Slam device, slam attack countermeasure method, and slam attack countermeasure program

Abstract

The LiDAR SLAM unit (111) performs SLAM using measurement data obtained from a LiDAR sensor mounted on the mobile body, and generates a LiDAR SLAM map. The camera SLAM unit (112) performs SLAM using measurement data obtained from a distance image camera mounted on the mobile body, and generates a camera SLAM map. The attack determination unit (122) determines whether the LiDAR sensor has been attacked based on the difference between the LiDAR SLAM map and the camera SLAM map.

Description

Technical Field

[0001] This disclosure relates to countermeasures against attacks against SLAM. Background Technology

[0002] Autonomous driving systems utilize multiple sensors to achieve safe driving control of a moving vehicle. These sensors work together to compensate for each other's strengths and weaknesses while measuring the surrounding environment of the moving vehicle.

[0003] This use of multiple sensors is called sensor fusion.

[0004] SLAM exists as a technology that utilizes sensor data in autonomous driving systems.

[0005] SLAM is a technique that simultaneously estimates the position of a mobile object and generates a map of its environment. By utilizing SLAM, mobile objects can drive safely even in unfamiliar locations.

[0006] SLAM is short for Simultaneous Localization and Mapping.

[0007] In certain circumstances, there is a risk of attacking autonomous driving systems via sensors.

[0008] For example, an attack could be launched against SLAM, an application used in autonomous driving systems. Due to a SLAM attack, the victim (the moving body) could misorient itself, causing sudden acceleration or deceleration.

[0009] Patent document 1 discloses a technology that addresses the aforementioned problem.

[0010] Specifically, Patent Document 1 discloses a SLAM interpolation technique that utilizes measurement data from LiDAR (Light Detection and Ranging) and cameras. This technique addresses attacks in applications such as synchronization spoofing of LiDAR and using its data to influence SLAM.

[0011] LiDAR is short for Light Detection And Ranging.

[0012] Existing technical documents

[0013] Patent documents

[0014] Patent Document 1: International Publication No. 2023-139793 Summary of the Invention

[0015] The problem that the invention aims to solve

[0016] The purpose of this disclosure is to determine whether there is an attack targeting a lidar sensor used in SLAM.

[0017] Methods for solving problems

[0018] The SLAM device disclosed herein includes: a lidar SLAM unit that performs SLAM using measurement data obtained from a lidar sensor mounted on a mobile body and generates a lidar SLAM map; a camera SLAM unit that performs SLAM using measurement data obtained from a distance image camera mounted on the mobile body and generates a camera SLAM map; and an attack determination unit that determines whether the lidar sensor has been attacked based on the difference between the lidar SLAM map and the camera SLAM map.

[0019] Invention Effects

[0020] According to this disclosure, it is possible to determine whether there is an attack targeting the lidar sensor used in SLAM. Attached Figure Description

[0021] Figure 1 This is a structural diagram of the SLAM device 100 in Embodiment 1.

[0022] Figure 2 This is a structural diagram of the movable body 200 in Embodiment 1.

[0023] Figure 3 This is a functional structure diagram of the SLAM device 100 in Implementation Method 1.

[0024] Figure 4 This is a flowchart of the SLAM attack countermeasure method in Implementation 1.

[0025] Figure 5 This is a diagram illustrating an example of the processing flow of the SLAM attack countermeasure method in Implementation 1.

[0026] Figure 6 This is a hardware structure diagram of the SLAM device 100 in Implementation Method 1. Detailed Implementation

[0027] In the embodiments and accompanying drawings, the same or corresponding elements are labeled with the same reference numerals. Descriptions of elements labeled with the same reference numerals as those already described are appropriately omitted or simplified. Arrows in the figures primarily indicate data flow or processing flow.

[0028] Implementation method 1.

[0029] based on Figures 1 to 6 This section explains countermeasures against SLAM attacks.

[0030] Structural description

[0031] based on Figure 1 Explain the structure of the SLAM device 100.

[0032] The SLAM device 100 is a computer equipped with hardware such as a processor 101, a memory 102, an auxiliary storage device 103, and an input / output interface 104. These hardware components are interconnected via signal lines.

[0033] Processor 101 is an IC that performs computational processing and controls other hardware. For example, processor 101 is a CPU.

