e are represented by hatched pentagons. The acute angles included in these pentagons indicate the directions in which the vehicles 700a to 700e are traveling. Hereinafter, the virtual vehicles 700a to 700e may be referred to as virtual vehicles 700. The virtual vehicle 700 may also be simply referred to as vehicle 700. The virtual vehicle 700 may also be referred to as a "virtual moving object" or a "virtual moving body."

[0102] The travel trajectory planning system 10e uses the trajectory predicted for the actually detected vehicle 500 and the trajectory predicted for the virtual vehicle 700 to determine a planned trajectory along which the vehicle VH will travel.

[0103] The virtual placement unit 155 determines the traveling direction and traveling speed for each of the virtually placed virtual vehicles 700. The traveling direction and traveling speed determined for the virtual vehicle are used by the trajectory prediction unit 152 in processing to predict multiple trajectory candidates for the virtual vehicle 700. Details of the processing by the virtual placement unit 155 will be described later.

[0104] The trajectory prediction unit 152 predicts a plurality of trajectory candidates for each vehicle 500. In this embodiment, the trajectory prediction unit 152 further predicts a plurality of trajectory candidates for each virtual vehicle 700. The trajectory candidates predicted for the vehicle 700 are also referred to as "virtual trajectory candidates."

[0105] Fig. 11 shows a flowchart of the processing executed by the trajectory planning system. The same processes as those shown in Fig. 2 in the first embodiment are denoted by the same reference numerals. For example, when the vehicle VH enters the acceleration lane L4, the processing shown in Fig. 11 is started.

[0106] In step S110, the trajectory prediction unit 152 generates a plurality of trajectory candidates for each of the vehicles 500 surrounding the vehicle VH. The processing in step S110 is the same as in the first embodiment (see step S110 in FIG. 2).

[0107] In step S112, the virtual placement unit 155 virtually places one or more virtual vehicles in a specific region RP. The specific region RP includes a blind spot region RB and a boundary region RE.

[0108] The blind spot area RB is an area of ​​the detectable area RD where a moving object cannot be detected due to an obstacle. In the example shown in Fig. 10, the sensor group 200 cannot detect a moving object in the blind spot area RB due to the presence of a vehicle 500a in the driving lane L1. In the example shown, the blind spot area RB is an area behind the vehicle 500a as seen from the vehicle VH, within a sector formed by an imaginary line SL1 passing through the center of the vehicle VH and the front end of the vehicle 500a in the driving lane L1, an imaginary line SL2 passing through the center of the vehicle VH and the rear end of the vehicle 500a in the driving lane L1, and an arc sandwiched between the lines SL1 and SL2 that are part of the boundary of the detectable area RD. In Fig. 10, the blind spot area RB is cross-hatched.

[0109] In this embodiment, to facilitate understanding of the technology, a vehicle in the driving lane L1, which is the lane closer to the vehicle VH out of the driving lane L1 and the passing lane L2, is treated as an obstacle. A vehicle in the passing lane L2, which is the lane farther from the vehicle VH out of the driving lane L1 and the passing lane L2, is not considered an obstacle. The same applies to a vehicle in an oncoming lane (not shown) on the opposite side of the passing lane L2 from the side adjacent to the driving lane L1.

[0110] The boundary area RE is a limited area within the undetectable area RN. The boundary area RE is a range that extends away from the detectable area RD by a predetermined distance D1 from the boundary between the detectable area RD and the undetectable area RN. Although the detectable area RD is encroached upon by the blind spot area RB, to facilitate understanding of the technology, the boundary area RE is defined as a range that extends by the distance D1 from the boundary between the detectable area RD and the undetectable area RN in the case where the detectable area RD does not exist. The distance D1 is also referred to as the "first distance." In the illustrated example, the range of the boundary area RE is represented by a dashed circle centered on the vehicle VH.

[0111] The virtual placement unit 155 places one or more virtual vehicles 700 in the center of the lane in the area where the blind spot area RB overlaps with the lane. The area where the blind spot area RB overlaps with the lane is determined based on the map data DT and position information indicating the current position of the vehicle VH measured by the position detection unit 151. In the example shown in the figure, a virtual vehicle 700d is placed in the area where the blind spot area RB overlaps with the lane. The map data DT is also referred to as "map information." Note that the virtual vehicle 700 is not placed in the area of ​​the blind spot area RB that does not overlap with the lane.

[0112] Furthermore, the virtual placement unit 155 places one or more virtual vehicles 700 in the center of the lane in the area where the boundary area RE overlaps with the lane. The area where the boundary area RE overlaps with the lane is determined based on the map data DT and position information indicating the current position of the vehicle VH measured by the position detection unit 151. In the example shown in the figure, vehicles 700a and 700b are placed in the area where the boundary area RE overlaps with the driving lane L1. Vehicles 700c and 700e are placed in the area where the boundary area RE overlaps with the passing lane L2. Note that the virtual vehicles 700 are not placed in the area of ​​the boundary area RE that does not overlap with the lane.

