Estimation device
By detecting the positions and relative velocities of multiple reflection points on a moving body and calculating the object's angular velocity, the problem of insufficient tracking and angular velocity accuracy when an object turns in the prior art is solved, and high-precision object state estimation is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- DENSO CORP
- Filing Date
- 2025-02-19
- Publication Date
- 2026-07-24
AI Technical Summary
Existing technologies suffer from reduced tracking accuracy and insufficient angular velocity estimation accuracy when tracking an object as it changes from straight to turning, especially when the object passes through a road junction, making it impossible to accurately predict its direction of travel.
By using a sensor unit mounted on the moving body, sensor waves are sent and received to detect the position and relative velocity of multiple reflection points on the object. The angular velocity of the object is calculated using the distance between the reflection points and the relative rotational velocity, avoiding dependence on the shape of the road. The calculation is performed using a reflection point distance calculation unit, a velocity vector calculation unit, a relative rotational velocity calculation unit, and an angular velocity calculation unit.
It achieves high-precision estimation of angular velocity when an object turns, and can accurately track the object without depending on the road shape, improving the tracking performance and the accuracy of angular velocity calculation, and reducing the delay of filtering processing.
Smart Images
Figure CN122459705A_ABST
Abstract
Description
[0001] Cross-references to related applications
[0002] This international application claims the benefit of Japanese Patent Application No. 2024-028628, filed with the Japan Patent Office on February 28, 2024, the disclosure of which is incorporated herein by reference in its entirety. Technical Field
[0003] This disclosure relates to an estimation device mounted on a moving body and used to estimate the state of surrounding objects. Background Technology
[0004] When tracking objects using vehicle-mounted radar, there is sometimes a tracking delay when the object's movement changes from straight to turning. This is because, in vehicle-mounted radar-based object observation, a filtering process that mixes prediction and observation is required to suppress observation errors. In other words, the reason is that during the filtering process, when transitioning from straight to turning, a prediction assuming straight-line movement is made, and the observation result deviates from the object's actual movement.
[0005] As a technique to mitigate this problem, the filter gain is typically adjusted to ensure tracking performance, but this approach may reduce stability. In contrast, Patent Document 1 proposes improving tracking performance by estimating the object's direction of travel and angular velocity based on the road shape.
[0006] Patent Document 1: Japanese Patent Application Publication No. 2021-60245.
[0007] In the estimation device described in Patent Document 1, the direction of travel and angular velocity of an object are estimated based on the shape of the road. Therefore, even if the state of the object changes from straight to turning, the change can be estimated instantly, and the object can be tracked with high precision.
[0008] However, in the estimation device described in Patent Document 1, for example, when an object enters an intersection or passes through a branch point of a road, it is impossible to accurately predict the direction of travel of the object, thus the estimation accuracy of the angular velocity is also reduced. Summary of the Invention
[0009] One aspect of this disclosure is to enable the high-precision estimation of the angular velocity of an object without relying on the shape of the road.
[0010] One aspect of the estimation device disclosed herein is an estimation device mounted on a moving body and estimating the state of surrounding objects, comprising a sensor unit, a velocity vector calculation unit, a reflection point distance calculation unit, a rotational relative velocity calculation unit, and an angular velocity calculation unit.
[0011] The sensor unit transmits sensor waves to the surrounding area and detects the positions and relative velocities of multiple reflection points on the object based on the reflected waves. Additionally, the velocity vector calculation unit calculates the velocity vector at any reference point on the object.
[0012] Furthermore, the reflection point distance calculation unit calculates the distance from the reference point to a straight line with the slope of the reflection point orientation detected by the sensor unit as the reflection point distance. Additionally, the rotational relative velocity calculation unit calculates the relative velocity of the reflection point by subtracting the relative velocity at the reflection point orientation from the relative velocity of the velocity vector calculated by the velocity vector calculation unit. This value is the rotational relative velocity of the reflection point.
[0013] Then, the angular velocity calculation unit calculates the angular velocity of the object based on the reflection point distance and rotational relative velocity of at least one reflection point calculated by the reflection point distance calculation unit and the rotational relative velocity calculation unit.
[0014] In this way, in the estimation device disclosed herein, the angular velocity of an object is estimated based on the position and relative velocity of multiple reflection points of surrounding objects detected by transmitting and receiving sensor waves, without utilizing the shape of the road.
[0015] The reason is that if an object turning is observed by transmitting and receiving sensor waves, a different relative velocity is detected at each reflection point. In other words, in an object turning, through rotational motion, a rotational velocity is generated at each reflection point corresponding to the distance from the center of the turn, and the magnitude of this rotational velocity varies depending on the angular velocity of the object.
[0016] Therefore, in the estimation device of this disclosure, the rotational velocity of the reflection point is obtained by subtracting the relative velocity at the reflection point's orientation from the relative velocity of the reference point's velocity vector, and the angular velocity is calculated based on this rotational relative velocity and the distance to the reflection point.
[0017] Therefore, according to the estimation device of this disclosure, the angular velocity of an object can be estimated without relying on the shape of the road, and the angular velocity can be estimated with high accuracy even when the object passes through a branch point of the road. Attached Figure Description
[0018] Figure 1 This is a block diagram illustrating the hardware structure of the presumed device for the implementation method.
[0019] Figure 2 This is a block diagram illustrating the functional structure of the presumed device in the implementation method.
[0020] Figure 3 This is a diagram illustrating an example of the mounting location and detection range of a sensor unit in a vehicle.
[0021] Figure 4This is another example of a diagram showing the mounting location and detection range of sensors in a vehicle.
[0022] Figure 5 is an explanatory diagram showing the relative velocity and rotational velocity at multiple reflection points of an object.
[0023] Figure 6 This is an explanatory diagram showing the calculation operation of the reflection point distance in the reflection point distance calculation unit.
[0024] Figure 7 This is an explanatory diagram showing the calculation operation of the relative rotational velocity in the relative rotational velocity calculation unit.
[0025] Figure 8 It is an explanatory diagram illustrating the method for calculating the angular velocity of an object.
[0026] Figure 9 It is a flowchart representing the control processes performed in the processing device.
[0027] Figure 10 It means in Figure 9 The flowchart shows the angular velocity calculation process performed in S50.
[0028] Figure 11 It means in Figure 10 The flowchart of the object determination process performed in S110.
[0029] Figure 12 It means in Figure 10 The flowchart of the object reflection point determination process performed in S140.
[0030] Figure 13 It means in Figure 9 The flowchart shows the first example of the state presumption process performed in S60.
[0031] Figure 14 It means in Figure 9 The flowchart for the second example of the state presumption process performed in S60.