[0034] IC is short for Integrated Circuit.

[0035] CPU is short for Central Processing Unit.

[0036] Memory 102 is a volatile or non-volatile storage device. Memory 102 is also referred to as main storage device or main memory. For example, memory 102 is RAM. Data stored in memory 102 is stored in auxiliary storage device 103 as needed.

[0037] RAM is short for Random Access Memory.

[0038] Auxiliary storage device 103 is a non-volatile storage device. For example, auxiliary storage device 103 is ROM, HDD, flash memory, or a combination thereof. Data stored in auxiliary storage device 103 is loaded into memory 102 as needed.

[0039] ROM is short for Read Only Memory.

[0040] HDD is short for Hard Disk Drive.

[0041] Input / output interface 104 is a port for connecting input devices and output devices. Input / output interface 104 is connected, for example, to sensor group 210 and control device 220, which will be described later.

[0042] USB is short for Universal Serial Bus.

[0043] The SLAM device 100 includes elements such as a signal processing unit 110 and a control unit 120. These elements are implemented through software.

[0044] The auxiliary storage device 103 stores a SLAM attack countermeasure program for enabling the computer to function as a signal processing unit 110 and a control unit 120. The SLAM attack countermeasure program is loaded into the memory 102 and executed by the processor 101.

[0045] The OS is also stored in the auxiliary storage device 103. At least a portion of the OS is loaded into the memory 102 and executed by the processor 101.

[0046] The processor 101 executes the OS and the SLAM attack countermeasure program simultaneously.

[0047] OS is short for Operating System.

[0048] The input and output data of the SLAM attack countermeasure program are stored in storage unit 190.

[0049] The memory 102 functions as the storage unit 190. However, auxiliary storage devices such as the auxiliary storage device 103, registers within the processor 101, and cache memory within the processor 101 may also replace the memory 102 or function as the storage unit 190 together with the memory 102.

[0050] SLAM attack countermeasures can be recorded (saved) in a computer-readable manner on non-volatile recording media such as optical discs or flash memory.

[0051] The SLAM device 100 is mounted on a mobile body 200 that moves automatically using SLAM.

[0052] Examples of mobile entities 200 include self-driving vehicles, autonomous mobile robots, and drones.

[0053] based on Figure 2 Explain the structure of the moving body 200.

[0054] The mobile unit 200 is equipped with a SLAM device 100, a sensor group 210 and a control device 220.

[0055] Sensor group 210 consists of multiple sensors.

[0056] The sensor group 210 includes a LiDAR sensor 211, a distance imaging camera 212, and a satellite positioning sensor 213.

[0057] The distance image camera 212 is a camera that determines the distance to an object based on the time-of-flight (ToF) of light such as infrared or visible light. The distance image camera 212 can output image data containing distance data. That is, the distance image camera 212 can perform odometry without special point cloud processing. On the other hand, in order to determine distance using a conventional camera, additional information such as the amount of movement is required. The distance image camera 212 is also called a distance image sensor.

[0058] Satellite positioning sensor 213 is a sensor that uses a satellite positioning system for positioning. An example of a satellite positioning system is the Global Positioning System (GPS).

[0059] In addition to the LiDAR sensor 211, the distance image camera 212, and the satellite positioning sensor 213, the sensor group 210 may also include radar and sonar.

[0060] The control device 220 is a computer that includes elements such as the motion control unit 221.

[0061] The motion control unit 221 performs control over the movement, stopping, and direction change of the mobile body 200 (motion control). Specifically, the motion control unit 221 controls the power unit and steering device of the mobile body 200. Examples of power units include wheels and motors. Illustrations of the power unit and steering device are omitted.

[0062] Figure 3 The functional structure of the SLAM device 100 is shown.

[0063] The signal processing unit 110 includes elements such as a LiDAR SLAM unit 111, a camera SLAM unit 112, and an accuracy determination unit 113.

[0064] The control unit 120 includes a comparison unit 121, an attack judgment unit 122, an output unit 123, and a correction unit 124.

[0065] Description of the action

[0066] The steps of the operation of the SLAM device 100 are equivalent to the steps of a SLAM attack countermeasure method. Furthermore, the steps of the operation of the SLAM device 100 are equivalent to the steps of a processing procedure based on a SLAM attack countermeasure program.

