Obstacle monitoring system, obstacle monitoring device, obstacle monitoring method, and program

Abstract

To provide a technique capable of improving the detection accuracy of an obstacle while ensuring the moving speed of a mobile body when detecting the obstacle in the advancing direction of the mobile body on the basis of the sensing result of sensing means mounted on the mobile body.SOLUTION: This obstacle monitoring system comprises: sensing means; first obstacle detection means; and control means for selectively executing a plurality of control modes including a first control mode in which the moving speed of the mobile body is a first moving speed and the sensing view angle of the sensing means is a first view angle and a second control mode in which the moving speed of the mobile body is a second moving speed and the sensing view angle of the sensing means is a second view angle. The control means executes the second control mode so as to capture the obstacle within the second view angle when the first obstacle detection means detects the obstacle while the control means is executing the first control mode.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to an obstacle monitoring system, an obstacle monitoring device, an obstacle monitoring method, and a program. [Background technology]

[0002] Patent Document 1 discloses a vehicle control system that, when an obstacle is detected ahead of the vehicle, searches for an evacuation space for the vehicle and guides the vehicle to the searched evacuation space and stops the vehicle there. As means for detecting an obstacle ahead of the vehicle, a stereo camera, a Radar (Radio Detection and Ranging) device, and a LiDAR (Light Detection and Ranging) device are exemplified. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] International Publication No. 2018 / 179958 Summary of the Invention [Problem to be solved by the invention]

[0004] However, when detecting an obstacle in the traveling direction of a moving body based on the sensing results of a sensing means mounted on the moving body, there is a trade-off between the moving speed of the moving body and the accuracy of obstacle detection, and it is considered difficult to achieve both simultaneously.

[0005] That is, when a moving object moves at high speed, the time from when the moving object detects an obstacle until when the moving object actually collides with the obstacle is limited. Therefore, the time is insufficient to detect the obstacle with high accuracy. Therefore, the obstacle may be falsely detected. If the moving object's speed is reduced due to the false detection of an obstacle, the time when the moving object arrives at the destination will be delayed.

[0006] An object of the present disclosure is to provide a technology that improves the accuracy of obstacle detection while maintaining the movement speed of a moving body when detecting an obstacle in the moving direction of the moving body based on the sensing results of a sensing means mounted on the moving body. [Means for solving the problem]

[0007] A sensing means for sensing a forward direction in a moving direction of the moving body; a first obstacle detection means for detecting an obstacle based on a sensing result of the sensing means; a control means for selectively executing a plurality of control modes including a first control mode in which the moving speed of the moving body is a first moving speed and the sensing field angle of the sensing means is a first field angle, and a second control mode in which the moving speed of the moving body is a second moving speed lower than the first moving speed and the sensing field angle of the sensing means is a second field angle narrower than the first field angle; Including, when the first obstacle detection means detects the obstacle while the control means is executing the first control mode, the control means executes the second control mode so as to capture the obstacle within the second viewing angle. An obstacle monitoring system is provided.

[0008] A sensing means for sensing a forward direction in a moving direction of the moving body; a first obstacle detection means for detecting an obstacle based on a sensing result of the sensing means; a control means for selectively executing a plurality of control modes including a first control mode in which the moving speed of the moving body is a first moving speed and the sensing field angle of the sensing means is a first field angle, and a second control mode in which the moving speed of the moving body is a second moving speed lower than the first moving speed and the sensing field angle of the sensing means is a second field angle narrower than the first field angle; Including, when the first obstacle detection means detects the obstacle while the control means is executing the first control mode, the control means executes the second control mode so as to capture the obstacle within the second viewing angle. An obstacle monitoring device is provided.