[0113] In the example shown, a virtual vehicle 700 is not placed in the area where the boundary area RE and the ramp way L3 overlap, but a virtual vehicle 700 may be placed in the area where the boundary area RE and the ramp way L3 overlap.

[0114] The virtual placement unit 155 determines the traveling direction of the vehicle 700 based on the position where the virtual vehicle 700 is placed and the map data DT. Specifically, the virtual placement unit 155 determines the traveling direction specified for the lane where the virtual vehicle 700 is placed as the traveling direction of the vehicle 700. Furthermore, the virtual placement unit 155 determines the traveling speed of the vehicle 700 based on laws and regulations regarding vehicle speed limits. For example, the virtual placement unit 155 determines the maximum speed established by laws and regulations regarding vehicle speed limits for the lane where the vehicle 700 is placed as the traveling speed of the vehicle 700. In this case, the traveling speed set for all of the multiple virtual vehicles 700 placed in the specific region RP is the same. In this way, the traveling direction and traveling speed of the virtual vehicle 700 can be easily determined. The virtual placement unit 155 outputs information indicating the position, traveling direction, and traveling speed of each vehicle 700 to the trajectory prediction unit 152.

[0115] As shown in FIG. 11, in step S115, the trajectory prediction unit 152 generates a plurality of trajectory candidates for each of the virtually placed vehicles 700.

[0116] 12 is an explanatory diagram of the trajectory candidates generated for each of the vehicle 500 and the virtual vehicle 700. The trajectory candidates generated for the vehicle 700 represent the trajectory predicted for the vehicle 700 from the present time until a predetermined time has elapsed. The method for generating the trajectory candidates for the vehicle 700 is the same as the method described for the vehicle 500 in the first embodiment.

[0117] As shown in FIG. 11 , in step S120e, the search processing unit 153 causes the quantum-inspired machine 160 to execute a process of minimizing an objective function formulated as a combinatorial optimization problem. Necessary parameters defining the objective function and the like are passed to the quantum-inspired machine 160. In this embodiment, an objective function formulated as a combinatorial optimization problem is used to select MP (MP is an integer equal to or greater than 2) trajectory candidates from multiple trajectory candidates for each vehicle 500, select MP (MP is an integer equal to or greater than 2) trajectory candidates from multiple trajectory candidates for each vehicle 700, and select one trajectory candidate from multiple trajectory candidates for vehicle VH. As in the first embodiment, an objective function is set that takes into consideration ensuring safety, reducing the occurrence of traffic congestion, and the like. As the quantum-inspired machine 160 finds a solution, MP (MP is an integer equal to or greater than 2) trajectory candidates for each vehicle 500, MP (MP is an integer equal to or greater than 2) trajectory candidates for each vehicle 700, and one trajectory candidate for vehicle VH are selected. One of the trajectory candidates selected for the vehicle VH is used as the planned trajectory that the vehicle VH will travel.

[0118] The processing in step S140 is the same as in the first embodiment. However, the planned trajectory output to the vehicle VH is determined by the combination search in step S120e. Trajectory information indicating the determined planned trajectory is output to the vehicle control device of the vehicle VH. When the information indicating the planned trajectory is received, the vehicle VH travels along the planned trajectory.

[0119] In step S150, it is determined whether or not the process should be ended. When the vehicle VH merges into the driving lane, it is determined that the process should be ended (step S150; YES), and the process shown in Fig. 11 is ended. On the other hand, if it is determined that the process should not be ended (step S150; NO), the process from step S110 onwards is executed again.

[0120] In this embodiment, too, the planned trajectory of the vehicle VH may be determined using a known trajectory planning algorithm, as described in the first embodiment. In this case, in step S120e, the search processing unit 153 selects MP (MP is an integer greater than or equal to 2) trajectory candidates from a set of trajectory candidates for each vehicle 500, and causes the quantum-inspired machine 160 to execute a process of minimizing an objective function formulated as a combinatorial optimization problem in order to select MP (MP is an integer greater than or equal to 2) trajectory candidates from the set of trajectory candidates for each vehicle 700. The trajectory planning unit 154 determines a planned trajectory for the vehicle VH to travel using a known trajectory planning algorithm, based on the MP (MP is an integer greater than or equal to 2) trajectory candidates determined for each vehicle 500 and the MP (MP is an integer greater than or equal to 2) trajectory candidates determined for each vehicle 700.

[0121] E2. Formulation of combinatorial optimization problems: The method of formulating the objective function is as explained in the first to third embodiments. In this embodiment, the virtual vehicle 700 is treated in the same way as the vehicle 500. For this reason, the trajectory candidates of the vehicle 500 and the virtual vehicle 700 are represented by the variable q i,j Assign to q i,j The subscript "i" represents a value that identifies the vehicle 500 or the virtual vehicle 700. The subscript "j" represents a value that identifies the trajectory candidate.

