Object detection device, object detection method, and computer program for object detection

Abstract

To provide an object detection device capable of appropriately detecting an emergency vehicle.SOLUTION: A object detection device comprises: a lane number detection unit that detects the number of lanes included in a road on which a vehicle runs; a peripheral vehicle detection unit that detects one or more peripheral vehicles positioned around the vehicle from a peripheral image representing the surroundings of the vehicle, generated through a camera mounted on the vehicle; and a determination unit that determines that, when peripheral vehicles, exceeding the number of lanes detected in a predetermined range in the rear of the vehicle, are detected side by side in the width direction of the vehicle, it is more likely that any of the peripheral vehicles is an emergency vehicle that is allowed to travel without following a traffic segment on a road than when peripheral vehicles exceeding the number of lanes are not detected side by side in the width direction of the vehicle.SELECTED DRAWING: Figure 7

Description

[Technical Field]

[0001] The present disclosure relates to an object detection device, an object detection method, and an object detection computer program for detecting an object from an image. [Background technology]

[0002] When approaching from behind an emergency vehicle (such as an ambulance, fire engine, or police vehicle) engaged in a high-priority task, non-emergency vehicles are required to take prescribed measures, such as moving to the shoulder of the road, so as not to obstruct the emergency vehicle's passage.

[0003] Patent Document 1 describes a driving assistance server that assists the driving of emergency vehicles and ordinary vehicles other than emergency vehicles. The driving assistance server described in Patent Document 1 identifies affected vehicles that may affect the driving of the emergency vehicle from the driving routes of the emergency vehicle and ordinary vehicles. The driving assistance server described in Patent Document 1 then determines an evacuation control method for the affected vehicles based on lane change information stored in a map database. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2019-032721 Summary of the Invention [Problem to be solved by the invention]

[0005] Ordinary vehicles are not necessarily capable of communicating with the driving assistance server described in Patent Document 1. Such ordinary vehicles are required to autonomously and appropriately detect emergency vehicles and perform driving control in response to the approach of an emergency vehicle.

[0006] An object of the present disclosure is to provide an object detection device that can appropriately detect emergency vehicles. [Means for solving the problem]

[0007] The object detection device according to the present disclosure includes a lane number detection unit that detects the number of lanes included in the road on which the vehicle is traveling, a surrounding vehicle detection unit that detects one or more surrounding vehicles located around the vehicle from a surrounding image representing the surroundings of the vehicle generated by a camera mounted on the vehicle, and a determination unit that, when more surrounding vehicles than the number of lanes detected in a predetermined range behind the vehicle are detected lined up in the width direction of the vehicle, determines that one of the surrounding vehicles is more likely to be an emergency vehicle that has been permitted to travel in a manner that does not comply with traffic lanes on the road, than when more surrounding vehicles than the number of lanes are not detected lined up in the width direction of the vehicle.

[0008] In the object detection device according to the present disclosure, it is preferable that the determination unit determine that, among the surrounding vehicles, a surrounding vehicle traveling at a higher speed is more likely to be an emergency vehicle.

[0009] In the object detection device according to the present disclosure, it is preferable that the determination unit determines that, among surrounding vehicles, a surrounding vehicle for which no surrounding vehicles are detected in front or behind is more likely to be an emergency vehicle than a surrounding vehicle for which other surrounding vehicles are detected in front or behind.

[0010] In the object detection device according to the present disclosure, it is preferable that the determination unit determines that the surrounding vehicle is less likely to be an emergency vehicle when the aspect ratio, which represents the ratio of the left-right length to the up-down length of the area corresponding to the surrounding vehicle in the surrounding image, is smaller than a predetermined aspect ratio threshold value than when the aspect ratio is larger than the aspect ratio threshold value.

[0011] In the object detection device according to the present disclosure, when an emergency sound generated when an emergency vehicle is traveling is detected from audio information around the vehicle, it is preferable that the determination unit determines that one of the surrounding vehicles is more likely to be an emergency vehicle than when no emergency sound is detected from the audio information.