[0009] The computer Detecting an obstacle based on a sensing result of a sensing means that senses a front area in a moving direction of the moving body; selectively executing a plurality of control modes including a first control mode in which the moving speed of the moving body is a first moving speed and the sensing field angle of the sensing means is a first field angle, and a second control mode in which the moving speed of the moving body is a second moving speed lower than the first moving speed and the sensing field angle of the sensing means is a second field angle narrower than the first field angle; When the obstacle is detected while the first control mode is being executed, the second control mode is executed so as to capture the obstacle within the second viewing angle. A method for monitoring obstacles is provided. Effect of the Invention

[0010] According to the present disclosure, when detecting an obstacle in the traveling direction of a moving body based on the sensing results of a sensing means mounted on the moving body, it is possible to improve the accuracy of obstacle detection while maintaining the moving speed of the moving body. [Brief description of the drawings]

[0011] [Figure 1] FIG. 1 is a block diagram of an obstacle monitoring system. [Diagram 2] FIG. 2 is a block diagram of an obstacle monitoring device. [Diagram 3] FIG. [Figure 4] FIG. 11 is an explanatory diagram of a threshold table. [Diagram 5] 4 is a control flow of the obstacle monitoring device. [Figure 6]1 is a diagram showing a case where a processing circuit provided in an obstacle monitoring device is configured with a processor and a memory. [Figure 7] 1 is a diagram showing a case where a processing circuit provided in an obstacle monitoring device is configured with dedicated hardware. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] (Summary of the Disclosure) An outline of the present disclosure will be described below. Fig. 1 shows a block diagram of an obstacle monitoring system 100. As shown in Fig. 1, the obstacle monitoring system 100 includes a sensing means 101, a first obstacle detection means 102, and a control means 103.

[0013] The sensing means 101 senses the area ahead in the moving direction of the moving body.

[0014] The first obstacle detection means 102 detects an obstacle based on the sensing result of the sensing means 101 .

[0015] The control means 103 selectively executes a plurality of control modes including a first control mode and a second control mode.

[0016] The first control mode is a control mode in which the moving speed of the moving object is a first moving speed, and the sensing field angle of the sensing means is a first field angle.

[0017] The second control mode is a control mode in which the moving speed of the moving body is a second moving speed that is lower than the first moving speed, and the sensing field angle of the sensing means is a second field angle that is narrower than the first field angle.

[0018] If the first obstacle detection means 102 detects an obstacle while the control means 103 is executing the first control mode, the control means 103 executes the second control mode so as to capture the obstacle within the second viewing angle.

[0019] According to the above configuration, when detecting an obstacle in the traveling direction of a moving body based on the sensing results of a sensor mounted on the moving body, it is possible to improve the accuracy of obstacle detection while maintaining the moving speed of the moving body.

[0020] (Embodiment) An embodiment of the present disclosure will be described below. Fig. 2 is a block diagram of an obstacle monitoring device 2 mounted on a train 1.

[0021] As shown in FIG. 2, the obstacle monitoring device 2 includes a LiDAR device 3, a rainfall sensor 4, an acceleration sensor 5, a first obstacle detection unit 6, a second obstacle detection unit 7, and a control unit 8. In addition to the obstacle monitoring device 2, the train 1 further includes a motor 10 that drives wheels 9, and a braking device 11 that brakes the wheels 9. The train 1 runs along a rail 12. The rail 12 is a specific example of a moving track along which the train 1 runs. The train 1 is a specific example of a moving body. The moving body is not limited to the train 1, and may be a car or a drone.

[0022] The LiDAR device 3 is a specific example of a sensing means for sensing the front in the running direction of the train 1. The LiDAR device 3 of this embodiment is a direct ToF (Time of Flight) type. That is, the LiDAR device 3 emits laser light forward in the running direction of the train 1 and measures the time required to receive the reflected light, thereby generating a three-dimensional point cloud forward in the running direction of the train 1. However, instead of this, the LiDAR device 3 may be an FMCW (Frequency Modulated Continuous Wave) type that generates the above three-dimensional point cloud based on the frequency difference between the laser light emitted forward in the running direction of the train 1 and the reflected light. In addition, the LiDAR device 3 may be an indirect ToF type that generates the above three-dimensional point cloud based on the phase difference between the laser light emitted forward in the running direction of the train 1 and the reflected light. The LiDAR device 3 corrects the generated three-dimensional point cloud based on the current location of the train 1, which changes from moment to moment. The LiDAR device 3 stores the corrected three-dimensional point cloud expressed in the global coordinate system in the latest point cloud storage unit 3a.