[0122] In this embodiment, the following formula (M10) is a cost term that can be adopted as the second term (quadratic term) on the right side of formula (M1).

[0123]

number

[0124] Equation (M10) represents the compatibility between the trajectory of the leading vehicle and the trajectory of the following vehicle. By increasing the value represented by this term, the possibility of a collision between the leading vehicle and the following vehicle can be reduced. TTC represents the time to collision. The combinations of the leading vehicle (leading vehicle) and the following vehicle (following vehicle) are as follows. Note that in the following, vehicle 500 represents any of vehicles 500a to 500c, and vehicle 700 represents any of vehicles 700a to 700e. (1) Leading vehicle: Vehicle VH Following vehicle: Vehicle 500 (2) Leading vehicle: vehicle VH Following vehicle: virtual vehicle 700 (3) Leading vehicle: Vehicle 500, Following vehicle: Vehicle VH (4) Leading vehicle: Vehicle 500 Following vehicle: Vehicle 500 (different from the leading vehicle) (5) Leading vehicle: vehicle 500 Following vehicle: virtual vehicle 700 (6) Leading vehicle: Virtual vehicle 700 Following vehicle: Vehicle VH (7) Leading vehicle: Virtual vehicle 700 Following vehicle: Vehicle 500 (8) Leading vehicle: Virtual vehicle 700 Following vehicle: Virtual vehicle 700 (different from the leading vehicle)

[0125] In equation (M10), the position S of the preceding vehicle in the same lane l and the position of the following vehicle S f Among these, the position of the leading vehicle S l The value representing the position of the following vehicle S f The fixed value A is set to a value greater than the value representing the timeout. Any value can be set as the fixed value A, but it is desirable to set a value that is somewhat large. For example, the fixed value A is set to 20 (seconds).

[0126] Cost expressed by formula (M10) TTC is the reciprocal of the minimum collision margin time for each fixed time interval from the present until a fixed period has elapsed, multiplied by the coefficient W TTC Therefore, Cost TTCIt is desirable to maximize the value representing the time to collision (TTC). A certain period is, for example, 5 seconds. A certain time interval is, for example, 1 second. For example, the minimum value of the time to collision predicted 1 second, 2 seconds, ..., 5 seconds from the current time is set to the value of min(TTC).

[0127] If the preceding vehicle is actually the detected vehicle 500, a coefficient W for modifying the likelihood of the trajectory prediction is TTC If the preceding vehicle is the virtual vehicle 700, the coefficient W TTC The value of "0 or more and 1.0 or less" is set to the coefficient W TTC Setting W closer to "0" means that the time to collision is not important. TTC Increasing W to "1.0" means that the collision margin time is emphasized, and in this case, a more careful trajectory planning is performed. TTC is also called the "weighting factor."

[0128] When the preceding vehicle is the virtual vehicle 700, the coefficient W TTC can be a fixed value less than or equal to 1, or the coefficient W TTC may be a value that changes depending on the situation of the vehicle 700.

[0129] When the preceding vehicle is the virtual vehicle 700, the coefficient W TTC The coefficient W may be changed depending on the traveling speed of the leading vehicle, the distance between the leading vehicle and the trailing vehicle, and the traveling speed of the trailing vehicle. TTC Bringing it closer to "1.0" means that the virtual vehicle 700 is given more importance. By placing more importance on the virtual vehicle 700 that has not yet been detected at this time, it is possible to carry out careful trajectory planning. By weighting in accordance with the traveling speed of the preceding vehicle, the distance between the vehicles, and the traveling speed of the following vehicle, it is possible to carry out trajectory planning in accordance with the traveling conditions of the vehicle 700.

[0130] When the preceding vehicle is the virtual vehicle 700, the coefficient W TTCmay be changed depending on the environmental conditions of the vehicle VH that may affect the sensing results of the sensor group 200. For example, the accuracy of the sensing results of the sensor group 200 may be reduced depending on the weather around the vehicle VH.

[0131] The LiDAR device included in the sensor group 200 uses infrared light. Infrared light has the property of being absorbed and scattered when it comes into contact with water. For this reason, the detectable distance of the LiDAR device is shortened in weather conditions such as rain, fog, and snow. Furthermore, the amount of information contained in an image captured by a camera included in the sensor group 200 at night when there is little light is less than the amount of information contained in an image captured during the day when there is more light.

[0132] In this way, the accuracy of the sensing results by the sensor group 200 may decrease depending on the weather around the vehicle VH. Even if the detectable region RD is narrowed depending on the degree of decrease in the accuracy of the sensing results, it is not easy to quantify the degree of decrease in the accuracy of the sensing results by the sensor group 200 due to weather such as rain, fog, or snow. Therefore, if the accuracy of the sensing results by the sensor group 200 may decrease and the preceding vehicle is the virtual vehicle 700, the coefficient W TTC By increasing the weighting on , careful trajectory planning can be performed.