[0032] Figure 15 This is an explanatory diagram illustrating the switching action of the filter gain based on the calculated value of angular velocity.
[0033] Figure 16 It means in Figure 9 The flowchart for the third example of state presumption processing performed in S60.
[0034] Figure 17 This is an explanatory diagram illustrating the correction action for the predicted travel angle based on the calculated angular velocity.
[0035] Figure 18This is a flowchart illustrating the control processing performed by the processing apparatus of the second embodiment.
[0036] Figure 19 It means in Figure 18 The flowchart of the collision determination process performed in S180. Detailed Implementation
[0037] [First Implementation Method]
[0038] [structure]
[0039] Reference Figure 1 as well as Figure 2 The structure of the estimation device 10 in this embodiment will be described. The estimation device 10 includes a sensor unit 11 and a processing unit 20, and is mounted on a vehicle 50, which is a moving body.
[0040] Sensor unit 11 includes radar, lidar, sonar, etc. The radar transmits electromagnetic waves such as millimeter waves as sensor waves and receives reflected waves generated when radar waves are reflected by an object 60. The lidar transmits light as sensor waves and receives reflected waves generated when light is reflected by an object. The sonar transmits sound waves as sensor waves and receives reflected waves generated when sound waves are reflected by an object.
[0041] like Figure 3 As shown, the sensor unit 11 can also be mounted at the front center of the vehicle 50 (e.g., the center of the front bumper), having a detection area A1 at the front center of the vehicle 50. Alternatively, as... Figure 4 As shown, in addition to the front center of the vehicle 50, the sensor unit 11 can also be mounted on the left front, right front, left rear, and right rear of the vehicle 50 (for example, the left and right ends of the front bumper and the left and right ends of the rear bumper). That is, in addition to the detection area A1, the sensor unit 11 can also have detection areas A2 on the left front, right front, left rear, and right rear of the vehicle 50. The sensor unit 11 only needs to be mounted on at least one of the following locations: the front center, left front, right front, left rear, and right rear of the vehicle 50.
[0042] The sensor unit 11 acquires multiple observation values by transmitting and receiving sensor waves around the vehicle 50. The sensor unit 11 is a high-resolution sensor; by transmitting a single sensor wave, it can acquire multiple reflected waves reflected from different reflection points on a single object 60, and acquire observation values based on each reflected wave. Therefore, these multiple observation values correspond to different reflection positions of the single object 60.
[0043] In addition, multiple observations include the position of the reflection point (i.e., the distance from the sensor unit 11 to the reflection point, and the orientation of the reflection point relative to the sensor unit 11) and the relative speed between the reflection point and the vehicle 50 (more specifically, the sensor unit 11) as physical quantities.
[0044] Next, the processing device 20 includes a microcomputer comprising a CPU, ROM, RAM, etc. The processing device 20 executes a program stored in a non-transitional physical recording medium via the CPU to implement the functions of the reflection point distance calculation unit 21, the velocity vector calculation unit 23, the rotational relative velocity calculation unit 25, and the angular velocity calculation unit 27. Furthermore, this function calculates the angular velocity of the object 60, in other words, the yaw rate, based on observations of multiple reflection points of a single object 60 acquired by the sensor unit 11.
[0045] In addition, the processing device 20 not only performs the angular velocity calculation function based on the above-mentioned calculation units 21, 23, 25, and 27, but also performs the function of the state estimation unit 30 that estimates the state of the object 60 based on the observation values of multiple reflection points obtained by the sensor unit 11.
[0046] In this embodiment, the state estimation unit 30 performs extended object tracking using a shape model of the object 60. Furthermore, this function calculates physical quantities representing the motion state of the object 60, such as the x-direction position, y-direction position, velocity, direction of travel, and angular velocity of the reference point, as estimated values.
[0047] Furthermore, the x-direction corresponds to the length direction of vehicle 50, and the y-direction corresponds to the width direction of vehicle 50. Additionally, in the reference point of object 60, assuming object 60 is a car, the center positions of the left and right rear wheels (hereinafter, the rear wheel axle centers) are set.
[0048] Extended object tracking is a method that assumes the object has a shape and models the object, then estimates the object's motion state according to a time series. In this embodiment, it is performed in the order described in document A below. The specific estimation steps will be explained in the state estimation process described later.
[0049] Reference A: Soichiro Tokisawa, Keisuke Yoneda, and Naoki Suganuma, “Extended Object Tracking with Both Stability and Real-Time Performance for Autonomous Driving”, Proceedings of the Automotive Technology Conference, Vol. 52, No. 5, September 2021.
[0050] Next, the reflection point distance calculation unit 21, velocity vector calculation unit 23, rotational relative velocity calculation unit 25, and angular velocity calculation unit 27, which realize the angular velocity calculation function, will be explained.
[0051] First, the reflection point distance calculation unit 21 calculates the distance between the positions of multiple reflection points of the object 60 detected by the sensor unit 11 and the reference point of the object 60 as the reflection point distance. Next, the velocity vector calculation unit 23 calculates the velocity vector at the reference point of the object 60 based on the latest state of the object 60 detected by the state estimation unit 30. Here, the velocity vector is a predicted value obtained by predicting the velocity vector at the current moment based on the velocity vector of the past reference point estimated in the state estimation process. For example, if the object's movement is assumed to be constant-speed linear motion, the predicted value is the sum of the previously estimated velocity vector and the velocity vector at the current moment. On the other hand, for example, if turning motion is assumed, in addition to the previously estimated velocity vector, the previously estimated angular velocity is also used, and the value after adding the change in direction corresponding to the angular velocity is used as the predicted value of the velocity vector at the current moment.
[0052] Furthermore, the rotational relative velocity calculation unit 25 calculates the rotational relative velocity of each reflection point by subtracting the relative velocity at the reflection point orientation of the velocity vector calculated by the velocity vector calculation unit 23 from the relative velocities of multiple reflection points detected by the sensor unit 11. In addition, the relative velocity at the reflection point orientation of the velocity vector represents the velocity vector of the reference point calculated by the velocity vector calculation unit 23 converted into a relative velocity with respect to the vehicle 50.
[0053] Then, the angular velocity calculation unit 27 calculates the angular velocity of the object 60 based on the reflection point distance calculated by the reflection point distance calculation unit 21 and the rotational relative velocity calculated by the rotational relative velocity calculation unit 25.
[0054] The calculation result of the angular velocity based on the angular velocity calculation unit 27 is output to the state estimation unit 30, which reflects the calculated value of the angular velocity calculated by the angular velocity calculation unit 27 in the state estimation result of the object 60. As a result, through filtering processing such as the Kalman filter used for state estimation, even if the state estimation of the state estimation unit 30 is delayed when the object 60 turns, this delay can be corrected by the angular velocity calculated by the angular velocity calculation unit 27. Furthermore, this correction method will be explained in the state estimation processing described later.