[0023] The sensing means is not limited to the LiDAR device 3. Any device capable of sensing the area ahead in the running direction of the train 1 may be used as the sensing means. For example, the sensing means may be configured to generate a three-dimensional point cloud using a radar device (Radio Detection And Ranging), an ultrasonic sensor, a stereo camera, or a combination of these. The sensing means may also be configured to generate a three-dimensional point cloud using Structure from Motion (SfM) from a plurality of two-dimensional images obtained by capturing an image of the area ahead in the running direction of the train 1. In any sensing means, the reliability of the sensing results is improved if a longer measurement time can be secured.

[0024] The rainfall sensor 4 is a weather sensor that detects whether the current weather is rainy or sunny. The rainfall sensor 4 outputs the sensing result to the control unit 8.

[0025] The acceleration sensor 5 detects the acceleration acting on the train 1. The acceleration sensor 5 outputs the sensing result to the control unit 8.

[0026] The first obstacle detection unit 6 detects an obstacle ahead in the running direction of the train 1 based on the three-dimensional point cloud stored in the latest point cloud storage unit 3a. Typically, the first obstacle detection unit 6 compares the three-dimensional point cloud stored in the latest point cloud storage unit 3a with the three-dimensional point cloud stored in the past point cloud storage unit 3b, and detects an obstacle ahead in the running direction of the train 1 based on the difference. The three-dimensional point cloud stored in the past point cloud storage unit 3b is a three-dimensional point cloud generated and accumulated before the train 1 runs. The obstacles are, for example, people, animals, stones, and fallen trees.

[0027] The second obstacle detection unit 7 is configured with a neural network that inputs the three-dimensional point cloud stored in the latest point cloud storage unit 3a and outputs a detection score for each class of obstacle detected by the first obstacle detection unit 6. The detection score for each class of obstacle detected by the first obstacle detection unit 6 is output to the output layer of the neural network. The detection score indicates the probability that a corresponding obstacle belongs to the corresponding class. In this embodiment, the probability is expressed in the range from 0 to 1. The detection score for each class is normalized by a softmax function so that the sum of the detection scores of all classes becomes 1.

[0028] The control unit 8 selectively executes a plurality of measurement modes including a low reliability measurement mode 8a and a high reliability measurement mode 8b. The measurement modes are a specific example of a control mode. The low reliability measurement mode 8a is a specific example of a first control mode. The high reliability measurement mode 8b is a specific example of a second control mode. In this embodiment, the plurality of measurement modes include the low reliability measurement mode 8a and the high reliability measurement mode 8b. The plurality of measurement modes may include other measurement modes, such as a medium reliability measurement mode.

[0029] The low reliability measurement mode 8a is a measurement mode in which the running speed of the train 1 is a first running speed, and the sensing field of view (FoV: Field of View) of the LiDAR device 3 is a first field of view. The high reliability measurement mode 8b is a measurement mode in which the running speed of the train 1 is a second running speed lower than the first running speed, and the sensing field of view of the LiDAR device 3 is a second field of view narrower than the first field of view. FIG. 3 shows a first field of view FV1 when the control unit 8 executes the low reliability measurement mode 8a, and a second field of view FV2 when the control unit 8 executes the high reliability measurement mode 8b. As shown in FIG. 3, the second field of view FV2 is a field of view that is more inward than the first field of view FV1. The first field of view FV1 includes the second field of view FV2.

[0030] Then, when the first obstacle detection unit 6 detects an obstacle while the control unit 8 is executing the low reliability measurement mode 8a, the control unit 8 executes the high reliability measurement mode 8b so as to capture the obstacle within the second viewing angle FV2. That is, when the control unit 8 executes the high reliability measurement mode 8b, the LiDAR device 3 intensively scans the obstacle detected by the first obstacle detection unit 6. This allows the LiDAR device 3 to increase the point cloud density of the three-dimensional point cloud corresponding to the obstacle detected by the first obstacle detection unit 6. In addition, when the control unit 8 executes the high reliability measurement mode 8b, the running speed of the train 1 is the second running speed, so that the time that the LiDAR device 3 can spend scanning the obstacle detected by the first obstacle detection unit 6 can be secured for a long time. In this sense, too, the LiDAR device 3 can increase the point cloud density of the three-dimensional point cloud corresponding to the obstacle detected by the first obstacle detection unit 6.