[0012] The object detection method according to the present disclosure includes detecting the number of lanes on a road on which a vehicle is traveling, detecting one or more surrounding vehicles located around the vehicle from a surrounding image representing the surroundings of the vehicle generated by a camera mounted on the vehicle, and determining that if more surrounding vehicles than the number of lanes detected in a predetermined range behind the vehicle are detected lined up in the width direction of the vehicle, it is more likely that one of the surrounding vehicles is an emergency vehicle that has been permitted to travel in a manner that does not comply with traffic lanes on the road, than if more surrounding vehicles than the number of lanes are not detected lined up in the width direction of the vehicle.

[0013] The object detection computer program disclosed herein causes a processor mounted on the vehicle to detect the number of lanes on a road on which a vehicle is traveling, detect one or more surrounding vehicles located around the vehicle from a surrounding image representing the surroundings of the vehicle generated by a camera mounted on the vehicle, and, if more surrounding vehicles than the number of lanes detected in a specified range behind the vehicle are detected lined up in the width direction of the vehicle, determine that one of the surrounding vehicles is more likely to be an emergency vehicle that has been permitted to travel in a manner that does not comply with traffic lanes on the road, than if more surrounding vehicles than the number of lanes are not detected lined up in the width direction of the vehicle.

[0014] According to the object detection device according to the present disclosure, it is possible to appropriately detect emergency vehicles. [Brief explanation of the drawings]

[0015] [Figure 1] 1 is a schematic configuration diagram of a vehicle in which an object detection device is implemented. [Figure 2] FIG. 1 is a hardware schematic diagram of an ECU. [Figure 3] FIG. 2 is a functional block diagram of a processor included in the ECU. [Figure 4] 1A is a schematic diagram illustrating a first example of object detection, and FIG. 1B is a peripheral image generated in the first example of object detection. [Figure 5] 10(a) is a schematic diagram illustrating a second example of object detection, and FIG. 10(b) is a peripheral image generated in the second example of object detection. [Figure 6] 10(a) is a schematic diagram illustrating a third example of object detection, and FIG. 10(b) is a peripheral image generated in the third example of object detection. [Figure 7] 10 is a flowchart of an object detection process. DETAILED DESCRIPTION OF THE INVENTION

[0016] An object detection device capable of appropriately detecting an emergency vehicle will be described in detail below with reference to the drawings. The object detection device detects the number of lanes included in a road on which a vehicle is traveling. The object detection device also detects one or more surrounding vehicles located around the vehicle from a surrounding image representing the surroundings of the vehicle generated by a camera mounted on the vehicle. When the object detection device detects more surrounding vehicles lined up in the width direction of the vehicle than the number of lanes detected in a predetermined range behind the vehicle, the object detection device determines that one of the surrounding vehicles is more likely to be an emergency vehicle than when the object detection device does not detect more surrounding vehicles lined up in the width direction of the vehicle than the number of lanes detected. An emergency vehicle is a vehicle that is observed to be traveling in a manner that does not comply with traffic classifications on the road. When it is determined that one of the surrounding vehicles is more likely to be an emergency vehicle, the vehicle control device suggests to the driver a predetermined response (e.g., changing the autonomous driving level) in response to the approach of the emergency vehicle.

[0017] FIG. 1 is a schematic diagram of a vehicle in which an object detection device is implemented.

[0018] The vehicle 1 includes a peripheral camera 2, a meter display 3, a GNSS receiver 4 (Global Navigation Satellite System), a storage device 5, and an ECU 6 (Electronic Control Unit). The ECU 6 is an example of an object detection device. The peripheral camera 2, the meter display 3, the GNSS receiver 4, and the storage device 5 are communicably connected to the ECU 6 via an in-vehicle network that complies with a standard such as a controller area network.

