CN111737384A - Track management system and method based on cache space - Google Patents
Track management system and method based on cache space Download PDFInfo
- Publication number
- CN111737384A CN111737384A CN202010472466.7A CN202010472466A CN111737384A CN 111737384 A CN111737384 A CN 111737384A CN 202010472466 A CN202010472466 A CN 202010472466A CN 111737384 A CN111737384 A CN 111737384A
- Authority
- CN
- China
- Prior art keywords
- data detected
- millimeter wave
- wave radar
- data
- valid
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title abstract description 26
- 230000008859 change Effects 0.000 claims abstract description 4
- 238000007726 management method Methods 0.000 claims description 24
- 238000004364 calculation method Methods 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 5
- 238000001914 filtration Methods 0.000 claims description 4
- 230000009191 jumping Effects 0.000 abstract description 6
- 230000008447 perception Effects 0.000 abstract description 6
- 230000000007 visual effect Effects 0.000 description 14
- 238000010586 diagram Methods 0.000 description 8
- 238000001514 detection method Methods 0.000 description 5
- 230000004927 fusion Effects 0.000 description 4
- 230000007246 mechanism Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 230000006378 damage Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000003139 buffering effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000012938 design process Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000011895 specific detection Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2455—Query execution
- G06F16/24552—Database cache management
-
- G06Q50/40—
Abstract
The invention relates to a track management system and method based on a cache space, wherein the track management system comprises a millimeter wave radar, a camera and an ECU (electronic control unit), a buffer cache interval is opened in the ECU, data detected by the millimeter wave radar comprises a target distance, a target azimuth angle and a target distance change rate, data detected by the camera comprises target vehicle information, lane line information and traffic marking information, the buffer cache interval comprises M storage bits, a state bit and failure times, and M is a natural number greater than 2. The invention provides a track management system and method based on a cache space, which can effectively and reliably generate and destroy tracks, effectively reduce perception errors while solving the problem of target ID jumping, respectively provide two different track management modes aiming at the conditions of normal operation of a camera and failure of the camera, and improve the quality of the generated tracks.
Description
Technical Field
The invention relates to the field of automobiles, in particular to a track management system and method based on a cache space.
Background
Nowadays, no matter unmanned driving or auxiliary driving systems, the unmanned driving or auxiliary driving systems become new trends of the current technology development, and the core difficulty of the technology is the design of a perception system. In the face of complex driving road conditions, the requirements on a sensing system of a vehicle are extremely strict, and a sensing scheme of multi-sensor data fusion becomes a unique way for overcoming the technical difficulty. At present, a relatively common sensor architecture is "a camera, a millimeter wave radar", although this kind of technical scheme has been generally adopted, still has certain limitation: first, the camera may fail at night due to insufficient lighting; secondly, although the millimeter wave radar can normally operate at night, a large number of false alarm points, false tracks, problems of tracking target ID jumping and the like are mixed in feedback data. All the above problems bring an unsolved perception error to the system.
The method for managing the flight path can alleviate the problems to a certain extent, and the main principle is as follows: the target points fed back by the sensors at a new moment are matched by recording the tracks of the known vehicles around the self-vehicle, so that the perception error is reduced, and the problem of tracking the target ID jumping is solved. The categories of track management algorithms mainly include GNN (nearest neighbor matching), JPDA (joint probability density association), MHT (multi hypothesis tracking), and the like. The main work is also focused on the matching of the flight path and the updating of the flight path in the process, and the work specially aiming at the aspects of the generation and the destruction of the flight path is still blank. In the current track management algorithm, the generation and the destruction of the track are not well defined, and the track management algorithm can only acquire data fed back by the millimeter wave radar when the vehicle is in a driving state at night, so that the definition of the track generation is more difficult.
Disclosure of Invention
In view of this, the invention provides a track management system and method based on a cache space, which effectively and reliably generate and destroy tracks, effectively reduce perception errors while solving target ID jump, and respectively provide two different track management modes for the conditions of "normal operation of a camera" and "failure of the camera" to improve the quality of the generated tracks.
The purpose of the invention is realized by the following technical scheme:
the utility model provides a track management system based on buffer space, includes millimeter wave radar, camera and ECU, open up buffer interval in the ECU, the data that millimeter wave radar detected include target distance, target azimuth, target distance rate of change, and the data that the camera detected include target vehicle information, lane marking information and traffic sign information, buffer interval includes M storage bit, status bit and the number of times of failure, and M is for being greater than 2 natural number. And M is 3.