[0133] As described above, in this embodiment, a virtual vehicle is placed in a specific region RP where the sensor group 200 cannot detect moving objects, and a combination solution of predicted trajectory candidates for other actually detected vehicles and the virtual vehicle is searched for. Even if another vehicle is currently outside the detectable region RD, which is the region detectable by the sensor group 200, it is possible that the other vehicle will enter the detectable region RD in the near future depending on the vehicle's traveling speed. It is also possible that a vehicle traveling in the blind spot region RB may exist. In this embodiment, a combination solution of trajectory candidates is searched for using a set of predicted trajectory candidates for the virtual vehicle 700 located outside the region detectable by the sensor group 200 in addition to a set of predicted trajectory candidates for the actually detected vehicle 500. This enables, for example, trajectory planning to be performed that is robust against the uncertainty of the future trajectories of other vehicles located in the blind spot of the vehicle VH (host vehicle). This prevents other vehicles located outside the region detectable by the sensor group 200 from interfering with the smooth execution of automatic merging by the vehicle VH.

[0134] F. Sixth embodiment: In the fifth embodiment, an example of a passenger car has been described as the vehicle 700, which is a virtual moving object. However, the virtual moving object is not limited to this, and may be a two-wheeled vehicle or a three-wheeled vehicle. Furthermore, the virtual moving object may include a pedestrian on a sidewalk adjacent to the road on which the vehicle VH is traveling, or a bicycle on a bicycle lane adjacent to the road on which the vehicle VH is traveling. Here, pedestrians include not only pedestrians walking on the sidewalk but also pedestrians standing on the sidewalk. Bicycles include not only bicycles traveling on the bicycle lane but also bicycles stopped on the bicycle lane.

[0135] 13 is an explanatory diagram of a method for arranging virtual pedestrians in the sixth embodiment. The position detection unit 151 detects pedestrians and bicycles around the vehicle VH, as well as other vehicles 500 around the vehicle VH, based on the detection results of the sensor group 200. In the illustrated example, a vehicle 500d parked at the left edge of the driving lane L1 and pedestrians 550a and 550b walking on the sidewalk L5 adjacent to the driving lane L1 are detected.

[0136] In the illustrated example, a blind spot is created by vehicle 500d parked at the left edge of travel lane L1. In this case, sensor group 200 cannot detect a moving object in blind spot area RB due to vehicle 500d. Therefore, virtual placement unit 155 virtually places a virtual pedestrian 750 as a virtual moving object in blind spot area RB for vehicle 500d. Virtual placement unit 155 determines the traveling direction and traveling speed of virtual pedestrian 750. For example, virtual placement unit 155 determines the traveling direction of pedestrian 550a or 550b detected around vehicle VH as the traveling direction of virtual pedestrian 750. Furthermore, virtual placement unit 155 determines the walking speed of pedestrian 550a or 550b detected around vehicle VH as the walking speed of virtual pedestrian 750. In this way, the traveling direction and traveling speed of the virtual pedestrian can be easily determined. The virtual placement unit 155 outputs information indicating the position, moving direction, and walking speed of the virtual pedestrian 750 to the trajectory prediction unit 152. The virtual pedestrian 750 is also called a "virtual pedestrian."

[0137] The trajectory prediction unit 152 predicts multiple trajectory candidates for each of the detected vehicle 500d and pedestrians 550a and 550b. The trajectory prediction unit 152 also predicts multiple trajectory candidates for a virtual pedestrian 750. The method for generating the trajectory candidates predicted for the vehicle 500d is the same as in the first embodiment. The trajectory prediction unit 152 generates multiple trajectory candidates predicted for each of the pedestrians 550a and 550b and the virtual pedestrian 750 using a mathematical model. For example, a social force model can be used as the mathematical model. When predicting multiple trajectory candidates for the vehicle VH, the trajectory prediction unit 152 generates the multiple trajectory candidates for the vehicle VH so that the multiple trajectory candidates for the vehicle VH do not intersect with either the multiple trajectory candidates for the pedestrians 550a and 550b or the multiple trajectory candidates for the virtual pedestrian 750. For example, a pedestrian may be in a blind spot created by a stopped vehicle. By virtually placing a virtual pedestrian 750 in a blind spot, it is possible to predict a sudden sudden jump of a pedestrian or the like that is not detected by the sensor group 200, and to execute careful trajectory planning.

[0138] G. Seventh embodiment: 14 is an explanatory diagram of a method for arranging a virtual vehicle 700 in the seventh embodiment. The virtual arranging unit 155 does not need to arrange the virtual vehicle 700 in an area where it is clear that no vehicle exists even within the specific area RP.