[0031] 2, the control unit 8 has a threshold value table 8c. The control unit 8 refers to the threshold value table 8c when controlling the running of the train 1 according to the type of obstacle, weather conditions, and the shape of the rail 12.

[0032] FIG. 4 illustrates an example of values ​​in the threshold table 8c. As shown in FIG. 4, the threshold table 8c specifies a score threshold for each class. The multiple classes (types, categories) are roughly divided into a harmful class group 20 that impedes the running of the train 1 and a harmless class group 21 that does not impede the running of the train 1. The harmful class group 20 includes classes A1 to A5. The harmless class group 21 includes classes B1 to B5. Classes A1 to A5 are, for example, people, stones, and fallen trees. Classes B1 to B5 are, for example, animals. The score threshold of each class belonging to the harmless class group 21 is set higher than the score threshold of each class belonging to the harmful class group 20.

[0033] The score threshold for each class is set in more detail depending on the weather conditions and the shape of the rail 12. That is, for example, when the score threshold for class A1 is taken into consideration, the score threshold for class A1 is set as follows. When the weather is good and rail 12 is curved, the score threshold is set to 0.60. When the weather is good and the rail 12 is straight, the score threshold is set to 0.70. When the weather is bad and track 12 is curved, the score threshold is set to 0.40. When the weather is bad and rail 12 is straight, the score threshold is set to 0.50.

[0034] Then, when the detection score exceeds the score threshold in any class, the control unit 8 slows down or stops the train 1.

[0035] As described above, the score threshold in bad weather is set lower than the score threshold in good weather. In the example of FIG. 4, the score threshold in bad weather is set 0.2 points lower than the score threshold in good weather. That is, in bad weather, there is a problem that the number of points of the three-dimensional point cloud that the LiDAR device 3 can generate per unit time is reduced. In other words, in bad weather, the reliability of sensing by the LiDAR device 3 decreases. Therefore, since it is considered that the reliability of the detection score output by the second obstacle detection unit 7 is reduced in bad weather, compared to good weather, prevention of collision between the train 1 and an obstacle is prioritized, that is, it is made easier to decelerate or stop the train 1.

[0036] In addition, the score threshold when the rail 12 on which the train 1 is currently traveling is curved is set lower than the score threshold when the rail 12 on which the train 1 is currently traveling is straight. In the example of FIG. 4, the score threshold when the rail 12 on which the train 1 is currently traveling is curved is set 0.1 lower than the score threshold when the rail 12 on which the train 1 is currently traveling is straight. That is, when the rail 12 on which the train 1 is currently traveling is curved, it is difficult to track the rail 12, and the reliability of the detection score output by the second obstacle detection unit 7 is considered to be reduced. Therefore, when the rail 12 on which the train 1 is currently traveling is curved, the prevention of a collision between the train 1 and an obstacle is prioritized, that is, the train 1 is made easier to decelerate or stop, compared to when the rail 12 on which the train 1 is currently traveling is straight.

[0037] Furthermore, when the detection scores of classes A1 to A5 belonging to the harmful class group 20 increase even slightly, that is, when the possibility that the obstacle detected by the first obstacle detection unit 6 will be classified into a class that will interfere with the running of the train 1 increases even slightly, the control unit 8 quickly decelerates or stops the train 1. On the other hand, when the detection scores of classes B1 to B5 belonging to the harmless class group 21 increase, that is, when the possibility that the obstacle detected by the first obstacle detection unit 6 will be classified into a class that will not interfere with the running of the train 1 increases, the control unit 8 prioritizes the running of the train 1 and does not decelerate or stop the train 1. This makes it possible to prevent collisions between the train 1 and the obstacles while maintaining the regular operation of the train 1 as much as possible.

[0038] In this embodiment, the control unit 8 determines the weather at the location where the train 1 is currently traveling based on the sensing result of the rainfall sensor 4. However, instead of this, the control unit 8 may determine the weather at the location where the train 1 is currently traveling by accessing an external server that provides meteorological information.

[0039] Furthermore, in this embodiment, the control unit 8 determines whether the rail 12 on which the train 1 is currently traveling is curved or straight, based on the sensing result of the acceleration sensor 5. However, instead of this, the control unit 8 may determine whether the rail 12 on which the train 1 is currently traveling is curved or straight, based on current location information indicating the current location of the train 1 and map information including the track of the rail 12.