A track management method, based on the track management system based on the cache space of claim 1 or 2, specifically comprising:
step one, setting a specific numerical value of a failure threshold;
reading data detected by the millimeter wave radar;
judging whether the data detected by the millimeter wave radar is valid or not, and distinguishing the valid data from invalid data;
if the data is valid data, storing the valid data into a storage bit which is not written with the data in the buffer cache interval to form current track information; if the data is invalid data, discarding the data, and increasing the number of failures by 1;
judging the relationship between the failure times and the failure threshold value; when the failure times are larger than the failure threshold value, destroying the current track information, and returning to the step two to carry out the next group of circulation; and when the failure times are less than or equal to the failure threshold value, keeping the current track information, and returning to the step two to carry out the next group of circulation.
In the third step, whether the data detected by the millimeter wave radar is valid is judged according to the states of the state bits, wherein the state bits comprise four states of 0, 1, 2 and 3.
Preferably, when the status bit is 0, there are two cases according to whether the data detected by the camera is valid: 1. when data detected by the camera fails, the validity of the data detected by the millimeter wave radar is judged through the ROI, when the data detected by the millimeter wave radar falls into the ROI, the data detected by the millimeter wave radar is considered to be valid, otherwise, the data detected by the millimeter wave radar is considered to be invalid and discarded; 2. when the data detected by the camera is valid, the data detected by the millimeter wave radar is subjected to validity judgment through the ROI (region of interest), when the data detected by the millimeter wave radar falls into the ROI, the data detected by the millimeter wave radar is considered to be valid and is matched with the data detected by the camera, if the matching is successful, the data detected by the millimeter wave radar is valid, otherwise, the data detected by the millimeter wave radar is invalid and discarded; if the data detected by the millimeter wave radar is effective, filling the data detected by the millimeter wave radar into a first storage position of a buffer cache interval, modifying the state bit to be 1, and resetting the failure times to be 0; if the data detected by the millimeter wave radar is judged to be invalid, the data detected by the millimeter wave radar is not filled into the buffer cache interval, the state bit value is kept unchanged, and the failure times are increased by 1; and when the failure times are larger than the failure threshold value, resetting all the storage bits, the state bits and the failure times to be empty.
Preferably, when the status bit is 1, there are two cases according to whether the data detected by the camera is valid: 1. when data detected by the camera fails, the validity of the data detected by the millimeter wave radar is judged through the ROI, when the data detected by the millimeter wave radar falls into the ROI, the data detected by the millimeter wave radar is considered to be valid, otherwise, the data detected by the millimeter wave radar is considered to be invalid and discarded; 2. when the data detected by the camera is valid, the data detected by the millimeter wave radar is subjected to validity judgment through the ROI (region of interest), when the data detected by the millimeter wave radar falls into the ROI, the data detected by the millimeter wave radar is considered to be valid and is matched with the data detected by the camera, if the matching is successful, the data detected by the millimeter wave radar is valid, otherwise, the data detected by the millimeter wave radar is invalid and discarded; if the data detected by the millimeter wave radar is effective, filling the data detected by the millimeter wave radar into a second storage position of the buffer cache interval, modifying the state bit to be 2, and resetting the failure times to be 0; if the data detected by the millimeter wave radar is judged to be invalid, the data detected by the millimeter wave radar is not filled into the buffer cache interval, the state bit value is kept unchanged, and the failure times are increased by 1; and when the failure times are larger than the failure threshold value, resetting all the storage bits, the state bits and the failure times to be empty.