[0055] [Principle of Angular Velocity Calculation]
[0056] Next, the calculation principle of angular velocity based on the reflection point distance calculation unit 21, velocity vector calculation unit 23, rotational relative velocity calculation unit 25, and angular velocity calculation unit 27 will be explained.
[0057] like Figure 5A As shown, when object 60 is traveling straight, the relative velocities of the multiple reflection points detected by sensor unit 11 are approximately the same. Conversely, when object 60 is turning, as... Figure 5B As shown, the relative velocities of the multiple reflection points are different for each reflection point.
[0058] Thus, the different relative velocities at each reflection point during the turning of object 60 are due to the rotational motion of object 60, which generates a rotational velocity (V) at each reflection point corresponding to the distance (r) from the center of the turn. Moreover, it can be seen from the expression for the velocity of circular motion, "V = r•ω", that this rotational velocity (V) varies according to the angular velocity (ω) of object 60, in other words, the yaw rate.
[0059] Therefore, the yaw rate can be derived from the relationship between the position and velocity vector of any reference point Pr within the object and the relative velocity and orientation of a reflection point at a different position from the reference point Pr. Therefore, in this embodiment, the angular velocity of the object 60 is calculated based on the position and velocity vector of the reference point Pr and the relative velocity and orientation of the reflection point.
[0060] Here, the translational velocity components based on translational movement are the same for the reference point Pr and the reflection point, but a difference arises regarding the rotational velocity components based on rotational motion, caused by the difference in position.
[0061] That is, if the position of the reflection point of the object is set as x and y, the position of any reference point Pr in the object is set as x0 and y0, the velocity of the reference point in the x and y directions is set as Vx0 and Vy0, and the angular velocity of the object is set as ω, then the velocity of the reflection point in the x and y directions, Vx and Vy, can be described as follows (1) and (2), respectively.
[0062] Vx=Vx0-ω(y-y0)…(1)
[0063] Vy=Vy0+ω(x-x0)…(2)
[0064] In addition, the relative velocity Vr of the reflection point can be described as follows (3).
[0065] Vr=Vxcosθ+Vysinθ…(3)
[0066] Then, if we substitute equations (1) and (2) into equation (3), we get equation (4).
[0067] Vr=(Vx0-ω(y-y0))cosθ+(Vy0+ω(x-x0))sinθ
[0068] =Vx0cosθ+Vy0sinθ
[0069] + (-ycosθ + xsinθ)ω
[0070] +(y0cosθ-x0sinθ)ω…(4)
[0071] Here, according to x = rcosθ, y = rsinθ, and xsinθ = ycosθ, equation (4) can be expressed as in equation (5). Then, according to equation (5), equation (6) holds.
[0072] Vr=Vx0cosθ+Vy0sinθ+(y0cosθ-x0sinθ)ω…(5)
[0073] (y0cosθ-x0sinθ)ω=Vr-(Vx0cosθ+Vy0sinθ)…(6)
[0074] In equation (6), the part in parentheses on the left corresponds to the distance from the reference point Pr to the straight line with the slope of the reflection point's orientation; in other words, it corresponds to the positional offset of the reference point Pr towards the reflection point in the circumferential direction. The part on the right corresponds to the velocity obtained by subtracting the relative velocity of the reference point Pr at the reflection point's orientation from the relative velocity Vr of the reflection point (hereinafter, rotational relative velocity); in other words, it corresponds to the velocity obtained by eliminating the circumferential velocity component from the velocity difference between the reference point Pr and the reflection point.
[0075] [Angular velocity calculation action]
[0076] In this embodiment, the angular velocity of the object 60 is calculated based on the velocity vector of the reference point Pr of the object 60 calculated by the state estimation unit 30.
[0077] In other words, such as Figure 6 As shown, the reflection point distance calculation unit 21 calculates the distance between the reference point Pr and the straight line with the slope of the orientation of each reflection point detected by the sensor unit 11.
[0078] In addition, Figure 6 In this context, the distance to the reflection point represents the length of the line orthogonal to the line in the direction of the reflection point, as observed from the reference point Pr. However, as... Figure 6 As shown by the dashed line, the distance to the reflection point can also be calculated as the length of an arc with the radius equal to the distance from the sensor unit 11, which serves as the origin for orientation measurement, to the reference point Pr. In this way, the distance of the straight line from the reference point Pr to the orientation of the reflection point can also be calculated. That is, the distance to the reflection point can be approximately calculated as the length of a straight line orthogonal to the straight line of the orientation of the reflection point, or the length of an arc with the radius equal to the distance from the reference point Pr.
[0079] In addition, the rotational relative velocity calculation unit 25 subtracts the relative velocity Vr from each reflection point. Figure 7 The relative velocity at the reflection point of the velocity vector of the reference point Pr is given as (Vx0cosθ+Vy0sinθ), which is used to calculate the rotational relative velocity.
[0080] like Figure 8 As indicated by the × mark, the angular velocity calculation unit 27 determines the position corresponding to each reflection point in a coordinate space with the aforementioned distance between reflection points as the horizontal axis and the relative rotational velocity as the vertical axis. Then, through... Figure 2 The regression processing unit 27A shown performs linear regression based on the distribution of each reflection point in the coordinate space, and calculates the angular velocity ω of the object 60 based on the slope of the distribution of each reflection point.
[0081] Therefore, the angular velocity calculation unit 27 can calculate the angular velocity of the object 60 without using filtering processes such as Kalman filters, and can instantly determine the angular velocity of the object 60 without response delay. In other words, the angular velocity calculation unit 27 can calculate the instantaneous angular velocity of the object 60, and then calculate the instantaneous yaw rate.
[0082] [Control Processing]
[0083] Next, the control processing performed in the processing device 20 to realize the angular velocity calculation function and the state estimation function will be described. Furthermore, this control processing is repeatedly executed in the processing device 20 at a predetermined processing cycle.
[0084] like Figure 9 As shown, if control processing begins in the processing unit 20, firstly in S10 (S indicates step), the sensor unit 11 transmits and receives sensor waves, thereby acquiring multiple observation values of the object 60 that is being tracked from the sensor unit 11. Furthermore, the object 60 that is being tracked is an object located in front of the vehicle 50 in the direction of travel, and is preset based on the estimation result of the state estimation processing performed in S60, which will be described later.
[0085] Next, in S20, the processing of the velocity vector calculation unit 23 is performed: the velocity vector of the reference point Pr of the object 60 is calculated based on the velocity vector of the reference point Pr of the object 60 that was last estimated in the state estimation process of S60.