[0040] In addition, in this embodiment, the control unit 8 controls the running speed of the train 1 by outputting a drive command to a motor driver (not shown) that drives the motor 10, and by outputting a drive command to a brake driver (not shown) that drives the brake device 11.

[0041] Next, we will explain the operation flow of the obstacle monitoring device 2. Fig. 5 shows the control flow of the obstacle monitoring device 2.

[0042] S100: First, the control unit 8 executes the low reliability measurement mode 8a.

[0043] S110: Next, the control unit 8 determines whether the first obstacle detection unit 6 has detected an obstacle. If the first obstacle detection unit 6 has detected an obstacle (S110: YES), the control unit 8 advances the process to S120. On the other hand, if the first obstacle detection unit 6 has not detected an obstacle (S110: NO), the control unit 8 returns the process to S100.

[0044] S120: Next, the control unit 8 determines whether the obstacle detected by the first obstacle detection unit 6 is on the rail 12. If the obstacle detected by the first obstacle detection unit 6 is off the rail 12 (S120: NO), the control unit 8 returns the process to S100. On the other hand, if the obstacle detected by the first obstacle detection unit 6 is on the rail 12 (S120: YES), the control unit 8 advances the process to S130.

[0045] S130: Next, the control unit 8 executes the high reliability measurement mode 8b instead of the low reliability measurement mode 8a. As a result, the train 1 decelerates and the LiDAR device 3 intensively scans the obstacles detected by the first obstacle detection unit 6.

[0046] S140: Next, the second obstacle detection unit 7 performs classifying of the obstacles detected by the first obstacle detection unit 6 based on the three-dimensional point cloud stored in the latest point cloud storage unit 3a.

[0047] S150: Next, the control unit 8 judges the current weather and the shape of the rail 12 based on the sensing results of the rain sensor 4 and the acceleration sensor 5. Then, the control unit 8 refers to the threshold table 8c to judge whether the detection score exceeds the score threshold in any class. If it is judged that the detection score exceeds the score threshold in any class (S150: YES), the control unit 8 advances the process to S160. On the other hand, if it is judged that the detection score does not exceed the score threshold in any class (S150: NO), the control unit 8 returns the process to S100.

[0048] For example, if the detection score, which is the probability that the obstacle detected by the first obstacle detection unit 6 is classified into class A2, is 0.6, the current weather is bad, and the track 12 on which the train 1 is currently traveling is straight, the detection score exceeds the score threshold of 0.54. In this case, the control unit 8 advances the process from S150 to S160.

[0049] In addition, if the detection score, which is the probability that the obstacle detected by the first obstacle detection unit 6 is classified into class B4, is 0.7, and the current weather is fine and the track 12 on which the train 1 is currently traveling is curved, the detection score falls below the score threshold of 0.72. In this case, the control unit 8 returns the process from S150 to S100.

[0050] S160: Next, the control unit 8 ends the process by decelerating or stopping the train 1. This makes it possible to mitigate or avoid a collision between the train 1 and an obstacle.

[0051] The preferred embodiment of the present disclosure has been described above. The above embodiment has the following features.

[0052] The obstacle monitoring device 2 is a specific example of an obstacle monitoring system. The obstacle monitoring device 2 includes a LiDAR device 3 (sensing means), a first obstacle detection unit 6 (first obstacle detection means), and a control unit 8 (control means). The LiDAR device 3 senses the front in the running direction (movement direction) of the train 1 (moving body). The first obstacle detection unit 6 detects an obstacle based on the sensing result of the LiDAR device 3. The control unit 8 selectively executes a plurality of measurement modes (control modes) including a low reliability measurement mode 8a (first control mode) and a high reliability measurement mode 8b (second control mode). Then, when the first obstacle detection unit 6 detects an obstacle while the control unit 8 is executing the low reliability measurement mode 8a, the control unit 8 executes the high reliability measurement mode 8b so as to capture the obstacle within the second viewing angle FV2. According to the above configuration, when detecting obstacles in the traveling direction of the train 1 based on the sensing results of the LiDAR device 3 mounted on the train 1, the detection accuracy of the train 1 can be improved while maintaining the movement speed of the train 1.