Preferably, when the status bit is 2, there are two cases according to whether the data detected by the camera is valid: 1. when data detected by the camera fails, the validity of the data detected by the millimeter wave radar is judged through the ROI, when the data detected by the millimeter wave radar falls in the ROI, oval wave gate calculation is carried out on the data detected by the millimeter wave radar, if the data detected by the millimeter wave radar falls in the ROI, the data detected by the millimeter wave radar is considered to be valid, otherwise, the data detected by the millimeter wave radar is considered to be invalid and discarded; 2. when the data detected by the camera is valid, the data detected by the millimeter wave radar is subjected to validity judgment through the ROI (region of interest), when the data detected by the millimeter wave radar falls into the ROI, the data detected by the millimeter wave radar is considered to be valid and is matched with the data detected by the camera, if the matching is successful, the data detected by the millimeter wave radar is valid, otherwise, the data detected by the millimeter wave radar is invalid and discarded; if the data detected by the millimeter wave radar is effective, filling the data detected by the millimeter wave radar into a third storage position of the buffer cache interval, modifying the state bit to be 3, and resetting the failure times to be 0; if the data detected by the millimeter wave radar is judged to be invalid, the data detected by the millimeter wave radar is not filled into the buffer cache interval, the state bit value is kept unchanged, and the failure times are increased by 1; and when the failure times are larger than the failure threshold value, resetting all the storage bits, the state bits and the failure times to be empty.
Preferably, when the status bit is 3, there are two cases according to the fact that the data detected by the camera at this time is valid: 1. when data detected by the camera fails, the validity of the data detected by the millimeter wave radar is judged through the ROI, when the data detected by the millimeter wave radar falls in the ROI, oval wave gate calculation is carried out on the data detected by the millimeter wave radar, if the data detected by the millimeter wave radar falls in the ROI, the data detected by the millimeter wave radar is considered to be valid, otherwise, the data detected by the millimeter wave radar is considered to be invalid and discarded; 2. when the data detected by the camera is valid, the data detected by the millimeter wave radar is subjected to validity judgment through the ROI (region of interest), when the data detected by the millimeter wave radar falls into the ROI, the data detected by the millimeter wave radar is considered to be valid and is matched with the data detected by the camera, if the matching is successful, the data detected by the millimeter wave radar is valid, otherwise, the data detected by the millimeter wave radar is invalid and discarded; if the data detected by the millimeter wave radar is effective, covering the data on the second storage position of the buffer cache interval to the first storage position, covering the data on the third storage position to the second storage position, filling the data detected by the millimeter wave radar into the third storage position of the buffer cache interval, keeping the state position to be 3, and resetting the failure times to be 0; if the data detected by the millimeter wave radar is judged to be invalid, the data detected by the millimeter wave radar is not filled into the buffer cache interval, the state bit value is kept unchanged, and the failure times are increased by 1; and when the failure times are larger than the failure threshold value, resetting all the storage bits, the state bits and the failure times to be empty.
Preferably, the calculation of the elliptic wave gate is as follows:
a, reading data of a first storage bit and a second storage bit in a buffer interval, and linearly predicting a prediction value Z of the second storage bitpredictWherein the length of the prediction time is determined by the number of times n of failure between the second memory bit and the third memory bitfailureAnd is increased;
step b, calculating observation data Z detected by the millimeter wave radar to be written into the third storage bit of the buffer cache intervalobserveAnd the predicted value ZpredictThe specific formula of the residual d between the two is as follows: d = | Zobserve-ZpredictAnd finally, calculating the mahalanobis distance D between the observed value and the predicted value, wherein the specific formula is as follows:
S=H*Ppre*H’+R
D=d*inv(S)*d
in the above formula, H is an observation matrix, and R is observation noise.
Step c, comparing the Mahalanobis distance D with a set threshold, if the Mahalanobis distance D is smaller than the set threshold, determining that the observation data falls in a 'wave gate', writing the observation data into a third storage position in a buffer cache interval, setting the state bit to be 3, and resetting the failure times to be 0; at the same time, for the predicted value ZpredictAnd number of observationsAccording to ZobservePerforming Kalman filtering tracking to obtain an estimated value as an initial value entering a continuous tracking state;
d, if the Mahalanobis distance D is larger than a set threshold value, the observation data is considered to fall outside a 'wave gate', the observation data is judged to be invalid, the failure times are increased by 1, and the observation data is discarded; meanwhile, comparing the failure times with a failure threshold, if the failure times is greater than the failure threshold, resetting and emptying all the storage bits, the state bits and the failure times, and entering a next group of observation data judgment circulation; and if the failure times are less than or equal to the failure threshold value, directly entering a loop of next group of observation data judgment.
Compared with the prior art, the invention has the beneficial effects that:
based on the current situation, the invention provides a track management method based on a cache space, which effectively reduces perception errors while solving the problem of target ID jumping and is an effective and reliable track generation and destruction method. Moreover, two different track management modes are respectively provided for the conditions of normal running of the camera and failure of the camera, so that the generated track quality is improved.