[0067] The following explains the methods for countering SLAM attacks.

[0068] LiDAR sensor 211 performs LiDAR measurements on the periphery of the moving object 200 and outputs measurement data. The output measurement data is input to LiDAR SLAM unit 111.

[0069] The distance image camera 212 performs Time-of-Flight (ToF) measurements on the periphery of the moving object 200 and outputs the measurement data. The output measurement data is input to the camera SLAM unit 112.

[0070] The measurement time interval of the distance image camera 212 is shorter than the measurement time interval of the LiDAR sensor 211, and more than two measurement data of the distance image camera 212 are obtained relative to one measurement data of the LiDAR sensor 211.

[0071] The satellite positioning sensor 213 performs satellite positioning and outputs satellite positioning data. The output satellite positioning data is input to the accuracy judgment unit 113.

[0072] based on Figure 4 Explain the steps involved in countering SLAM attacks.

[0073] The steps of the SLAM attack countermeasures method are repeated.

[0074] In step S101, the LiDAR SLAM unit 111 performs SLAM using measurement data obtained from the LiDAR sensor 211.

[0075] In SLAM, measurement data is used to perform point cloud processing and odometry.

[0076] Point cloud processing is the process of generating point clouds that represent multiple measured locations (measurement points).

[0077] Odometry is a method for deriving a location trajectory. Odometry can use well-known methods.

[0078] By executing SLAM, the position of the moving body 200 is estimated, and a map of the surrounding area of ​​the moving body 200 is generated.

[0079] SLAM performed using measurement data from the LiDAR sensor 211 is called LiDAR-based SLAM.

[0080] Maps generated by LiDAR-based SLAM are called LiDAR SLAM maps or LiDAR-based SLAM maps.

[0081] The location estimated by LiDAR-based SLAM is called the LiDAR SLAM location or LiDAR-based SLAM location.

[0082] The LiDAR SLAM unit 111 outputs LiDAR SLAM data.

[0083] LiDAR SLAM data represents LiDAR SLAM locations and LiDAR SLAM maps.

[0084] The output LiDAR SLAM data is input into the comparison unit 121.

[0085] In step S102, the accuracy judgment unit 113 determines whether the positioning accuracy of the satellite positioning sensor 213 is poor based on the satellite positioning data obtained from the satellite positioning sensor 213.

[0086] For example, the accuracy judgment unit 113 judges whether the positioning accuracy of the satellite positioning sensor 213 is poor.

[0087] Satellite positioning data includes information indicating the number of captured satellites. The number of captured satellites refers to the number of positioning satellites that have been captured. Captured positioning satellites are those that are the sources of the positioning signals received by satellite positioning sensor 213. For example, when the moving object 200 enters a tunnel, satellite positioning sensor 213 may be unable to capture positioning satellites. With a low number of captured satellites, the positioning accuracy of satellite positioning sensor 213 cannot be guaranteed, and its accuracy decreases. Alternatively, the position of satellite positioning sensor 213 may be uncertain.

[0088] The accuracy judgment unit 113 determines whether the positioning accuracy of the satellite positioning sensor 213 is poor based on the number of captured satellites shown in the satellite positioning data.

[0089] When the number of captured satellites is small (when the number of captured satellites is less than the threshold), the accuracy judgment unit 113 judges that the positioning accuracy of the satellite positioning sensor 213 is poor.

[0090] Then, the accuracy judgment unit 113 outputs the judgment result. The output judgment result is input to the camera SLAM unit 112 and the comparison unit 121.

[0091] If the positioning accuracy of the satellite positioning sensor 213 is determined to be good, the process proceeds to step S103.

[0092] In this case, the LiDAR SLAM unit 111 can also regenerate LiDAR SLAM data by performing SLAM using satellite positioning results. As a result, the accuracy of its own position estimation is improved.

[0093] The satellite positioning result represents the position estimated by satellite positioning through satellite positioning sensor 213, which is included in the satellite positioning data.

[0094] The regenerated LiDAR SLAM data is input into the comparison unit 121.

[0095] In step S103, the comparison unit 121 receives the judgment result that the accuracy of the LiDAR SLAM data and the satellite positioning sensor 213 is not different, and outputs the LiDAR SLAM data. The output LiDAR SLAM data is input to the output unit 123.