[0053] The obstacle monitoring device 2 further includes a second obstacle detection unit 7 (second obstacle detection means). When the control unit 8 is executing the high reliability measurement mode 8b, the second obstacle detection unit 7 is configured with a neural network that inputs the sensing result of the LiDAR device 3 and outputs a detection score for each class of obstacle detected by the first obstacle detection unit 6. With the above configuration, high reliability of obstacle classification can be achieved. Note that PointNet, PointNet++, and VoteNet may be adopted as neural networks that perform class classification based on a three-dimensional point cloud. Also, the reliability of object detection may be improved by sensor fusion of a two-dimensional image and a three-dimensional point cloud.

[0054] 4, a score threshold is set for each class in the threshold table 8c. When the detection score exceeds the score threshold in any class, the control unit 8 slows down or stops the train 1. With the above configuration, it is possible to finely adjust the running control of the train 1 for each class, by giving priority to either the operation of the train 1 or avoiding a collision between the train 1 and an obstacle.

[0055] Furthermore, a first score threshold is set as the score threshold for a class in which the running of the train 1 is hindered. A second score threshold higher than the first score threshold is set as the score threshold for a class in which the running of the train 1 is not hindered. With the above configuration, depending on whether the running of the train 1 is hindered or not, the running control of the train 1 can be finely adjusted to give priority to either the operation of the train 1 or avoiding a collision between the train 1 and an obstacle.

[0056] In addition, for each class, the score threshold for bad weather is set lower than the score threshold for good weather. With the above configuration, it becomes easier to slow down or stop the train 1 in bad weather, so preventing a collision between the train 1 and an obstacle is given priority.

[0057] Furthermore, in each class, the score threshold when the rail 12 (traveling track) on which the train 1 is currently traveling is curved is set lower than the score threshold when the rail 12 is straight. According to the above configuration, since it becomes easier to decelerate or stop the train 1 when the rail 12 is curved, prevention of collision between the train 1 and an obstacle is prioritized.

[0058] Furthermore, when the obstacle detected by the first obstacle detection unit 6 has deviated from the rail 12 of the train 1 (S120: NO), the control unit 8 does not switch the control mode. According to the above configuration, the train 1 is not unnecessarily decelerated, so that the on-time operation of the train 1 can be prioritized.

[0059] Next, a description will be given of the hardware configuration of the obstacle monitoring device 2. In the obstacle monitoring device 2, the first obstacle detection unit 6, the second obstacle detection unit 7, and the control unit 8 are realized by processing circuits. The processing circuit may be a processor and memory that executes a program stored in a memory, or may be dedicated hardware.

[0060] FIG. 6 is a diagram showing an example of a processing circuit provided in the obstacle monitoring device 2, which is configured with a processor and a memory. When the processing circuit is configured with a processor 1000 and a memory 1001, each function of the processing circuit of the obstacle monitoring device 2 is realized by software, firmware, or a combination of software and firmware. The software or firmware is written as a program and stored in the memory 1001. In the processing circuit, each function is realized by the processor 1000 reading and executing the program stored in the memory 1001. That is, the processing circuit includes the memory 1001 for storing a program that results in the processing of the obstacle monitoring device 2 being executed. It can also be said that these programs cause a computer to execute the procedures and methods of the obstacle monitoring device 2.

[0061] Here, the processor 1000 may be a CPU (Central Processing Unit), a processing device, an arithmetic device, a microprocessor, a microcomputer, or a DSP (Digital Signal Processor), etc. Also, the memory 1001 may be, for example, a RAM (Random Access Memory), a ROM (Read Only Memory), a flash memory, an EPROM (Erasable Programmable ROM), an EEPROM (Electrically EPROM), or other non-volatile or volatile semiconductor memory, a magnetic disk, a flexible disk, an optical disk, a compact disk, a mini disk, or a DVD (Digital Versatile Disc), etc.

[0062] Fig. 7 is a diagram showing an example in which the processing circuit of the obstacle monitoring device 2 is configured with dedicated hardware. When the processing circuit is configured with dedicated hardware, the processing circuit 1002 shown in Fig. 7 corresponds to, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a combination of these. Each function of the obstacle monitoring device 2 may be realized by the processing circuit 1002 on a function-by-function basis, or each function may be realized collectively by the processing circuit 1002.