Drawings
FIG. 1 is a block diagram of a track management system based on cache space according to an embodiment of the present invention.
Fig. 2 is a schematic diagram of detection data of the millimeter wave radar and the camera in an embodiment of the present invention.
FIG. 3 is a flow chart illustrating the status differentiation between buffer intervals according to an embodiment of the present invention.
FIG. 4 is a flowchart illustrating an initialization process of a buffer interval according to an embodiment of the invention.
FIG. 5 is a flowchart illustrating the A/B determination method under the condition of failure of visual information according to an embodiment of the present invention.
FIG. 6 is a flowchart illustrating the A/B/C determination method under the condition of valid visual information according to an embodiment of the present invention.
FIG. 7 is a diagram of a data point failure processing mechanism in accordance with an embodiment of the present invention.
FIG. 8 is a diagram of a discrimination failure processing mechanism at C when visual information fails according to an embodiment of the present invention.
FIG. 9 is a flowchart illustrating a continuous tracking procedure of a buffer interval according to an embodiment of the present invention.
Fig. 10 is a flowchart illustrating a data point validity determination at point D according to an embodiment of the present invention.
FIG. 11 is a diagram illustrating output data type selection according to an embodiment of the present invention.
FIG. 12 is a track chart of first embodiment of the present invention.
FIG. 13 is a flight path diagram according to the second embodiment of the present invention.
FIG. 14 is a flight path diagram of embodiment c of the present invention.
FIG. 15 is a track chart of the embodiment (iv) of the present invention.
Detailed Description
To facilitate understanding of those skilled in the art, the present invention will be described in further detail below with reference to specific embodiments and the accompanying drawings.
Referring to fig. 1-15, an embodiment of the invention includes:
the invention relates to a target level fusion sensing technology based on a sensor architecture of a millimeter wave radar and a camera. The visual data is taken charge of by a camera, the original video is processed into a target through an image algorithm and then transmitted to an ECU (electronic processing unit) through CAN communication, and the radar data is transmitted to the ECU through a CAN bus by a millimeter wave radar. While the sensor data fusion algorithm is mainly performed in the ECU (a specific sensor architecture is shown in fig. 1), this architecture reduces the sensor redundancy cost overhead but at the same time increases the risk of sensing system failure. Aiming at the point, the method can still remove false alarm information only by depending on a single millimeter wave radar under the condition that the visual information is in a failure state, and the track confidence is maintained.
In the target detection scheme (a specific detection scheme is shown in fig. 2), the detection distance of the millimeter wave radar is greater than that of the camera, and the detection angle of the camera is greater in terms of the detection angle. The millimeter wave radar feedback information includes a target distance (range), a target azimuth (angle), and a target distance change rate (range rate). And the camera feedback data includes target vehicle information (position information/speed information), lane line information, and traffic marking information.
The most important core point of the invention is that a buffer interval is opened up, 64 same buffer intervals exist in total, and 3 data point information (the data point information comes from different moments under the same ID) can be stored in the interval. When the data point information is not full in the buffer interval, defining the flight path to be in an initialization state; and when the data point information is full of the buffer interval, the track is considered to be in a mature state, and then continuous tracking is started.
The specific flow diagram is as shown in the attached figure 3:
in the buffer interval state distinguishing flow chart, the buffer is divided into two states according to whether the buffer is filled or not, wherein the 'initialization state' indicates that a new data point is detected by a track algorithm and cannot be matched with any existing track, but at the moment, whether the data point is a real value fed back by a new target or is generated due to noise interference cannot be distinguished. And after entering a continuous tracking state, the algorithm system considers that the track is mature, and then enters a data fusion module.
The following describes "initialization state" and "tracking state", respectively.
An initialization state:
as shown in fig. 4, after the data point enters the initialization state, the occupation status of the buffer interval is first determined, which includes three situations, namely, occupation of 0 data point, occupation of 1 data point, and occupation of 2 data points. The following are explained separately:
a) the buffer interval is free from occupation: when the buffer status bit is determined to be 0 (if the status bit is not 0, it is determined whether the status bit is 1 next), the validity determination of the data point is entered, and the determination is performed by using the method a, which is shown in fig. 5.