[0139] Prior to the process of placing the virtual vehicle 700, the virtual placement unit 155 determines whether or not the virtual vehicle 700 needs to be placed. In this embodiment, the sensor group 200 further includes a telephoto camera. The telephoto camera is provided, for example, above the windshield. The telephoto camera captures an image ahead of the vehicle VH. The virtual placement unit 155 determines whether or not the virtual vehicle 700 needs to be placed based on an image captured by the telephoto camera.

[0140] In the illustrated example, the virtual placement unit 155 determines whether or not to place the virtual vehicle 700 in a candidate range for placing the virtual vehicle 700. The candidate range for placing the virtual vehicle 700 is a range where the boundary region RE and the lane overlap.

[0141] FIG. 14 shows a range R1 where the boundary region RE and the driving lane L1 overlap in front of the vehicle VH as a candidate range for locating the virtual vehicle 700. The range R1 is cross-hatched. Based on the image captured by the telephoto camera, the virtual placement unit 155 determines not to locate the virtual vehicle 700 in the range R1 if a white line on the road is detected on the far side of the range R1 on a virtual straight line SL3 extending from the vehicle VH and passing through the center of the range R1. Detection of a white line on the road on the far side of the range R1 on the virtual straight line SL3 passing through the center of the range R1 clearly indicates that no vehicle is present in the range R1. FIG. 14 shows an example in which the telephoto camera is installed above the windshield and a virtual straight line SL2 extends from the front end of the vehicle VH. In the illustrated example, a white line is detected in the range R2 represented by the two-dot chain ellipse. On the other hand, if no white line on the road surface is detected on a virtual straight line SL2 that extends from the vehicle VH and passes through the center of the range R1, the virtual placement unit 155 determines that the virtual vehicle 700 is to be placed in the range R1. Note that the range R2 is a range in which moving objects, including other vehicles, cannot be detected due to limitations in the performance of the sensor group 200 to detect moving objects, but white lines, etc. can be detected.

[0142] 14 shows the range R1 in front of the vehicle VH where the boundary area RE and the driving lane L1 overlap as a candidate range for placing the virtual vehicle 700, but the virtual placement unit 155 can similarly determine whether to place the virtual vehicle 700 in the range in front of the vehicle VH where the boundary area RE and the overtaking lane L2 overlap. In this case, the white line to be detected is the white line on the outer side of the roadway that separates the shoulder or side of the road from the overtaking lane L2.

[0143] In this way, in this embodiment, the virtual vehicle 700 is not placed in an area where it is clear that no moving object is present based on the detection results of any of the sensors included in the sensor group 200. By not placing the virtual vehicle 700 in an area where it is clear that no moving object is present, processing such as the generation of trajectory candidates is not performed for unnecessary virtual vehicles 700, thereby improving processing efficiency.

[0144] H. Eighth embodiment: In the fifth embodiment, an example has been described in which the maximum speed established by laws and regulations regarding vehicle speed limits is determined as the travel speed of the virtually placed vehicle 700. However, the travel speed of the virtually placed vehicle 700 may be determined by other methods. The travel speed of another vehicle 500 detected around the vehicle VH may be determined as the travel speed of the virtually placed vehicle 700. For example, the travel speed of a vehicle 500 traveling in front of the virtual vehicle 700 on the same lane as the lane on which the virtual vehicle 700 is placed may be determined as the travel speed of the virtually placed vehicle 700. In this way, the travel speed of the virtual vehicle 700 can be easily determined.

[0145] When two or more vehicles 500 are detected in the lane in which the virtual vehicle 700 is located, the traveling speed of the faster of the two or more vehicles 500 may be set as the traveling speed of the virtual vehicle 700. Alternatively, when two or more vehicles 500 are detected in the lane in which the virtual vehicle 700 is located, the traveling speed of the vehicle 500 that is closer to the virtual vehicle 700 may be set as the traveling speed of the virtual vehicle 700.

[0146] The determined traveling speed of the vehicle 700 can also be corrected. The determined traveling speed of the vehicle 700 can be corrected so that the greater the distance between the position where the virtual vehicle 700 is virtually placed and the position of the vehicle VH, the greater the speed. Another vehicle that is currently outside the detectable area of ​​the sensor group 200 and is far away from the vehicle VH may be traveling at high speed and enter the detectable area RD in a short time. By correcting the determined traveling speed of the vehicle 700 as described above, a safe trajectory can be planned. This correction method is applicable both when the maximum speed established by laws and regulations regarding vehicle speed limits is determined as the traveling speed of the vehicle 700 and when the traveling speed of another vehicle 500 detected around the vehicle VH is determined as the traveling speed of the vehicle 700.