[0019] The surrounding camera 2 is an example of a surrounding sensor for generating surrounding data according to the surrounding conditions of the vehicle 1. The surrounding camera 2 has a two-dimensional detector configured with an array of photoelectric conversion elements, such as a CCD or C-MOS, that are sensitive to visible light, and an imaging optical system that forms an image of the area to be photographed on the two-dimensional detector. The surrounding camera 2 is disposed, for example, at the upper front of the vehicle interior, facing forward. The surrounding camera 2 photographs the surrounding conditions of the vehicle 1 through the windshield at a predetermined photographing interval (for example, 1 / 30 to 1 / 10 seconds), and outputs a surrounding image showing the surrounding conditions as surrounding data.

[0020] The vehicle 1 may further include a microphone (not shown) as a surrounding sensor that outputs audio information corresponding to sounds around the vehicle 1.

[0021] The meter display 3 is an example of an output device, and includes, for example, a liquid crystal display. The meter display 3 displays information about an emergency vehicle approaching the vehicle 1 in a manner that is visible to the driver, in accordance with a signal received from the ECU 6 via the in-vehicle network.

[0022] The GNSS receiver 4 receives GNSS signals from GNSS satellites at predetermined intervals and, based on the received GNSS signals, determines the position of the vehicle 1. The GNSS receiver 4 outputs, at predetermined intervals, a positioning signal representing the positioning result of the vehicle 1 based on the GNSS signals to the ECU 6 via the in-vehicle network.

[0023] The storage device 5 is an example of a storage unit, and includes, for example, a hard disk drive or a non-volatile semiconductor memory. The storage device 5 stores map data including information about features such as lane markings in association with positions.

[0024] The ECU 6 determines the possibility that a surrounding vehicle traveling around the vehicle 1 detected from the surrounding image generated by the surrounding camera 2 is an emergency vehicle. If it is determined that there is a high possibility that any of the surrounding vehicles is an emergency vehicle, the ECU 6 suggests to the driver a predetermined response in response to the approach of the emergency vehicle.

[0025] 2 is a hardware schematic diagram of the ECU 6. The ECU 6 includes a communication interface 61, a memory 62, and a processor 63.

[0026] The communication interface 61 is an example of a communication unit, and includes a communication interface circuit for connecting the ECU 6 to an in-vehicle network. The communication interface 61 supplies received data to the processor 63. The communication interface 61 also outputs data supplied from the processor 63 to the outside.

[0027] The memory 62 includes a volatile semiconductor memory and a nonvolatile semiconductor memory. The memory 62 stores various data used in processing by the processor 63, such as a group of parameters (number of layers, layer configuration, kernel, weighting coefficients, etc.) for defining a neural network that operates as a classifier for detecting surrounding vehicles and lane markings from a surrounding image. The memory 62 also stores an aspect ratio threshold for determining the likelihood that a surrounding vehicle is an emergency vehicle based on the aspect ratio of the area corresponding to the surrounding vehicle in the surrounding image. The memory 62 also stores emergency sound data that represents the characteristics of emergency sounds generated when an emergency vehicle is traveling. The memory 62 also stores various application programs, such as an object detection program that executes object detection processing.

[0028] The processor 63 is an example of a control unit and includes one or more processors and their peripheral circuits. The processor 63 may further include other arithmetic circuits such as a logic unit, a numerical calculation unit, or a graphics processing unit.

[0029] FIG. 3 is a functional block diagram of the processor 63 included in the ECU 6. As shown in FIG.

[0030] The processor 63 of the ECU 6 has, as functional blocks, a lane number detection unit 631, a surrounding vehicle detection unit 632, and a determination unit 633. Each of these units in the processor 63 is a functional module implemented by a computer program stored in the memory 62 and executed on the processor 63. A computer program that realizes the functions of each unit in the processor 63 may be provided in a form recorded on a computer-readable portable recording medium such as a semiconductor memory, a magnetic recording medium, or an optical recording medium. Alternatively, each of these units in the processor 63 may be implemented in the ECU 6 as an independent integrated circuit, a microprocessor, or firmware.

[0031] The lane number detection unit 631 detects the number of lanes included in the road on which the vehicle 1 is traveling.