[0063] It should be noted that the functions of the obstacle monitoring device 2 may be partially realized by dedicated hardware and partially realized by software or firmware. In this way, the processing circuit can realize each of the above-mentioned functions by dedicated hardware, software, firmware, or a combination of these.

[0064] Although the present disclosure has been described above with reference to the embodiments, the present disclosure is not limited to the above-described embodiments. Various modifications that can be understood by a person skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure.

[0065] Each drawing is merely an example for describing one or more embodiments. Each drawing may relate not only to one particular embodiment, but to one or more other embodiments. As can be understood by those skilled in the art, various features or steps described with reference to any one drawing may be combined with features or steps shown in one or more other drawings to create an embodiment not explicitly shown or described, for example. Not all features or steps shown in any one drawing are necessary to describe an exemplary embodiment, and some features or steps may be omitted. The order of steps described in any drawing may be changed as appropriate.

[0066] A part or all of the above-described embodiments can be described as, but is not limited to, the following supplementary notes. (Appendix 1) A sensing means for sensing a forward direction in a moving direction of the moving body; a first obstacle detection means for detecting an obstacle based on a sensing result of the sensing means; a control means for selectively executing a plurality of control modes including a first control mode in which the moving speed of the moving body is a first moving speed and the sensing field angle of the sensing means is a first field angle, and a second control mode in which the moving speed of the moving body is a second moving speed lower than the first moving speed and the sensing field angle of the sensing means is a second field angle narrower than the first field angle; Including, when the first obstacle detection means detects the obstacle while the control means is executing the first control mode, the control means executes the second control mode so as to capture the obstacle within the second viewing angle. Obstacle monitoring system. (Appendix 2) and a second obstacle detection means configured with a neural network that receives a sensing result from the sensing means and outputs a detection score for each class of the obstacle detected by the first obstacle detection means when the control means is executing the second control mode. 2. An obstacle monitoring system as described in claim 1. (Appendix 3) A score threshold is set for each class. The control means decelerates or stops the moving object when the detection score exceeds the score threshold in any class. 3. An obstacle monitoring system as described in claim 2. (Appendix 4) A first score threshold is set as the score threshold for a class that impedes the movement of the moving object; A second score threshold higher than the first score threshold is set as the score threshold for a class in which the movement of the moving object is not hindered. 4. An obstacle monitoring system as described in claim 3. (Appendix 5) For each class, the score threshold for bad weather is set lower than the score threshold for good weather. 5. An obstacle monitoring system according to claim 3 or 4. (Appendix 6) In each class, a score threshold when the moving trajectory currently being moved by the moving object is a curved line is set lower than a score threshold when the moving trajectory is a straight line. 5. An obstacle monitoring system according to claim 3 or 4. (Appendix 7) the control means does not switch the control mode when the obstacle detected by the first obstacle detection means is off the moving trajectory of the moving object. 2. An obstacle monitoring system as described in claim 1. (Appendix 8) A sensing means for sensing a forward direction in a moving direction of the moving body; a first obstacle detection means for detecting an obstacle based on a sensing result of the sensing means; a control means for selectively executing a plurality of control modes including a first control mode in which the moving speed of the moving body is a first moving speed and the sensing field angle of the sensing means is a first field angle, and a second control mode in which the moving speed of the moving body is a second moving speed lower than the first moving speed and the sensing field angle of the sensing means is a second field angle narrower than the first field angle; Including, when the first obstacle detection means detects the obstacle while the control means is executing the first control mode, the control means executes the second control mode so as to capture the obstacle within the second viewing angle. Obstacle monitoring device. (Appendix 9) The computer Detecting an obstacle based on a sensing result of a sensing means that senses a front area in a moving direction of the moving body; selectively executing a plurality of control modes including a first control mode in which the moving speed of the moving body is a first moving speed and the sensing field angle of the sensing means is a first field angle, and a second control mode in which the moving speed of the moving body is a second moving speed lower than the first moving speed and the sensing field angle of the sensing means is a second field angle narrower than the first field angle; When the obstacle is detected while the first control mode is being executed, the second control mode is executed so as to capture the obstacle within the second viewing angle. Obstacle monitoring method. (Appendix 10) A program that causes a computer to execute the obstacle monitoring method described in appendix 9.