[0147] I. Other Embodiments: (I1) In the first embodiment, an example was described in which the predicted number of multipaths (the value of MP) was fixed. The search processing unit 153 may change the predicted number of multipaths depending on the situation around the vehicle VH. For example, the situation around the vehicle VH is determined depending on conditions such as the number of other vehicles around the vehicle VH, the time of day the vehicle VH is traveling, and weather. For example, if the number of other vehicles traveling in the driving lane and the passing lane is equal to or greater than a first threshold, the predicted number of multipaths may be increased from a reference value. Also, if the number of other vehicles traveling in the driving lane and the passing lane is less than a second threshold, the predicted number of multipaths may be decreased from a reference value. Note that the second threshold is set to a value smaller than the first threshold. Also, for example, if the time period during which the vehicle VH is traveling is a time period during which there are many other vehicles traveling in the driving lane and the passing lane, the predicted number of multipaths may be increased from a reference value. If the time period during which the vehicle VH is traveling is a time period during which there are few other vehicles traveling in the driving lane and the passing lane, the predicted number of multipaths may be decreased from a reference value. The search processing unit 153 may specify the predicted number of multipaths determined according to the surrounding conditions of the vehicle VH to the quantum-inspired machine 160. The search processing unit 153 is also referred to as a "specifying unit."

[0148] (I2) In the first embodiment, an example was described in which the search process for selecting MP trajectory candidates from a plurality of trajectory candidates is repeated until the vehicle VH merges into the driving lane (see step S150 in FIG. 2). However, the search process does not necessarily have to be repeated. For example, if the number of other vehicles around the vehicle VH is less than a threshold, the search process may be performed only once. If the number of other vehicles around the vehicle VH is equal to or greater than a threshold, the search process may be performed repeatedly.

[0149] 2, if it is determined not to end the process, the process from step S110 onward may be executed again after waiting for a certain period of time, for example, three seconds.

[0150] (I3) In the first embodiment, an example was described in which the vehicle VH was an electric vehicle, but the vehicle VH may also be a car powered by an internal combustion engine, or a hybrid vehicle equipped with an internal combustion engine and a motor.

[0151] (I4) In the fifth embodiment, the number of virtually placed vehicles 700 may be limited depending on the processing capacity of the trajectory planning system 10. For example, assume that the trajectory planning system 10 assumes trajectory candidates for a maximum of L vehicles (L is an integer greater than or equal to 1) and can determine a planned trajectory for the vehicle VH using the assumed trajectory candidates. Assume that N vehicles 500 (N is an integer greater than or equal to 1) are actually detected around the vehicle VH. In this case, a maximum of (LN) virtual vehicles 700 are virtually placed.

[0152] If the trajectory planning system 10 is made to execute processing that exceeds its processing capacity, the trajectory planning system 10 itself may fail. Therefore, it is preferable to limit the number of virtually placed vehicles 700 in accordance with the processing capacity of the trajectory planning system 10. Furthermore, it is preferable that the processing capacity assumed for the trajectory planning system 10 is not the maximum processing capacity of the trajectory planning system 10, but a processing capacity that has a certain margin.

[0153] (I5) In the fifth embodiment, if a vehicle corresponding to a vehicle 700 virtually placed in a specific region RP is not detected by the sensor group 200 for a certain period of time, the virtual placement unit 155 may virtually remove the vehicle 700 from the specific region RP. The certain period of time refers to a period during which a processing loop including the processing of steps S110 to S140 shown in FIG. 11 is repeated a predetermined number of times. In step S112 of FIG. 11, when the virtual placement unit 155 virtually places the virtual vehicle 700 in the specific region RP, if the vehicle 700 is also a vehicle that was virtually placed in the immediately preceding step S112, the virtual placement unit 155 can store information indicating this. Using this information, the virtual placement unit 155 can determine that a vehicle corresponding to the vehicle 700 has not been detected by the sensor group 200 for a certain period of time.

[0154] If a vehicle corresponding to the virtually placed vehicle 700 is not detected by the sensor group 200 for a certain period of time, this means that a vehicle corresponding to the assumed vehicle 700 does not exist. It is inefficient to treat such a non-existent vehicle as a processing target. By excluding virtual vehicles that have not been detected for a certain period of time from the processing targets, it is possible to improve the efficiency of the processing.

[0155] (I6) In the fifth embodiment, an example was described in which one virtual vehicle 700 was placed in the blind spot area RB. The virtual placement unit 155 may place two or more virtual vehicles in the blind spot area RB depending on the size or type of the obstacle. For example, if the obstacle is a truck or a large bus, two or more vehicles, motorcycles, etc., may be present in the blind spot area RB. The larger the size of the obstacle, the more virtual moving objects may be placed in the blind spot area RB created by the obstacle. By assuming multiple moving objects present in the blind spot area RB, a safe trajectory plan can be executed. Furthermore, as the maximum speed established by laws and regulations regarding vehicle speed limits increases, it is expected that the inter-vehicle distance between the leading vehicle and the following vehicle will increase. Whether to place one or more virtual vehicles 700 in the blind spot area RB may be determined depending on the expected inter-vehicle distance between the leading vehicle and the following vehicle.