[0032] The lane number detection unit 631 acquires lane information indicating the number of lanes included in the road on which the vehicle 1 is traveling, in the vicinity of the current position of the vehicle 1 identified by the positioning signal received from the GNSS receiver 4, from the storage device 5 that stores map data. Then, the lane number detection unit 631 detects the number of lanes included in the road on which the vehicle 1 is traveling, based on the acquired lane information.

[0033] The lane number detection unit 631 may detect the number of lanes included in the road on which the vehicle 1 is traveling by inputting the peripheral image generated by the peripheral camera 2 into a classifier that has been trained in advance to detect lane markings. The lane number detection unit 631 can obtain the number of lanes included in the road on which the vehicle 1 is traveling by subtracting 1 from the number of lane markings in the direction perpendicular to the traveling direction of the vehicle 1.

[0034] The classifier can be, for example, a convolutional neural network (CNN) with multiple convolution layers connected in series from the input side to the output side. Images containing lane markings are used as training data in advance, and the CNN is trained using a predetermined learning method such as backpropagation, so that the CNN operates as a classifier that detects lane markings from surrounding images.

[0035] The surrounding vehicle detection unit 632 detects one or more surrounding vehicles located around the vehicle 1 from a surrounding image generated by the surrounding camera 2 mounted on the vehicle 1 .

[0036] The surrounding vehicle detection unit 632 detects surrounding vehicles and lane markings located around the vehicle 1 by inputting the surrounding image into a classifier that has been trained in advance to detect vehicles and lane markings. The classifier can be, for example, a CNN. Images containing vehicles and lane markings are used as training data in advance to train the CNN using a predetermined learning method such as backpropagation, so that the CNN operates as a classifier that detects vehicles and lane markings from the surrounding image. The lane number detection unit 631 may detect lane markings using the classifier used by the surrounding vehicle detection unit 632 to detect surrounding vehicles.

[0037] FIG. 4(a) is a schematic diagram illustrating a first example of object detection, and FIG. 4(b) is a peripheral image generated in the first example of object detection.

[0038] In a first example of object detection, the number of lanes included in road R1 around vehicle 1 detected by lane number detection unit 631 from storage device 5 that stores map data is two. Vehicle 1 is traveling in lane L11 on road R1, which has lane L11 divided by lane markings LL11 and LL12 and lane L12 divided by lane markings LL12 and LL13. In lane L11, surrounding vehicles V11 and V13 are traveling behind vehicle 1, in order of proximity to vehicle 1. In lane L12, surrounding vehicles V12 and V14 are traveling behind vehicle 1, in order of proximity to vehicle 1.

[0039] In a first example of object detection, a surroundings camera 2 of a vehicle 1 generates a surroundings image P1 representing the situation behind the vehicle 1.

[0040] The surrounding vehicle detection unit 632 detects surrounding vehicles V11-V14 and lane identification lines LL11-LL13 by inputting the surrounding image P1 to a classifier. In Figure 4(b), the area B1 in the surrounding image P1 corresponding to the surrounding vehicle V11 is indicated by a double line, and areas corresponding to the other surrounding vehicles are omitted.

[0041] The surrounding vehicle detection unit 632 determines the width of lane L11, which is represented at the bottom of area B1 in the surrounding image P1. If the intersection of lane markings LL11-LL12 that define lane L11 and an extension of the bottom of area B1 in the surrounding image P1 is obscured by a surrounding vehicle or the like, the surrounding vehicle detection unit 632 determines the intersection of a lane marking model generated using the unobscured portions of lane markings LL11 and LL12 with the extension of the bottom of area B1, and determines the width of lane L11 from the intersection. The surrounding vehicle detection unit 632 estimates the distance from the surrounding camera 2 of vehicle 1 to the bottom of area B1 using the width of lane L11 included in the map data stored in storage device 5, the width of lane L11 represented in the surrounding image P1, and the focal length of the optical system of the surrounding camera 2 stored in memory 62. The surrounding vehicle detection unit 632 similarly estimates the distances to other surrounding vehicles.