[0067] Some or all of the elements (e.g., configurations and functions) described in Supplementary Notes 2 to 7 that are dependent on Supplementary Note 1 may also be dependent on Supplementary Notes 8, 9, and 10 in the same dependent relationship as Supplementary Notes 2 to 7. Some or all of the elements described in any Supplementary Note may be applied to various hardware, software, recording means for recording software, systems, and methods. [Explanation of symbols]

[0068] 1 Train 2 Obstacle monitoring device 3 LiDAR device 3a Latest point cloud storage 3b Past point cloud storage section 4 Rain Sensor 5. Acceleration Sensor 6 First Obstacle Detection Unit 7 Second Obstacle Detection Unit 8. Control Unit 8a Low-confidence measurement mode 8b High reliability measurement mode 8c Threshold Table 9 wheels 10 Motor 11 Brake system 12 Rail 20 Harmful Classes 21 Harmless Classes FV1 First viewing angle FV2 Second viewing angle

Claims

1. A sensing means for sensing a forward direction in a moving direction of the moving body; a first obstacle detection means for detecting an obstacle based on a sensing result of the sensing means; a control means for selectively executing a plurality of control modes including a first control mode in which the moving speed of the moving body is a first moving speed and the sensing field angle of the sensing means is a first field angle, and a second control mode in which the moving speed of the moving body is a second moving speed lower than the first moving speed and the sensing field angle of the sensing means is a second field angle narrower than the first field angle; Including, when the first obstacle detection means detects the obstacle while the control means is executing the first control mode, the control means executes the second control mode so as to capture the obstacle within the second viewing angle. Obstacle monitoring system.

2. a second obstacle detection means configured with a neural network that receives a sensing result from the sensing means and outputs a detection score for each class of the obstacle detected by the first obstacle detection means when the control means is executing the second control mode; The obstacle monitoring system according to claim 1 .

3. A score threshold is set for each class. The control means decelerates or stops the moving object when the detection score exceeds the score threshold in any class.

3. The obstacle monitoring system according to claim 2.

4. a first score threshold is set as the score threshold for a class that impedes the movement of the moving object; A second score threshold higher than the first score threshold is set as the score threshold for a class in which the movement of the moving object is not hindered.

4. The obstacle monitoring system according to claim 3.

5. For each class, the score threshold for bad weather is set lower than the score threshold for good weather.

5. An obstacle monitoring system according to claim 3 or 4.

6. In each class, a score threshold when the moving trajectory currently being moved by the moving object is a curved line is set lower than a score threshold when the moving trajectory is a straight line.

5. An obstacle monitoring system according to claim 3 or 4.

7. the control means does not switch the control mode when the obstacle detected by the first obstacle detection means is off a moving trajectory of the moving object. The obstacle monitoring system according to claim 1 .

8. A sensing means for sensing a forward direction in a moving direction of the moving body; a first obstacle detection means for detecting an obstacle based on a sensing result of the sensing means; a control means for selectively executing a plurality of control modes including a first control mode in which the moving speed of the moving body is a first moving speed and the sensing field angle of the sensing means is a first field angle, and a second control mode in which the moving speed of the moving body is a second moving speed lower than the first moving speed and the sensing field angle of the sensing means is a second field angle narrower than the first field angle; Including, when the first obstacle detection means detects the obstacle while the control means is executing the first control mode, the control means executes the second control mode so as to capture the obstacle within the second viewing angle. Obstacle monitoring device.

9. The computer Detecting an obstacle based on a sensing result of a sensing means that senses a front area in a moving direction of the moving body; selectively executing a plurality of control modes including a first control mode in which the moving speed of the moving body is a first moving speed and the sensing field angle of the sensing means is a first field angle, and a second control mode in which the moving speed of the moving body is a second moving speed lower than the first moving speed and the sensing field angle of the sensing means is a second field angle narrower than the first field angle; when the obstacle is detected while the first control mode is being executed, the second control mode is executed so as to capture the obstacle within the second viewing angle; Obstacle monitoring method.

10. A program for causing a computer to execute the obstacle monitoring method according to claim 9.

Images