[0156] (I7) In the fifth embodiment, an example was described in which a vehicle in the passing lane L2, which is the lane farther from the vehicle VH of the driving lane L1 and the passing lane L2, is not considered an obstacle. A vehicle in the passing lane L2, which is the lane farther from the vehicle VH of the driving lane L1 and the passing lane L2, may be considered an obstacle.

[0157] As shown in FIG. 14, when viewed from vehicle VH, there is a blind spot CB to the right of vehicle 500b traveling in passing lane L2, and there is a blind spot CC to the right of vehicle 500c traveling in passing lane L2. Because the ranges of both blind spots CB and CC are not large, it is difficult to imagine that a vehicle is in blind spot CB or CC. However, a small moving object such as a motorcycle may be in blind spot CB or CC. Therefore, the virtual placement unit 155 may place a virtual motorcycle as a virtual moving object in blind spot CB and blind spot CC.

[0158] The estimation apparatus and method described herein may be implemented by a special-purpose computer configured with a processor and memory programmed to perform one or more functions embodied in a computer program. Alternatively, the estimation apparatus and method described herein may be implemented by a special-purpose computer configured with a processor comprising one or more dedicated hardware logic circuits. Alternatively, the estimation apparatus and method described herein may be implemented by one or more special-purpose computers configured with a combination of a processor and memory programmed to perform one or more functions and a processor configured with one or more hardware logic circuits. Furthermore, the computer program may be stored in a computer-readable non-transitory tangible storage medium as instructions executed by a computer.

[0159] The present disclosure is not limited to the above-described embodiments and can be realized in various configurations without departing from the spirit thereof. For example, the technical features in the embodiments corresponding to the technical features in each aspect described in the Summary of the Invention section can be appropriately replaced or combined to solve some or all of the above-described problems or achieve some or all of the above-described effects. Furthermore, if a technical feature is not described as essential in this specification, it can be appropriately deleted. [Explanation of symbols]

[0160] 10, 10d, 10e...Trajectory planning system, 130...First communication unit, 140...Second communication unit, 152...Trajectory prediction unit, 153...Search processing unit, 154...Trajectory planning unit, 155...Virtual placement unit, 160...Quantum-inspired machine, 200...Sensor group, 500...Vehicle, 700...Virtual vehicle, 750...Virtual pedestrian, D1...Distance, RB...Blind spot area, RE...Boundary area, RB...Undetectable area, VH...Vehicle

Claims

1. A travel trajectory planning system (10, 10d, 10e), a sensing unit (200, 130, 140) that senses the surrounding conditions of the vehicle (VH); a trajectory prediction unit (152) that outputs a plurality of predicted trajectory candidates for each of the one or more other vehicles (500) other than the subject vehicle detected around the subject vehicle by the sensing unit; a search unit (160) that searches for a solution to a combinatorial optimization problem formulated to select an optimal combination of trajectory candidates obtained by selecting two or more trajectory candidates for each of the other vehicles from the plurality of predicted trajectory candidates; a trajectory planning unit (154) that determines a planned trajectory along which the host vehicle will travel using the solution found by the search unit; A driving trajectory planning system comprising:

2. The travel trajectory planning system according to claim 1, a designation unit (153) that designates, to the search unit, a multipath prediction number representing the number of trajectory candidates selected from the plurality of trajectory candidates for each of the other vehicles, the designation unit changing the multipath prediction number depending on the situation around the vehicle; Furthermore, the search unit searches for the combination of trajectory candidates obtained by selecting, for each of the other vehicles, the trajectory candidates of the predicted number of multipaths from the plurality of predicted trajectory candidates; Trajectory planning system.

3. 3. The traveling trajectory planning system according to claim 2, the trajectory prediction unit further outputs a plurality of vehicle trajectory candidates predicted for the vehicle; The search unit Two or more trajectory candidates are selected for each of the other vehicles from the plurality of predicted trajectory candidates; one vehicle trajectory candidate is selected from the plurality of predicted vehicle trajectory candidates; a search for a solution to the combinatorial optimization problem formulated to select an optimal combination of the trajectory candidates obtained by: Trajectory planning system.

4. The travel trajectory planning system according to claim 2 or 3, a virtual placement unit (155) that virtually places one or more virtual moving objects (700) in a specific region (RP) around the vehicle, the specific region being a region in which the sensing unit cannot detect a moving object; the trajectory prediction unit outputs a plurality of predicted virtual trajectory candidates for each virtual moving body included in the one or more virtual moving bodies; The search unit Two or more trajectory candidates are selected for each of the other vehicles from the plurality of predicted trajectory candidates; two or more virtual trajectory candidates are selected for each of the virtual moving bodies from the plurality of predicted virtual trajectory candidates; and searching for a solution to the combinatorial optimization problem formulated to select an optimal combination of the trajectory candidates obtained by: Trajectory planning system.