[0042] The determination unit 633 determines whether the positions of the detected nearby vehicles V11-V14 are within a predetermined range D from the vehicle 1. The range D is set to, for example, twice the standard inter-vehicle distance on the road R1, and is stored in the memory 62. The range D may be set to be longer as the vehicle speed of the vehicle 1 increases.

[0043] In the first example of object detection, the determination unit 633 determines that the positions of the surrounding vehicles V11 and V12 are included in the range D. At this time, the number of surrounding vehicles detected lined up in the width direction of the vehicle is two, which is not more than the number of lanes, which is two.

[0044] FIG. 5(a) is a schematic diagram illustrating a second example of object detection, and FIG. 5(b) is a peripheral image generated in the second example of object detection.

[0045] The second example of object detection differs from the first example of object detection in that a nearby vehicle V25 is traveling near lane marking LL22 on road R2, but the rest of the example is the same as the first example of object detection, so a description of the common parts will be omitted. The nearby vehicle V25 is an emergency vehicle, and the position of the nearby vehicle V25 is included in range D.

[0046] In the second example of object detection, the number of surrounding vehicles detected by the determination unit 633 as lined up in the width direction of the vehicle is three, namely surrounding vehicles V21, V22, and V25, which is more than the number of lanes, which is 2. The determination unit 633 determines that there is a higher possibility that one of the surrounding vehicles V21, V22, and V25 in the second example of object detection is an emergency vehicle than there is a higher possibility that one of the surrounding vehicles V11 and V12 in the first example of object detection is an emergency vehicle.

[0047] The determination unit 633 may estimate the distance to the surrounding vehicle in each of a plurality of surrounding images generated at a predetermined time interval, and calculate the speed of each surrounding vehicle from the change in the estimated distance. Vehicles other than emergency vehicles are expected to travel slower than emergency vehicles because they are required to take measures not to interfere with the travel of the emergency vehicle. Therefore, the determination unit 633 may determine that, among the surrounding vehicles determined to have a high possibility of being an emergency vehicle, the faster the surrounding vehicle, the more likely it is to be an emergency vehicle. By determining the possibility that each of the surrounding vehicles is an emergency vehicle, the ECU 6 can suggest to the driver that the vehicle 1 be moved away from the location of the surrounding vehicle that is likely to be an emergency vehicle.

[0048] Vehicles other than emergency vehicles often move to the shoulder of the road to avoid interfering with the emergency vehicle's travel, thereby creating space for the emergency vehicle to travel. In this case, for example, as shown in FIG. 5(a), no other surrounding vehicles are detected in front of or behind the emergency vehicle V25, but other surrounding vehicles V23 and V24 are detected in front of or behind the other surrounding vehicles V21 and V22. Therefore, the determination unit 633 may determine that, among the surrounding vehicles determined to have a high probability of being an emergency vehicle, a surrounding vehicle for which no other surrounding vehicles are detected in front of or behind is more likely to be an emergency vehicle than a surrounding vehicle for which other surrounding vehicles are detected in front of or behind.

[0049] FIG. 6(a) is a schematic diagram illustrating a third example of object detection, and FIG. 6(b) is a peripheral image generated in the third example of object detection.

[0050] The third example of object detection differs from the first example of object detection in that a nearby vehicle V35 is traveling near lane marking LL32 on road R3, but is otherwise similar to the first example of object detection, and therefore a description of the common parts will be omitted. The nearby vehicle V35 is a motorcycle, and the position of the nearby vehicle V35 is included in range D.

[0051] Because motorcycles are narrower than four-wheeled vehicles, the width of the road required for motorcycles to travel is narrower than that required for four-wheeled vehicles to travel on. Especially during traffic jams, motorcycles may travel close to the lane markings that separate adjacent lanes where vehicles are traveling at low speeds. Other vehicles are not required to take measures to avoid obstructing the travel of motorcycles that are not emergency vehicles.