[0096] The output unit 123 receives LiDAR SLAM data and outputs a LiDAR SLAM map. The output LiDAR SLAM map is input to the motion control unit 221.

[0097] The motion control unit 221 receives the LiDAR SLAM map and uses the LiDAR SLAM map to control the movement of the mobile body 200.

[0098] After step S103, one step of the SLAM attack countermeasure method ends.

[0099] If the positioning accuracy of the satellite positioning sensor 213 is determined to be poor in step S102, the process proceeds to step S111.

[0100] In step S111, the camera SLAM unit 112 receives the judgment result of the positioning accuracy difference of the satellite positioning sensor 213 and controls the distance image camera 212 to perform an action.

[0101] Next, the camera SLAM unit 112 uses the measurement data obtained from the distance image camera 212 to perform SLAM.

[0102] SLAM performed using measurement data from the distance image camera 212 is called camera-based SLAM. The distance image camera 212 can measure both image and distance, thus enabling odometry without the need for special point cloud processing.

[0103] Maps generated using camera-based SLAM are called camera SLAM maps or camera-based SLAM maps.

[0104] The location estimated by camera-based SLAM is called camera SLAM location or camera-based SLAM location.

[0105] Next, the camera SLAM unit 112 outputs camera SLAM data.

[0106] Camera SLAM data represents camera SLAM location and camera SLAM map.

[0107] The output camera SLAM data is input to the comparison unit 121.

[0108] In addition, the camera SLAM unit 112 obtains two or more measurement data from the distance image camera 212, and outputs camera SLAM data for each measurement data from the distance image camera 212.

[0109] Then, the camera SLAM unit 112 controls the distance image camera 212 to stop the distance image camera 212.

[0110] In step S112, the comparison unit 121 receives the judgment result of the difference in positioning accuracy between LiDAR SLAM data, camera SLAM data, and satellite positioning sensor 213.

[0111] Next, the comparison unit 121 compares the LiDAR SLAM map with the camera SLAM map and calculates the difference between the LiDAR SLAM map and the camera SLAM map.

[0112] For example, the comparison unit 121 overlays the camera SLAM map onto the LiDAR SLAM map and calculates the difference between the point cloud in the camera SLAM map and the point cloud in the LiDAR SLAM map. The calculated difference becomes the difference between the LiDAR SLAM map and the camera SLAM map.

[0113] In addition, the comparison unit 121 calculates the difference by comparing the LiDAR SLAM map with the camera SLAM map according to the input SLAM data of each camera.

[0114] That is, the comparison unit 121 compares one LiDAR SLAM map with two or more camera SLAM maps respectively, and calculates two or more differences corresponding to the two or more camera SLAM maps.

[0115] The comparison unit 121 can also compare the LiDAR SLAM position with the camera SLAM position and calculate the difference (of the trajectory) between the LiDAR SLAM position and the camera SLAM position.

[0116] Then, the comparison unit 121 outputs LiDAR SLAM data, camera SLAM data, and differential data.

[0117] The difference data represents the calculated difference.

[0118] The output data is input into the attack judgment unit 122.

[0119] In step S113, the attack determination unit 122 receives LiDAR SLAM data, camera SLAM data, and differential data.

[0120] Next, the attack determination unit 122 determines whether the LiDAR sensor 211 has been attacked based on the difference between the LiDAR SLAM map and the camera SLAM map.

[0121] For example, the attack determination unit 122 calculates statistical values ​​of two or more differences corresponding to SLAM maps from two or more cameras. Then, the attack determination unit 122 uses the calculated statistical values ​​to determine whether the LiDAR sensor 211 has been attacked.

[0122] An example of a statistical value is the standard error. If the standard error deviates from a specified range, the attack determination unit 122 determines that the LiDAR sensor 211 has been attacked. For example, if the standard error σ is below -2.3 or above 2.3σ, the attack determination unit 122 determines that the LiDAR sensor 211 has been attacked. The specified range is appropriately set based on the performance of the LiDAR sensor 211 and the distance imaging camera 212, or the type of moving object 200.

[0123] The attack detection unit 122 can also determine whether the LiDAR sensor 211 has been attacked based on the difference between the LiDAR SLAM map and the camera SLAM map, and the difference between the LiDAR SLAM position and the camera SLAM position.