[0086] In S30, the process of the reflection point distance calculation unit 21 is performed: based on the multiple observation values of the object 60 obtained from the sensor unit 11 in S10, the distance from the reference point Pr of the object 60 to the straight line with the slope of the reflection point orientation of the multiple reflection points is calculated as the reflection point distance of each reflection point.
[0087] Then, in S40, based on the relative velocities of each reflection point obtained from the sensor unit 11 in S10 and the velocity vector calculated in S20, the rotational relative velocity of each reflection point is calculated by subtracting the relative velocity at the reflection point's orientation from the relative velocity of the reference point Pr. Furthermore, the processing in S40 performs the function of the rotational relative velocity calculation unit 25.
[0088] Thus, if the distance between the reflection points and the relative rotational speed are calculated for each reflection point of the object 60 detected by the sensor unit 11, the process moves to S50 and performs the angular velocity calculation unit 27: calculating the angular velocity of the object 60 according to the angular velocity calculation principle described above.
[0089] Furthermore, in the following S60, the processing of the state estimation unit 30 is performed: based on the observation value obtained from the sensor unit 11 in S10, the angular velocity (i.e., instantaneous angular velocity) calculated in S50, and the state of the object 60 previously estimated in S60, the latest state of the object 60 is estimated.
[0090] Furthermore, if the state estimation process is performed in S60, the process moves to S10, thereby repeatedly executing the above series of processes in a predetermined processing cycle.
[0091] [Angular velocity calculation and processing]
[0092] Next, use Figures 10-12 The angular velocity calculation process performed in S50 is explained.
[0093] like Figure 10 As shown, in the angular velocity calculation process, firstly, in S110, object determination processing is performed: determining whether the object 60 detected by the sensor unit 11 is an object that will be used to calculate the angular velocity. This object determination processing, for example, follows the steps described above. Figure 11 Perform the steps shown.
[0094] That is, in the object determination process, such as Figure 11 As shown, firstly, in S210, it is determined whether the distance between object 60 and the vehicle 50 is below a pre-set threshold for close-range determination. Then, if it is determined in S210 that the distance is below the threshold, the process moves to S250, where object 60 is determined to be a non-object object for which angular velocity is not calculated.
[0095] On the other hand, if it is determined in S210 that the distance is greater than the threshold, then proceed to S220 to determine whether the number of reflection points detected by the sensor unit 11 is below a preset threshold for determining the number of reflection points. If it is determined in S220 that the number of reflection points is below the threshold, then proceed to S250 to determine that the object 60 is a non-object. If it is determined in S220 that the number of reflection points is greater than the threshold, then proceed to S240.
[0096] In step S240, it is determined whether the absolute value of the acceleration of object 60 is above a pre-set threshold for acceleration / deceleration determination. If, in S240, the absolute value of the acceleration is determined to be above the threshold, and object 60 is being accelerated or decelerated, the process moves to S250, where object 60 is determined to be a non-object. Conversely, if, in S240, the acceleration is determined to be below the threshold, and object 60 is not being accelerated or decelerated, the process moves to S240. Then, in S240, object 60 is determined to be an object for angular velocity calculation, and the object determination process ends.
[0097] Thus, in the object identification process, if the number of reflection points detected by the sensor unit 11 is greater than the predetermined threshold when the object 60 separates from the vehicle 50, and the object 60 is not accelerated or decelerated, then the object 60 is identified as the object. This is because, under conditions where angular velocity cannot be calculated with high precision, the object 60 is considered a non-object and its angular velocity is not calculated.
[0098] In other words, when the distance between object 60 and vehicle 50 is short, it is assumed that the reflected wave from object 60 is prone to propagation, resulting in unwanted signals being included in the received signal, thus reducing the accuracy of angular velocity calculation. Furthermore, when the number of reflection points is small, it is assumed that the aforementioned linear regression becomes unstable, further reducing the accuracy of angular velocity calculation. Additionally, when object 60 is accelerated or decelerated, it is assumed that the velocity vector of the reference point Pr obtained in the state estimation processing of the state estimation unit 30 deviates from the actual velocity vector, further reducing the accuracy of angular velocity calculation.
[0099] Therefore, in this embodiment, under these conditions, object 60 is determined to be a non-object, and its angular velocity is not calculated. Furthermore, the determination criteria used to determine whether object 60 is an object or a non-object are not limited to the three criteria described above; two or one of the three criteria may also be used. Additionally, other determination criteria may be added as determination criteria.
[0100] like Figure 10 As shown, if the object determination process described above is performed in S110, the process moves to S120 to determine whether object 60 is determined to be a processing object (i.e., an object) in the object determination process. Then, if object 60 is determined not to be a processing object in S120, the angular velocity is not calculated in subsequent processes, and the angular velocity calculation process ends.
[0101] On the other hand, if it is determined in S120 that object 60 is the processing target, then proceed to S130 to determine whether the object reflection point determination process of S140 has been performed on all reflection points obtained from sensor unit 11. Then, if it is determined in S130 that the object reflection point determination process of S140 has been performed on all reflection points, then proceed to S170.
[0102] Furthermore, if it is determined in S130 that the object reflection point determination process in S140 has not been performed on all reflection points, then the process moves to S140 to perform the object reflection point determination process. The object reflection point determination process in S140 is the process of determining whether the reflection point of the object 60 detected by the sensor unit 11 is an object reflection point suitable for calculating angular velocity or a non-object reflection point unsuitable for calculating angular velocity.
[0103] In the object reflection point determination process, reflection points that have not undergone this determination process are obtained from the multiple reflection points detected by the sensor unit 11, and then... Figure 12 The steps shown determine whether the reflection point is the object's reflection point.
[0104] That is, in the object reflection point determination process, in S310, it is determined whether the received power of the reflected wave from the reflection point is below a pre-set threshold for receiving power determination. In S310, if it is determined that the received power of the reflection point is below the threshold, it is considered that the received signal from the reflection point is affected by noise due to micro-Doppler, multipath, etc., and therefore the process moves to S350. Then, in S350, the reflection point is determined to be a non-object reflection point where angular velocity is not calculated.
[0105] On the other hand, if it is determined in S310 that the received power of the reflection point exceeds a threshold, the process moves to S320 to determine whether the velocity difference between the reflection point and the reference point Pr is above a pre-set threshold for velocity difference determination. In S320, if it is determined that the velocity difference between the reflection point and the reference point Pr is above the threshold, it is considered that the velocity of the reflection point cannot be detected normally due to micro-Doppler, etc., and therefore the process moves to S350 to determine that the reflection point is a non-object reflection point.