[0052] When the surrounding vehicle detection unit 632 detects surrounding vehicles from the surrounding image, regions corresponding to the surrounding vehicles are identified. In the second example of object detection, the aspect ratio representing the ratio of the left-right length B2w to the up-down length B2h of the region B2 corresponding to the surrounding vehicle V25 detected from the surrounding image P2 is expressed as B2w / B2h. In the third example of object detection, the aspect ratio representing the ratio of the left-right length B3w to the up-down length B3h of the region B3 corresponding to the surrounding vehicle V35 detected from the surrounding image P3 is expressed as B3w / B3h.

[0053] Since emergency vehicles such as ambulances, fire engines, and police vehicles are generally not two-wheeled vehicles, the aspect ratio threshold may be set to a value equivalent to the maximum aspect ratio of a typical two-wheeled vehicle, or may be set to a value equivalent to the minimum aspect ratio of a typical four-wheeled vehicle, or a value between the maximum aspect ratio of a typical two-wheeled vehicle and the minimum aspect ratio of a typical four-wheeled vehicle.

[0054] Because the nearby vehicle V25 is a four-wheeled vehicle and the nearby vehicle V35 is a two-wheeled vehicle, the determination unit 633 determines that the aspect ratio B2w / B2h of the area B2 corresponding to the nearby vehicle V25 is greater than the aspect ratio threshold. Furthermore, the determination unit 633 determines that the aspect ratio B3w / B3h of the area B3 corresponding to the nearby vehicle V35 is smaller than the aspect ratio threshold. The determination unit 633 determines that the nearby vehicle V35 in the third example of object detection is less likely to be an emergency vehicle than the possibility that the nearby vehicle V25 in the second example of object detection is an emergency vehicle. Furthermore, if the aspect ratio of the area corresponding to the fastest nearby vehicle among the multiple nearby vehicles determined to have a high possibility of being an emergency vehicle is smaller than the aspect ratio threshold, the determination unit 633 may change the possibility that any of the multiple nearby vehicles is an emergency vehicle to be lower.

[0055] An emergency vehicle emits an emergency sound when traveling to alert surrounding vehicles of the approach of the emergency vehicle. Therefore, when an emergency vehicle is approaching vehicle 1, it is considered that the emergency sound is included in the audio information acquired by the peripheral microphone mounted on vehicle 1 as a peripheral sensor. When an emergency sound is detected from the audio information, determination unit 633 determines that there is a higher possibility that one of the surrounding vehicles is an emergency vehicle than when no emergency sound is detected from the audio information.

[0056] Fig. 7 is a flowchart of the object detection process. The ECU 6 repeatedly executes the process shown in Fig. 7 at predetermined time intervals (for example, every 1 second) while the vehicle 1 is traveling under automatic driving control.

[0057] First, the lane number detection unit 631 of the processor 63 of the ECU 6 detects the number of lanes included in the road on which the vehicle is traveling (step S1).

[0058] In addition, the surrounding vehicle detection unit 632 of the processor 63 of the ECU 6 detects one or more surrounding vehicles located around the vehicle 1 from a surrounding image representing the surroundings of the vehicle 1 generated by the surrounding camera 2 mounted on the vehicle 1 (step S2).

[0059] The determination unit 633 of the processor 63 of the ECU 6 determines whether or not more surrounding vehicles are detected lined up in the width direction of the vehicle than the number of lanes detected in a predetermined range behind the vehicle (step S3).

[0060] If it is determined that more surrounding vehicles have been detected lined up than the number of lanes (step S3: Y), the determination unit 633 determines that one of the surrounding vehicles is an emergency vehicle with a first probability (step S4), and terminates the object detection process.

[0061] If it is determined that more peripheral vehicles than the number of lanes are not detected lined up (step S3: N), the determination unit 633 determines that one of the peripheral vehicles is an emergency vehicle with a second probability (step S4), and ends the object detection process. The second probability is smaller than the first probability. In other words, if more peripheral vehicles than the number of lanes are detected lined up, the determination unit 633 determines that the vehicle is more likely to be an emergency vehicle than if more peripheral vehicles than the number of lanes are not detected.