[0124] When the LiDAR sensor 211 is attacked, an anomaly is considered to exist in the LiDAR SLAM map.

[0125] If the attack determination unit 122 determines that the LiDAR sensor 211 has been attacked, it outputs camera SLAM data. The output camera SLAM data is then input to the output unit 123.

[0126] In this case, the attack determination unit 122 also outputs LiDAR SLAM data. The output LiDAR SLAM data is input to the correction unit 124.

[0127] If the attack determination unit 122 determines that the LiDAR sensor 211 has not been attacked, it outputs LiDAR SLAM data. The output LiDAR SLAM data is then input to the output unit 123.

[0128] If it is determined that the LiDAR sensor 211 has not been attacked, the process proceeds to step S114.

[0129] In step S114, the output unit 123 receives LiDAR SLAM data and outputs a LiDAR SLAM map.

[0130] The output LiDAR SLAM map is input into the motion control unit 221.

[0131] The motion control unit 221 receives the LiDAR SLAM map and uses the LiDAR SLAM map to control the movement of the mobile body 200.

[0132] After step S114, one step of the SLAM attack countermeasure method ends.

[0133] When it is determined in step S113 that the LiDAR sensor 211 has been attacked, the process proceeds to step S121.

[0134] In step S121, the calibration unit 124 receives LiDAR SLAM data and calibrates the LiDAR SLAM map.

[0135] Specifically, the correction unit 124 removes point clouds with outliers from the LiDAR SLAM map. Outliers are values ​​that deviate from the standard error.

[0136] The corrected LiDAR SLAM map is considered a map created using LiDAR-based SLAM and is used for things like track planning and assessment of the surrounding environment.

[0137] The calibration unit 124 outputs the calibrated LiDAR SLAM map.

[0138] The output LiDAR SLAM map is input into the LiDAR SLAM unit 111 for use in the next LiDAR-based SLAM.

[0139] In step S122, the output unit 123 receives camera SLAM data and outputs a camera SLAM map.

[0140] The output camera SLAM map is input into the motion control unit 221.

[0141] The motion control unit 221 receives the camera SLAM map and uses the camera SLAM map to control the motion of the mobile body 200.

[0142] After step S122, one step of the SLAM attack countermeasure method ends.

[0143] based on Figure 5 An example illustrating the processing flow of SLAM attack countermeasures.

[0144] In this example, we assume that the positioning accuracy of the satellite positioning sensor 213 is poor.

[0145] The LiDAR sensor 211 sends measurement data, the LiDAR SLAM unit 111 performs LiDAR-based SLAM, and sends the LiDAR SLAM map to the comparison unit 121 (step S101).

[0146] Using LiDAR-based SLAM, the LiDAR sensor 211 generates a LiDAR SLAM map, for example, every 1 Hz and estimates its own position every 10 Hz. Although the LiDAR sensor 211 relies on laser output, it has a wide measurement range (approximately 100 meters). Therefore, a large-area LiDAR SLAM map is generated. By performing high-speed self-position estimation and low-speed map generation separately, the self-position shift can be reduced, ensuring sufficient time for feature extraction for map generation.

[0147] The distance image camera 212 sends measurement data, the camera SLAM unit 112 performs camera-based SLAM, and sends the camera SLAM map to the comparison unit 121 (step S103).

[0148] Using camera-based SLAM, the camera SLAM unit 112 generates a camera SLAM map, for example, every 5 Hz, and performs its own position estimation every 50 Hz. The distance image camera 212 has a small measurement range (approximately several meters). Therefore, a camera SLAM map of a small area is generated.

[0149] Typically, distance image sensors have a higher frame rate (measurement time interval) than LiDAR sensors. Therefore, camera SLAM maps can be generated with higher accuracy.

[0150] Assume that the frame rate of the distance image camera 212 is one-fifth of the frame rate of the LiDAR sensor 211.

[0151] The comparison unit 121 compares a LiDAR SLAM map with five camera SLAM maps and sends the differential data to the attack judgment unit 122 (step S112).

[0152] Additionally, by changing their respective frame rates, the number of maps used for comparison can be changed.

[0153] The attack determination unit 122 determines whether the LiDAR sensor 211 has been attacked based on the differential data (step S113).