[0106] Furthermore, if it is determined in S320 that the velocity difference between the reflection point and the reference point Pr is less than the threshold, then the process moves to S330, where the orientation of the reflection point of the sensor unit 11 is used to determine whether the reflected wave is a multiple wave.
[0107] Here, "multiple waves" in azimuth estimation refers to the state where multiple reflected waves are detected at the same distance during the azimuth estimation process. For example, in the azimuth estimation process of extracting peaks above a threshold from the angular FFT spectrum, the case where multiple peaks are above the threshold is equivalent to this.
[0108] Furthermore, in the case where there are multiple reflected waves in the estimation of the orientation of the reflected wave, it is believed that the detection accuracy of the orientation of the reflection point, or in other words, the location of the reflection point, is low. Therefore, it is moved to S350 and determined that the reflection point is a non-object reflection point.
[0109] Next, if it is determined in S330 that the reflected wave is not multiple waves in the azimuth estimation of the reflection point, the process moves to S340, where the reflection point is determined to be the object reflection point used for angular velocity calculation. In other words, in the object reflection point determination process, for each reflection point, it is confirmed that the received power exceeds a threshold, the velocity difference with the reference point Pr is less than a threshold, and the azimuth of the reflection point can be correctly estimated based on the reflected wave from the reflection point. Furthermore, when these three conditions are met, the reflection point is determined to be the object reflection point.
[0110] Furthermore, in the object reflection point determination process, after executing S340 or S350, the object reflection point determination process ends and moves to... Figure 10 S150. Then, when the object reflection point determination process begins, for reflection points that have not been determined to be object reflection points, the same steps as described above are followed to determine whether they are object reflection points.
[0111] In the above object reflection point determination process, by implementing the determination processes S310 to S330, the reflection point is determined to be an object reflection point when the above three determination conditions are met. However, in the object reflection point determination process, at least one of the determination processes S310 to S330 can also be executed. That is, in the object reflection point determination process, the reflection point can also be determined to be an object reflection point when one or two of the above three determination conditions are met.
[0112] Next, in Figure 10 In S150, it is determined whether the reflection point was determined to be a processing object (i.e., an object reflection point) in the object reflection point determination process of S140. Then, if it is determined in S150 that the reflection point is not a processing object, the process moves to S130.
[0113] Furthermore, if the reflection point is determined to be the object to be processed in S150, the process proceeds to S160. In S160, the angular velocity calculation information of the reflection point determined to be the object's reflection point in S140 is stored as reflection point information in a recording medium such as RAM, and the process proceeds to S130. Additionally, the reflection point information includes the reflection point distance and rotational relative velocity calculated in the reflection point distance calculation process in S30 and the rotational relative velocity calculation process in S40.
[0114] Next, in S170, which is executed when it is determined in S130 that object reflection point determination processing has been performed on all reflection points, the angular velocity of object 60 is calculated based on the reflection point distance and relative rotation speed of each reflection point stored in the recording medium in S160.
[0115] In other words, in S170, as described above, the angular velocity of object 60 is calculated according to the following steps: Within a coordinate space with the distance to the reflection point and the relative rotational velocity as parameters, the positions corresponding to each reflection point are determined, and the slope of the distribution of each reflection point is calculated using linear regression. Then, if the angular velocity is calculated in S170, the process ends. Figure 9 The angular velocity calculation process shown in S50 is followed by the state estimation process in S60.
[0116] [State estimation processing]
[0117] S60's state estimation process, for example, according to... Figure 13 Perform the steps shown.
[0118] That is, such as Figure 13 As shown, in the state estimation process, firstly, a prediction process is performed in S410. In this prediction process, the outline of the object 60 (hereinafter, the predicted outline) is predicted based on the predetermined shape model of the object 60 and the estimated value calculated in the past processing cycle (e.g., the last processing cycle).
[0119] Furthermore, the shape model of object 60 is determined, for example, by selecting from a number of pre-prepared models such as circular models, elliptical models, and rectangular models, a model suitable for estimating the size and shape of object 60 based on observations from multiple reflection points.
[0120] The predicted profile corresponds to the area of the object that is predicted based on past estimated values, i.e., the state of the reference point Pr.
[0121] Next, in S420, an association process is performed to associate each of the multiple predicted values calculated in S410 with the observation values obtained from the sensor unit 11 in S10, thereby generating an association set.
[0122] Next, in S430, the estimation process is performed: a Kalman filter or other filter is applied to the correlation set calculated in S40 to calculate the current estimated value P2.
[0123] Furthermore, in the estimation process of S430, an extended Kalman filter, or similar nonlinear filter, is used to calculate the update amount of the estimated value based on the observations correlated with the predicted value. Then, the estimated value is updated based on the update amounts of the predicted and estimated values. This updated estimated value becomes the estimated value for the current processing cycle. Additionally, the estimated value includes at least a parameter representing the velocity vector of the reference point Pr. For example, this could be any combination of longitudinal and lateral velocities in any coordinate system, or any combination of the magnitude and orientation of the velocities.
[0124] Next, in S440, an angular velocity mixing process is performed, which mixes the angular velocity of the reference point Pr (hereinafter, the angular velocity estimation value) included in the estimation process updated in S430 and the angular velocity calculated in the angular velocity calculation process in S50 (hereinafter, the angular velocity calculation value).
[0125] This angular velocity mixing process, for example, uses the following equation (7) which includes a coefficient α smaller than 1, to mix the estimated angular velocity value and the calculated angular velocity value, and update the estimated angular velocity value. Furthermore, in the following equation (7), ωfil represents the estimated angular velocity value, and ωobs represents the calculated angular velocity value.
[0126] ωfil=(1-α)•ωfil+α•ωobs…(7)
[0127] Then, if the angular velocity mixing process of S440 is performed, the state estimation process of S60 ends, and the process is moved to... Figure 9 S10 begins the angular velocity calculation and state estimation for the next processing cycle.
[0128] [Effect]
[0129] As explained above, in the estimation device 10 of this embodiment, the angular velocity calculation unit 27 calculates the angular velocity of the object 60 based on the distance between at least one reflection point of the object 60 and the reference point Pr, and the relative rotational velocity of the reflection point.
[0130] Therefore, according to the angular velocity calculation unit 27 of this embodiment, the angular velocity of the object 60 (i.e., the calculated angular velocity value) can be calculated without utilizing the road shape or filtering processes such as Kalman filters. Thus, the angular velocity of the object 60 can be calculated accurately and without response delay, corresponding to the actual movement state of the object 60.
[0131] Furthermore, the angular velocity calculation unit 27 performs object determination processing to determine whether any one of the following conditions is met: the distance to the vehicle 50 is below a threshold, the number of reflection points detected by the sensor unit 11 is below a threshold, or the absolute value of the object's acceleration is above a threshold. Moreover, if any one of these conditions is met, it is determined that the accuracy of the angular velocity calculation is reduced, and the angular velocity of the object 60 is not calculated.