5. The travel trajectory planning system according to claim 4, The specific area includes a blind spot area (RB) caused by an obstacle and a boundary area (RE) which is a limited area of ​​an undetectable area (RN) in which a moving object cannot be detected by the sensing unit, the boundary area includes a range that is spaced a predetermined first distance from a boundary between a detectable area (RD), which is an area in which the sensing unit can detect the moving object, and the undetectable area in a direction away from the detectable area, Trajectory planning system.

6. The travel trajectory planning system according to claim 5, The virtual placement unit determining a direction of travel of each of the virtual moving bodies based on the position where each of the virtual moving bodies is placed and map information; determining a traveling speed of each of the virtual moving objects based on a traveling speed detected for at least one of the one or more other vehicles detected around the host vehicle by the sensing unit or a law or regulation regarding a speed limit for vehicles; Trajectory planning system.

7. The travel trajectory planning system according to claim 6, the virtual placement unit does not place the virtual moving object in an area within the specific area where it is clear that the moving object does not exist; Trajectory planning system.

8. The traveling trajectory planning system according to claim 7, The virtual placement unit correcting the travel speed determined for the virtual moving object so that the speed increases as the distance between the position where the virtual moving object is virtually placed and the position of the host vehicle increases; Trajectory planning system.

9. The travel trajectory planning system according to claim 8, the virtual placement unit virtually places one or more virtual pedestrians (750) in the specific area as the one or more virtual moving bodies; the trajectory prediction unit outputs a plurality of predicted virtual trajectory candidates for each virtual pedestrian included in the one or more virtual pedestrians using a mathematical model; Trajectory planning system.

10. The traveling trajectory planning system according to claim 9, The larger the size of the obstacle, the more the number of the virtual moving objects to be placed in the blind spot area caused by the obstacle is increased. Trajectory planning system.

11. The traveling trajectory planning system according to claim 10, the number of the virtual moving objects virtually placed in the specific area is limited depending on the processing capacity of the system; Trajectory planning system.

12. The travel trajectory planning system according to claim 11, The sensing unit A function of sensing the surrounding situation of the vehicle using at least one of a camera, a LiDAR device, and a radar installed in the vehicle; a function of communicating with the one or more other vehicles via V2N communication; a function of communicating with the one or more other vehicles via V2X communication; A function of acquiring sensing results from sensors installed on the road from traffic infrastructure via the V2X communication; a function of acquiring the sensing results by sensors installed in the one or more other vehicles through V2I communication via the transportation infrastructure; At least one of the following functions is provided: Trajectory planning system.

13. The travel trajectory planning system according to claim 12, the trajectory prediction unit outputs the plurality of trajectory candidates by using values ​​that are set with variations in the values ​​indicating the positions of the one or more other vehicles detected by the sensing unit as the positions of the one or more other vehicles. Trajectory planning system.

14. The travel trajectory planning system according to claim 13, a process in which the trajectory prediction unit outputs the plurality of trajectory candidates for each of the other vehicles included in the one or more other vehicles; a process in which the search unit searches for the combinations related to the trajectory candidates; is repeatedly executed while the host vehicle is merging from the merging lane onto the main lane. Trajectory planning system.

15. The travel trajectory planning system according to claim 14, a process in which the virtual placement unit virtually places the one or more virtual moving objects; a process in which the trajectory prediction unit outputs the plurality of trajectory candidates for each of the other vehicles and outputs the plurality of virtual trajectory candidates for each of the virtual moving bodies; a process in which the search unit searches for the combinations related to the trajectory candidates; is executed repeatedly, the virtual placement unit virtually removes the virtual moving object from the specific area when the sensing unit does not detect a moving object corresponding to the virtual moving object for a certain period of time or longer. Trajectory planning system.

16. The travel trajectory planning system according to claim 15, causing a quantum-inspired machine installed in the vehicle to execute the processing executed by the search unit; Trajectory planning system.

17. The travel trajectory planning system according to claim 16, causing the quantum-inspired machine to execute the processing executed by the trajectory planning unit; Trajectory planning system.

18. The travel trajectory planning system according to claim 17, a cost term included in an objective function formulated in a QUBO (Quadratic Unconstrained Binary Optimization) format to select the optimal combination, wherein the cost term related to the one or more virtual moving objects is multiplied by a weighting coefficient between 0 and 1; Trajectory planning system.

19. The travel trajectory planning system according to claim 18, The weighting coefficient is changed according to the traveling speed of the host vehicle, the distance between the host vehicle and each of the other vehicles, and the traveling speed of each of the other vehicles. Trajectory planning system.

20. The travel trajectory planning system according to claim 18, The weighting coefficients are fixed values. Trajectory planning system.

21. The travel trajectory planning system according to claim 18, The weighting coefficient is changed depending on an environmental state around the vehicle that may affect the sensing result by the sensing unit. Trajectory planning system.

Images