[0154] Effects of Implementation Method 1

[0155] Implementation method 1 can prevent interference attacks on sensors and attacks such as deceptive sensor fusion.

[0156] In Implementation 1, when the positioning accuracy of the satellite positioning sensor 213 is low, the distance image camera 212 operates. This reduces unnecessary functions that would occur under normal conditions.

[0157] The measurement data from the distance image camera 212 can only be used within a small range, but the measurement accuracy is good.

[0158] When the positioning accuracy of the satellite positioning sensor 213 is low, the mobile object 200 is mostly located in narrow places such as inside three-dimensional buildings or tunnels. This condition of the mobile object 200 being located in a narrow place is most suitable for using the distance image camera 212. Therefore, even if the LiDAR sensor 211 is attacked, safe movement can be achieved using the camera SLAM map.

[0159] Supplement to Implementation Method 1

[0160] In order to enable the mobile object 200 to move safely, the SLAM device 100 estimates its own position and generates a map of the surrounding environment.

[0161] During normal movement (when not under attack), the SLAM device 100 uses the LiDAR sensor 211 and the satellite positioning sensor 213 to estimate its own position.

[0162] When the positioning accuracy of the satellite positioning sensor 213 is reduced due to reasons such as the inability to capture satellites inside the tunnel, a distance image camera 212 is used.

[0163] The signal processing unit 110 uses measurement data from the LiDAR sensor 211 and measurement data from the distance image camera 212 to perform SLAM.

[0164] Normally, the signal processing unit 110 only obtains measurement data (signals) from the LiDAR sensor 211 and performs LiDAR-based SLAM using the measurement data from the LiDAR sensor 211.

[0165] When measurement data is sent from the distance image camera 212, the signal processing unit 110 performs SLAM using the measurement data from the LiDAR sensor 211 and the measurement data from the distance image camera 212, respectively, and performs data processing (signal processing) for comparison.

[0166] The steps of the SLAM attack countermeasure method are equivalent to a sensor fusion SLAM algorithm that uses the distance image camera 212 and the LiDAR sensor 211.

[0167] In cases where GPS positioning accuracy is poor, the signal processing unit 110 uses the distance image camera 212 to perform countermeasures.

[0168] When GPS positioning accuracy is poor, the control unit 120 performs correction by comparing a large-area map based on LiDAR SLAM with a small-area map based on camera SLAM.

[0169] When GPS positioning accuracy is poor, the SLAM device 100 prioritizes the distance image camera 212 and performs sensor fusion signal processing. Since no additional processing is required for the distance image camera 212, it is easy to meet the time requirements during installation.

[0170] The SLAM device 100 utilizes a distance image camera 212. The distance image camera 212 performs measurements using pattern illumination at a frame rate 10 times that of the LiDAR sensor 211.

[0171] The control unit 120 compares one LiDAR-based SLAM map with five camera-based SLAM maps and performs attack detection through statistical processing.

[0172] based on Figure 6 Describe the hardware structure of the SLAM device 100.

[0173] The SLAM device 100 includes a processing circuit 109.

[0174] The processing circuit 109 is the hardware that implements the signal processing unit 110 and the control unit 120.

[0175] The processing circuit 109 can be dedicated hardware or a processor 101 that executes the program stored in the memory 102.

[0176] When the processing circuit 109 is dedicated hardware, the processing circuit 109 may be, for example, a single circuit, a composite circuit, a programmable processor, a parallel programmable processor, an ASIC, an FPGA, or a combination thereof.

[0177] ASIC is short for Application Specific Integrated Circuit.

[0178] FPGA is short for Field Programmable Gate Array.

[0179] The SLAM device 100 may also have multiple processing circuits that replace the processing circuit 109.

[0180] In the processing circuit 109, some functions may be implemented by dedicated hardware, while the remaining functions may be implemented by software or firmware.

[0181] Thus, the functions of the SLAM device 100 can be implemented through hardware, software, firmware, or a combination thereof.

[0182] Embodiment 1 is an example of a preferred embodiment and is not intended to limit the scope of this disclosure. Embodiment 1 may be implemented in part or in combination with other embodiments. The steps described using flowcharts or the like may also be appropriately modified.

[0183] The term "part" in the various elements of the SLAM device 100 can also be replaced with "process", "procedure", "circuit" or "line".