[0132] In addition, the angular velocity calculation unit 27 performs object reflection point determination processing on the object 60 that is determined to be an object in the object determination processing. Then, in the object reflection point determination processing, reflection points detected by the sensor unit 11 that have a received power below a threshold, reflection points whose velocity difference with the reference point Pr is above a threshold, and reflection points whose reflected waves are multiple waves in the azimuth estimation are excluded from the objects of angular velocity calculation.
[0133] Therefore, the angular velocity calculation unit 27 can select from the multiple reflection points detected by the sensor unit 11 a reflection point from which the sensor unit 11 can reliably receive the reflection signal and accurately detect the position and velocity of the reflection point. Then, based on the reflection point distance and the relative rotational velocity of the selected reflection point, the angular velocity is calculated, thereby improving the accuracy of the angular velocity calculation.
[0134] Furthermore, in this embodiment, the calculation result of the angular velocity based on the angular velocity calculation unit 27 (i.e., the calculated angular velocity value) is reflected in the state estimation result of the object 60 estimated by the state estimation unit 30. Specifically, in the above-described angular velocity mixing process, the calculated angular velocity value is used to correct the estimated angular velocity value in the state of the object 60 estimated by the state estimation unit 30.
[0135] As a result, the calculated angular velocity value, as an instantaneous value, can be used to correct the estimated angular velocity value obtained through filtering processes such as Kalman filters, thereby suppressing the estimation delay contained in the final estimated vehicle state. Consequently, in vehicle 50, the vehicle 50 can be controlled more appropriately based on this estimation result.
[0136] [Variation Example]
[0137] In the above embodiment, as a first example of making the calculated angular velocity value calculated by the angular velocity calculation unit 27 reflect the estimation result of the vehicle state based on the state estimation unit 30, it is described that in the angular velocity mixing process, the estimated angular velocity value is mixed with the calculated angular velocity value to correct the estimated angular velocity value.
[0138] However, in order to make the calculated angular velocity value reflect the estimated vehicle state and thus enable a more appropriate estimation of the vehicle state, it is not necessarily necessary to use the calculated angular velocity value to correct the estimated angular velocity value obtained by the state estimation unit 30.
[0139] Therefore, in this modified example, a second and a third example of a method for making the calculated angular velocity value calculated by the angular velocity calculation unit 27 reflect the estimation result of the vehicle state based on the state estimation unit 30 will be described.
[0140] like Figure 14 As shown, in the state estimation process of the second example, firstly, in S400, it is determined whether the absolute value of the angular velocity calculated in the angular velocity calculation process in S50 is above a preset threshold. Then, if it is determined in S400 that the absolute value of the angular velocity calculated is above the threshold, that is, if the yaw rate of object 60 is large, the process moves to S405, increases the filter gain from the initial value, and then moves to S410.
[0141] On the other hand, if the absolute value of the calculated angular velocity is determined to be less than the threshold in S400, that is, if the yaw rate of the object 60 is small, the filter gain is not increased, and the process is moved to S410.
[0142] This is because, when the object's angular velocity is large (60°), the reliability of predictions based on S410 is considered to decrease. Therefore, if... Figure 15 As shown, it is believed that the observed values at each reflection point, compared with the case where the angular velocity of the object is small at 60°, can improve the tracking performance by increasing the filter gain.
[0143] Then, after S410, execute... Figure 13 The processes shown in S410 to S430 end the state estimation process.
[0144] Thus, in the state estimation process of the second example, the filter gain used in the state estimation unit 30 is corrected based on the calculated angular velocity value calculated by the angular velocity calculation unit 27. This also improves the state estimation accuracy of the object 60 based on the state estimation unit 30.
[0145] Next, in the state presumption process of the third case, as follows: Figure 16 As shown, first, after performing the above prediction processing in S410, velocity vector prediction correction processing is performed in S415, and then moved to S420.
[0146] In velocity vector prediction correction processing, such as Figure 17 As shown, the change in the direction of the velocity of the reference point Pr of the object 60 (hereinafter, change in velocity direction) is corrected based on the angular velocity calculated by the angular velocity calculation unit 27, thereby correcting the velocity vector prediction value of this processing cycle.
[0147] Then, after S420, execute Figure 13 The processes shown in S420 and S430 end the state estimation process.
[0148] Thus, in the state estimation process of the third example, the change in velocity direction of the reference point Pr estimated in the previous processing cycle is corrected based on the calculated angular velocity value, thereby correcting the velocity vector prediction value. Therefore, the velocity and direction of travel of the reference point Pr of the object 60 can be estimated more appropriately. Consequently, the state estimation accuracy of the object 60 based on the state estimation unit 30 can be improved.
[0149] Furthermore, the correction of the filter gain in the second example and the correction of the velocity vector prediction in the third example can also be implemented in combination with other corrections such as the correction of the angular velocity estimation in the first example.
[0150] [Second Implementation]
[0151] The basic structure of the second embodiment is the same as that of the first embodiment; therefore, the differences from the first embodiment will be described below. Furthermore, the same reference numerals as in the first embodiment denote the same structures, as described previously.
[0152] The difference between the second implementation method and the first implementation method is that, for example Figure 2 As shown by the dashed line, the processing device 20 includes a collision determination unit 32. Furthermore, this collision determination unit 32 performs angular velocity calculation processing executed by the processing device 20.
[0153] That is, such as Figure 18 As shown, in the angular velocity calculation process of this embodiment, if the angular velocity is calculated in S170 or it is determined in S120 that the object 60 is not the processing object, the angular velocity calculation process ends after performing collision determination processing in S180.
[0154] like Figure 19 As shown, in the collision determination process of S180, firstly, in S510, it is determined whether the direction of the velocity vector of the object 60, specifically the direction of travel of the reference point Pr, which is estimated in the state estimation process based on the state estimation unit 30, intersects the travel path of the vehicle 50. Then, if the direction of the velocity vector of the object 60 does not intersect the travel path of the vehicle 50, in S560, it is determined that the vehicle 50 does not collide with the object 60, which is a "non-collision" condition, and the collision determination process ends.
[0155] Next, if it is determined in S510 that the direction of the velocity vector of object 60 intersects the travel path of vehicle 50, then proceed to S520. In S520, it is determined whether the time to collision (hereinafter, TTC), calculated by dividing the distance between vehicle 50 and object 60 by the difference in velocity (i.e., relative velocity), is below a preset threshold for collision determination. Furthermore, TTC is an abbreviation for Time to Collision.