As shown in fig. 6, the data point validity determination is divided into two determination methods: (1) when the visual information is invalid, the validity is judged directly through a region of interest (ROI), and when the data point falls into the ROI, the data point is considered as a valid point, otherwise, the valid point is discarded; (2) when the visual information normally runs, adding the image data on the basis of the method (1), when the data point passes through the ROI and the radar data and the image data are successfully matched, considering the point to be effective, and otherwise, abandoning the point.
After the point is authenticated as the valid point, the point is filled to the first position of the cache buffer, the state of the cache buffer is set to 1, which indicates that one position of the cache buffer is filled, and meanwhile, the number of the invalid points is reset to 0. Otherwise, the data point is discarded and the decision for the next data point is entered.
b) Buffer interval has 1 occupation: when the buffer status bit is judged to be 1, entering data point validity judgment, and judging by adopting a method B, wherein the specific method is the same as that in FIGS. 3 and 4, and is not repeated here.
And when the data point is judged to be the valid point, filling the data point to the second position of the buffer, setting the state of the buffer to be 2, indicating that the two positions of the buffer are filled, and resetting the number of the invalid points to be 0. Otherwise, discarding the data point, and entering a buffer failure judgment module, wherein the specific judgment mode is as follows:
firstly, recording the failure times between the failure point and the last effective point (if the point is the first failure point, the failure times is set to 1, and if the point is not the first failure point, the failure times are increased by 1). And then, performing cache buffer invalidation judgment, and resetting the whole cache buffer when the invalidation frequency is greater than a certain threshold (the threshold is determined to be 3 here). Wherein resetting the content comprises:
(1) caching the data points stored by the buffer;
(2) buffering the buffer status bit;
(3) buffer invalidation times are cached.
And finally entering the next data point cycle.
c) The buffer interval has 2 occupancies: when the buffer status bit of the cache is judged to be 2, entering data point validity judgment, and adopting a C method for judgment, wherein the specific method is shown in FIG. 6.
First, it is determined whether the current data point exists in the ROI, and if the current data point exists outside the ROI, the current data point is discarded, and the data point failure handling mechanism shown in fig. 5 is entered. If the data point is present within the ROI, the next step is entered, which is classified into two categories:
(1) when the visual information is in the valid state, entering a judgment mechanism of fig. 4, which is not described herein again;
(2) when the visual information is in a failure state, the validity judgment of the data point is performed again through the elliptic wave gate calculation shown in fig. 6, and the specific steps are as follows:
1. linearly predicting the position of the third point according to the data of the 1 st point and the 2 nd point of the bufferZ predict Wherein the predicted time duration should be based on the number of failure points spaced between 2-3 data points (n failure) And increases, namely:
2. calculating observation data Z detected by the millimeter wave radar to be written into the third storage bit of the buffer cache intervalobserveAnd the predicted value ZpredictThe specific formula of the residual d between the two is as follows: d = | Zobserve-ZpredictAnd finally, calculating the mahalanobis distance (D) between the observed value and the predicted value, wherein the specific formula is as follows:
S=H*Ppre*H’+R
D=d*inv(S)*d
in the above formula, H is an observation matrix, and R is observation noise.
3. And comparing the Mahalanobis distance with a set threshold, if the distance is smaller than the set threshold, determining that the data point falls in a 'wave gate', filling the data point to a third position of the cache buffer, setting the state of the cache buffer to be 3 (indicating that the flight path responsible for the cache buffer enters a 'mature' state), and resetting the number of the dead points to be 0. At the same time, for the predicted valueZ predict And observed valueZ observe And performing Kalman (or other algorithm) filtering tracking to obtain an estimated value as an initial value for entering a continuous tracking state.
4. On the contrary, if the distance is greater than the set threshold, the data point is considered to fall outside the 'wave gate', the data point is determined as an invalid point, the number of invalid data points flag is increased by 1, and the data point is discarded. Meanwhile, the number of the failure points is compared with the threshold value of the number of the failure points, if the number of the failure points is larger than the threshold value of the number of the failure points, the cache buffer is reset (the reset content is not described any more), and the next data point cycle is entered, and if the number of the failure points is smaller than the threshold value of the number of the failure points, the next data point cycle.