[0184] Label Explanation

[0185] 100 SLAM device, 101 processor, 102 memory, 103 auxiliary storage device, 104 input / output interface, 109 processing circuit, 110 signal processing unit, 111 LiDAR SLAM unit, 112 camera SLAM unit, 113 accuracy judgment unit, 120 control unit, 121 comparison unit, 122 attack judgment unit, 123 output unit, 124 correction unit, 190 storage unit, 200 moving body, 210 sensor group, 211 LiDAR sensor, 212 distance image camera, 213 satellite positioning sensor, 220 control device, 221 movement control unit.

Claims

1. A SLAM device comprising: The lidar SLAM unit uses measurement data obtained from lidar sensors mounted on a moving body to perform SLAM and generate a lidar SLAM map. A camera SLAM unit performs SLAM using measurement data obtained from a distance image camera mounted on the mobile body, and generates a camera SLAM map; and The attack detection unit determines whether the lidar sensor has been attacked based on the difference between the lidar SLAM map and the camera SLAM map.

2. The SLAM apparatus according to claim 1, wherein, The SLAM device includes an accuracy determination unit that determines the positioning accuracy of the satellite positioning sensor based on satellite positioning data obtained from the satellite positioning sensor mounted on the mobile body. When the camera SLAM unit determines that the positioning accuracy of the satellite positioning sensor is poor, it generates the camera SLAM map. If the attack determination unit determines that the positioning accuracy of the satellite positioning sensor is poor, it determines whether the lidar sensor has been attacked.

3. The SLAM apparatus according to claim 2, wherein, When the camera SLAM unit determines that the positioning accuracy of the satellite positioning sensor is poor, it causes the distance image camera to operate and obtain the measurement data from the distance image camera.

4. The SLAM apparatus according to claim 2 or 3, wherein, The mobile object is equipped with a control device that uses a SLAM map generated by SLAM to control its movement. The SLAM device has an output unit that outputs the camera SLAM map to the control device when it is determined that the lidar sensor has been attacked, and outputs the lidar SLAM map to the control device when it is not determined that the lidar sensor has been attacked.

5. The SLAM apparatus according to claim 4, wherein, If the positioning accuracy of the satellite positioning sensor is not determined to be poor, the output unit outputs the lidar SLAM map to the control device.

6. The SLAM apparatus according to any one of claims 1 to 5, wherein, The SLAM device includes a comparison unit that compares the lidar SLAM map with the camera SLAM map and calculates the difference between the lidar SLAM map and the camera SLAM map.

7. The SLAM apparatus according to claim 6, wherein, The measurement time interval of the distance image camera is shorter than the measurement time interval of the lidar sensor, and two or more measurement data points of the distance image camera are obtained for one measurement data point of the lidar sensor. The lidar SLAM unit uses the measurement data from the lidar sensor to generate a lidar SLAM map. The camera SLAM unit uses the two or more measurement data from the distance image camera to generate two or more camera SLAM maps. The comparison unit compares the one LiDAR SLAM map with the two or more camera SLAM maps respectively, and calculates two or more differences corresponding to the two or more camera SLAM maps. The attack determination unit calculates the statistical values ​​of the two or more differences and uses the calculated statistical values ​​to determine whether the lidar sensor has been attacked.

8. The SLAM apparatus according to any one of claims 1 to 7, wherein, The SLAM device includes a correction unit that corrects the lidar SLAM map when it is determined that the lidar sensor has been attacked.

9. A method for countering SLAM attacks, wherein, SLAM is performed using measurement data obtained from lidar sensors mounted on a moving object to generate a lidar SLAM map. SLAM is performed using measurement data obtained from a distance imaging camera mounted on the mobile body to generate a camera SLAM map. Based on the difference between the LiDAR SLAM map and the camera SLAM map, it is determined whether the LiDAR sensor has been attacked.

10. A SLAM attack countermeasure program that causes a computer to perform the following processing: LiDAR SLAM processing uses measurement data obtained from a LiDAR sensor mounted on a moving body to perform SLAM and generate a LiDAR SLAM map. Camera SLAM processing, using measurement data obtained from a distance image camera mounted on the moving body to perform SLAM, and generating a camera SLAM map; and The attack detection process determines whether the lidar sensor has been attacked based on the difference between the lidar SLAM map and the camera SLAM map.

Images