[0156] If TTC is determined to be greater than the threshold in S520, then it is determined to be "non-collision" in S560, and the collision determination process ends. If TTC is determined to be below the threshold in S520, then it moves to S530.
[0157] In S530, it is determined whether the angular velocity of object 60 has been calculated in the angular velocity calculation process. Then, if it is determined in S530 that the angular velocity has not been calculated, since the probability of collision between vehicle 50 and object 60 is high, the process moves to S550, determines "collision", and ends the collision determination process.
[0158] Furthermore, if an angular velocity is calculated in S530, the process moves to S540 to determine whether the absolute value of the calculated angular velocity is below the threshold used for collision determination. If the absolute value of the angular velocity is below the threshold, the probability of the vehicle 50 colliding with the object 60 is high, so the process moves to S550, determines a "collision," and ends the collision determination process.
[0159] On the other hand, if in S540 the absolute value of the angular velocity is determined to be greater than the threshold, it is determined that the object 60 has changed its travel path, the possibility of the vehicle 50 colliding with the object 60 is reduced, and the process is moved to S560, where it is determined to be "non-collision" and the collision determination process ends.
[0160] In this way, the estimation device 10 in this embodiment is the same as the estimation device 10 in the first embodiment. It can not only calculate the instantaneous acceleration of the surrounding object 60 and estimate the state of the object 60, but also determine whether there is a collision with the object 60 based on the estimation result.
[0161] Furthermore, since the instantaneous angular velocity calculated in the angular velocity calculation process is used in this collision determination, the instantaneous angular velocity of the object 60, which is determined to be of high probability of collision due to TTC, is large, thus reducing the probability of collision and suppressing false "collision" determinations. Therefore, it is possible to suppress the situation where warnings are output to the occupants of vehicle 50 or the automatic driving device due to false "collision" determinations, thereby preventing the steering control device of vehicle 50 from being erroneously controlled.
[0162] [Other Implementation Methods]
[0163] The embodiments of this disclosure have been described above, but this disclosure is not limited to the above embodiments and can be implemented in various ways.
[0164] In the above embodiment, it is described that the state estimation unit 30 models the shape of the object and performs state estimation including angular velocity according to the extended object tracking technology described in document A. However, the state estimation of the object does not necessarily need to be performed according to the steps described in the above embodiment, and can be appropriately modified. Moreover, even in this case, by correcting the estimation results based on the angular velocity, etc., of the state estimation unit 30 using the angular velocity calculation value described above, the tracking accuracy of the object can be improved.
[0165] In the above embodiments, it is described that the estimation device 10 is mounted on a vehicle 50 such as an automobile, but it can also be mounted on a mobile body other than an automobile. For example, the estimation device 10 can also be mounted on a mobile body such as a ship, an airplane, a motorcycle, or an unmanned aerial vehicle.
[0166] The estimation device 10 and method described in this disclosure can also be implemented using a dedicated computer, which is provided by comprising a processor and memory programmed to perform one or more functions embodied in a computer program. Alternatively, the estimation device 10 and method described in this disclosure can also be implemented using a dedicated computer provided by comprising a processor configured with one or more dedicated hardware logic circuits. Alternatively, the estimation device 10 and method described in this disclosure can also be implemented using one or more dedicated computers, which are composed of a combination of a processor and memory programmed to perform one or more functions and a processor configured with one or more hardware logic circuits. In addition, the computer program can also be stored as instructions executed by the computer on a computer-readable non-transitional tangible recording medium. In the method of implementing the functions of the various parts included in the estimation device 10, it is not necessary to include software, and all its functions can be implemented using one or more hardware components.
[0167] Multiple functions of a single component in the above embodiments can be achieved through multiple components, or a single function of a single component can be achieved through multiple components. Alternatively, multiple functions of multiple components can be achieved through a single component, or a single function achieved by multiple components can be achieved through a single component. Furthermore, a portion of the structure in the above embodiments may be omitted. Additionally, at least a portion of the structure in other above embodiments may be added to or replaced.
[0168] In addition to the estimating device 10 described above, this disclosure can also be implemented in various ways, such as a system incorporating the estimating device 10, a program for enabling a computer to function as the estimating device, a non-transitional physical recording medium such as a semiconductor memory containing the program, and an object estimating method.
[0169] [The technical concepts disclosed in this specification]
[0170] [Project 1]
[0171] An estimation device is a device mounted on a moving body that estimates the state of surrounding objects, wherein...
[0172] The aforementioned estimation device includes:
[0173] The sensor unit (11) is configured to send sensor waves to the surroundings and detect the position and relative velocity of multiple reflection points on the object based on the reflected waves of the sensor waves.
[0174] The velocity vector calculation unit (23) is configured to calculate the velocity vector of any reference point of the object.
[0175] The reflection point distance calculation unit (21) is configured to calculate the distance from the reference point to a straight line with the slope of the reflection point direction detected by the sensor unit as the reflection point distance;
[0176] The rotational relative velocity calculation unit (25) is configured to calculate the rotational relative velocity of the reflection point, which is a value obtained by subtracting the relative velocity at the orientation of the reflection point from the relative velocity of the reflection point detected by the sensor unit, based on the velocity vector of the reference point calculated by the velocity vector calculation unit; and
[0177] The angular velocity calculation unit (27) is configured to calculate the angular velocity of the object based on the reflection point distance and the rotational relative velocity of at least one of the reflection points calculated by the reflection point distance calculation unit and the rotational relative velocity calculation unit.
[0178] [Project 2]
[0179] According to the presumed apparatus described in Project 1, wherein,
[0180] The aforementioned angular velocity calculation unit is configured to calculate the angular velocity based on the magnitude of the relative rotational velocity of at least one of the aforementioned reflection points relative to the distance from the aforementioned reflection points.
[0181] [Project 3]
[0182] According to the presumed apparatus described in Project 2, among which,
[0183] The aforementioned angular velocity calculation unit includes a regression processing unit (27A), which is configured to perform linear regression based on the distribution of the reflection points in a coordinate space using the distance between the reflection points and the relative rotational velocity as parameters, and calculate the slope in the coordinate space.
[0184] The angular velocity calculation unit is configured to calculate the angular velocity based on the slope calculated by the regression processing unit.
[0185] [Project 4]
[0186] According to the presumed apparatus described in any one of Items 1 to 3, among which,
[0187] The angular velocity calculation unit performs an exclusion determination on the multiple reflection points detected by the sensor unit based on at least one of the following determination conditions a) to c):
[0188] a) The received power of the reflected wave is below the threshold.
[0189] b) The speed difference from the aforementioned benchmark point is above the threshold.
[0190] c) The above reflected waves are multiple waves.