[0062] By performing the object detection process in this manner, the ECU 6 can appropriately detect an emergency vehicle.

[0063] The vehicle 1 may have a LiDAR (Light Detection and Ranging) sensor or a RADAR (Radio Detection and Ranging) sensor as a surrounding sensor. The LIDAR sensor or RADAR sensor outputs, as surrounding data, a distance image in which each pixel has a value corresponding to the distance to the object represented by that pixel, based on the surrounding conditions of the vehicle 1.

[0064] It should be understood that those skilled in the art can make various changes, substitutions and alterations thereto without departing from the spirit and scope of the present invention. [Explanation of symbols]

[0065] 1 vehicle 6 ECU 631 Lane number detection unit 632 Surrounding vehicle detection unit 633 Judgment section

Claims

1. a lane number detection unit that detects the number of lanes included in the road on which the vehicle is traveling; a surrounding vehicle detection unit that detects one or more surrounding vehicles located around the vehicle from a surrounding image representing the surroundings of the vehicle generated by a camera mounted on the vehicle; a determination unit that, when a greater number of the surrounding vehicles than the number of lanes detected in a predetermined range behind the vehicle are detected lined up in the width direction of the vehicle, determines that there is a higher possibility that one of the surrounding vehicles is an emergency vehicle that has been permitted to travel in a manner that does not comply with traffic classifications on the road than when a greater number of the surrounding vehicles than the number of lanes are not detected lined up in the width direction of the vehicle; An object detection device comprising:

2. The object detection device according to claim 1 , wherein the determination unit determines that a nearby vehicle traveling at a higher speed is more likely to be the emergency vehicle.

3. 3. The object detection device according to claim 1, wherein the determination unit determines that, among the surrounding vehicles, a surrounding vehicle for which no surrounding vehicles are detected in front or behind is more likely to be the emergency vehicle than a surrounding vehicle for which other surrounding vehicles are detected in front or behind.

4. 3. The object detection device according to claim 1, wherein the determination unit determines that the surrounding vehicle is less likely to be the emergency vehicle when an aspect ratio representing the ratio of the left-right length to the up-down length of an area in the surrounding image corresponding to the surrounding vehicle is smaller than a predetermined aspect ratio threshold value, compared to when the aspect ratio is larger than the aspect ratio threshold value.

5. 3. The object detection device according to claim 1, wherein when an emergency sound generated when the emergency vehicle is traveling is detected from audio information generated around the vehicle by a microphone mounted on the vehicle, the determination unit determines that one of the surrounding vehicles is more likely to be the emergency vehicle than when the emergency sound is not detected from the audio information.

6. Detect the number of lanes on the road on which the vehicle is traveling, Detecting one or more surrounding vehicles located around the vehicle from a surrounding image representing the surroundings of the vehicle generated by a camera mounted on the vehicle; When a greater number of the surrounding vehicles than the number of lanes detected in a predetermined range behind the vehicle are detected lined up in the width direction of the vehicle, it is determined that there is a higher possibility that one of the surrounding vehicles is an emergency vehicle that is permitted to travel in a manner that does not comply with traffic classifications on the road, than when a greater number of the surrounding vehicles than the number of lanes are not detected lined up in the width direction of the vehicle. The object detection method includes:

7. Detect the number of lanes on the road on which the vehicle is traveling, Detecting one or more surrounding vehicles located around the vehicle from a surrounding image representing the surroundings of the vehicle generated by a camera mounted on the vehicle; When a greater number of the surrounding vehicles than the number of lanes detected in a predetermined range behind the vehicle are detected lined up in the width direction of the vehicle, it is determined that there is a higher possibility that one of the surrounding vehicles is an emergency vehicle that is permitted to travel in a manner that does not comply with traffic classifications on the road, than when a greater number of the surrounding vehicles than the number of lanes are not detected lined up in the width direction of the vehicle. A computer program for object detection that causes a processor mounted on the vehicle to execute the above.

Images