[0191] The aforementioned estimation device is configured to exclude the aforementioned reflection point that is determined to meet at least one of the aforementioned determination conditions a) to c) in the aforementioned exclusion determination from the objects of the aforementioned angular velocity calculation.
[0192] [Project 5]
[0193] According to the presumed apparatus described in any one of Items 1 to 4, among which,
[0194] The aforementioned angular velocity calculation unit is configured to either not perform the aforementioned angular velocity calculation or not output the aforementioned angular velocity calculation result if any one of the following conditions is met: the distance from the aforementioned object is less than or equal to a predetermined value; the number of the aforementioned reflection points is less than or equal to a predetermined value; or the absolute value of the acceleration of the aforementioned object is greater than or equal to a threshold value.
[0195] [Project 6]
[0196] According to the presumed apparatus described in any one of items 1 to 5, among which,
[0197] The system includes a state estimation unit (30) configured to estimate the state of the object corresponding to the plurality of reflection points.
[0198] The aforementioned state estimation unit is configured to reflect the calculated value of the angular velocity calculated by the aforementioned angular velocity calculation unit, i.e., the calculated value of the angular velocity, in the state estimation result of the aforementioned object.
[0199] [Project 7]
[0200] According to the presumed apparatus described in Project 6, among which,
[0201] The state estimation unit is configured to estimate a state including the angular velocity using a method different from that of the angular velocity calculation unit, and is configured to update the angular velocity estimation value by mixing the angular velocity estimation value with the angular velocity calculation value. The angular velocity estimation value is the estimation result of the angular velocity included in the state estimation result of the object, and the angular velocity calculation value is calculated by the angular velocity calculation unit.
[0202] [Project 8]
[0203] According to the presumed apparatus described in item 6 or item 7, wherein,
[0204] The state estimation unit is configured to estimate the state including the angular velocity using a method different from that of the angular velocity calculation unit, and is configured to increase the filter gain of the filter used by the state estimation unit to estimate the state of the object when the absolute value of the angular velocity calculated by the angular velocity calculation unit is above a predetermined value.
[0205] [Project 9]
[0206] According to the presumed apparatus described in any one of Items 6 to 8, among which,
[0207] The aforementioned state estimation unit is configured to use the calculated angular velocity value calculated by the aforementioned angular velocity calculation unit to correct the current velocity vector prediction value of the object calculated based on the past velocity vector of the object, which is estimated by the aforementioned state estimation unit to be one of the states of the object.
[0208] [Project 10]
[0209] According to the presumed apparatus described in any one of items 1 to 9, among which,
[0210] The system includes a collision determination unit (32) configured to determine the collision between the moving body and the object using the calculated angular velocity value calculated by the angular velocity calculation unit.
Claims
1. A estimation device, which is mounted on a moving body and estimates the state of surrounding objects, wherein, The estimation device includes: The sensor unit (11) is configured to send sensor waves to the surroundings and detect the position and relative velocity of multiple reflection points on the object based on the reflected waves of the sensor waves. The velocity vector calculation unit (23) is configured to calculate the velocity vector of any reference point of the object; The reflection point distance calculation unit (21) is configured to calculate the distance from the reference point to a straight line with the slope of the reflection point orientation detected by the sensor unit as the reflection point distance; The rotational relative velocity calculation unit (25) is configured to calculate the rotational relative velocity of the reflection point, which is a value obtained by subtracting the relative velocity at the orientation of the reflection point from the relative velocity of the reflection point detected by the sensor unit and the relative velocity at the orientation of the reflection point based on the velocity vector of the reference point calculated by the velocity vector calculation unit. as well as The angular velocity calculation unit (27) is configured to calculate the angular velocity of the object based on the reflection point distance and the rotational relative velocity of at least one of the reflection points calculated by the reflection point distance calculation unit and the rotational relative velocity calculation unit.
2. The estimation device according to claim 1, wherein, The angular velocity calculation unit is configured to calculate the angular velocity based on the relative rotational velocity of at least one of the reflection points relative to the distance from the reflection point.
3. The estimation device according to claim 2, wherein, The angular velocity calculation unit includes a regression processing unit (27A), which is configured to perform linear regression based on the distribution of the reflection points in a coordinate space using the distance to the reflection point and the relative rotational velocity as parameters, and calculate the slope in the coordinate space. The angular velocity calculation unit is configured to calculate the angular velocity based on the slope calculated by the regression processing unit.
4. The estimating device according to any one of claims 1 to 3, wherein, The angular velocity calculation unit performs an exclusion determination on the plurality of reflection points detected by the sensor unit based on at least one of the following determination conditions a) to c): a) The received power of the reflected wave is below a threshold. b) The velocity difference from the reference point is above the threshold value. c) The reflected wave is a plurality of waves. The estimation device is configured to exclude the reflection point that is determined to meet at least one of the determination conditions a) to c) in the exclusion determination from the calculation object of the angular velocity.
5. The estimating device according to any one of claims 1 to 3, wherein, The angular velocity calculation unit is configured to either not calculate the angular velocity or not output the calculation result of the angular velocity if any one of the following conditions is met: the distance from the object is below a predetermined value; the number of reflection points is below a predetermined value; or the absolute value of the object's acceleration is above a threshold value.
6. The estimating device according to any one of claims 1 to 3, wherein, The device includes a state estimation unit (30) configured to estimate the state of the object corresponding to the plurality of reflection points. The state estimation unit is configured to reflect the calculated value of the angular velocity calculated by the angular velocity calculation unit, i.e., the calculated value of the angular velocity, in the state estimation result of the object.
7. The estimation device according to claim 6, wherein, The state estimation unit is configured to estimate a state including the angular velocity using a method different from that of the angular velocity calculation unit, and is configured to update the angular velocity estimation value by mixing the estimated angular velocity value with the calculated angular velocity value. The estimated velocity value is the estimated result of the angular velocity included in the state estimation result of the object, and the calculated angular velocity value is calculated by the angular velocity calculation unit.
8. The estimation device according to claim 6, wherein, The state estimation unit is configured to estimate the state including the angular velocity using a method different from that of the angular velocity calculation unit, and is configured to increase the filter gain of the filter used by the state estimation unit to estimate the state of the object when the absolute value of the angular velocity calculated by the angular velocity calculation unit is above a predetermined value.
9. The estimation device according to claim 6, wherein, The state estimation unit is configured to use the angular velocity calculated by the angular velocity calculation unit to correct the current velocity vector prediction value of the object, which is calculated based on the past velocity vector of the object, which is estimated by the state estimation unit as one of the states of the object.
10. The estimating device according to any one of claims 1 to 3, wherein, The system includes a collision determination unit (32) configured to determine the collision between the moving body and the object using the calculated angular velocity value calculated by the angular velocity calculation unit.