Continuous tracking state:
when the buffer status is set to 3, the method enters a continuous tracking status, and the specific process is as shown in fig. 9:
(1) according to state transition matrixF(the matrix type can be various, and nonlinear iterative updating equation can also be adopted), and the sampling period of the sensorTPredicting the kalman estimated value of the last timeX predict Wherein the prediction step size isT * n failure 。
(2) Judging the validity of the data point at the moment, and judging by adopting a D method, wherein the judging process is as shown in the attached figure 10:
firstly, judging whether a data point is effective or not through the ROI, if the data point is not in the ROI, directly discarding the data point, increasing the number of failure points by 1 (if the number of failure points exceeds a threshold value, resetting the buffer), and if not, entering elliptical wave gate calculation and judging.
(3) If the data point falls on the outer side of the 'wave gate', the number of the invalid data points of the buffer is increased by 1, then the invalid data points are compared with the threshold value of the number of the invalid points, if the number of the invalid data points is larger than the threshold value, the buffer is reset, otherwise, the next data point circulation is directly carried out. At this time, the data point is not discarded, but is refilled to the first position of the new buffer, and a new initialization operation is started.
(4) If the data point falls within the "wave gate", then the predicted value is usedZ predict Effective observed value at current momentZ observe Performing kalman tracking filtering to obtain an estimated valueZ estimate Wherein:
and filling the estimated value to the third position of the buffer, respectively filling the effective data at the previous moment and the effective data at the previous moment to the second position of the buffer and the first position of the buffer according to the queue displacement (discarding the data point originally filled at the first position), resetting the number of the failure points to 0, and then entering the next data point cycle.
(5) Based on the validity judgment of the current data point, the output data type is determined: when the current data point is judged to be valid, outputting a kalman estimated valueZ estimate (ii) a When the current data point is judged to be invalid, outputting a predicted valueZ predict As shown in fig. 11:
compared with the prior art, the invention has the following advantages:
1. the invention discloses a method for managing a track, which aims at designing track management strategies under two driving states of 'effective visual information' and 'ineffective visual information'. When the visual information is normal and effective, the radar information is acquired, and then the effectiveness of the visual information evidence data is reused, so that the appearance of a false alarm target is reduced or basically stopped. And when the visual information fails, a judgment mode of 'gate' is adopted to check whether the continuous radar points generate obvious jumping, and if the continuous radar points generate obvious jumping, the false alarm points are considered to be generated.
2. Possible causes of the jump in effect 1 include the following two: a) the radar data point at the current moment jumps; b) in the buffer, false alarm points appear in the filling bits 1 and 2, and further the 'wave gate' cannot be matched. For the second case, the result is that the 3 rd point in the buffer cannot match the 1 st and 2 nd points (i.e. cannot fall inside the wave gate) all the time. Therefore, in the algorithm design process of the invention patent, it is specified that the whole buffer is reset when the 3 rd point in the buffer after a plurality of continuous points can not be matched, so as to avoid the second situation.
3. After the buffer enters the continuous tracking state, if the newly appeared radar data is in the ROI but cannot be matched with the track, the reasons for such phenomenon are as follows: a) the radar data are false alarm points, generally only a few data points jump and then return to normal; b) the old target disappears while the new target appears.
Aiming at the first phenomenon, the invention can continuously track the old track in a prediction mode when the original cache buffer fails to be matched successfully, and the data points which cannot be matched enter a new cache buffer.
In the second kind of phenomenon, the invention reserves the old track for several steps in a prediction mode, but if the threshold value is exceeded, the buffer is reset (namely, the old track is destroyed), and the new track can be continuously tracked after the new track is initialized.
The concrete effects are shown in figures 12-15
In FIG. 12, after several frames are required by the algorithm, the track is generated.
In the attached figure 13, observation data is lost at position two, but the continuous tracking is realized through a prediction mode by the patented algorithm.
In fig. 14, when noise disturbance occurs in the observed value, the algorithm can effectively suppress the noise disturbance.
In fig. 15, when the old track disappears, the track algorithm resets the buffer after continuing to track a plurality of points.
In the description of the present invention, it is to be understood that terms such as "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", and the like, which indicate orientations or positional relationships, are used based on the orientations or positional relationships shown in the drawings only for the convenience of describing the present invention and for the simplicity of description, and do not indicate or imply that the referenced devices or elements must have a particular orientation, be constructed in a particular orientation, and be operated, and thus, are not to be construed as limiting the present invention.
Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the present invention, "a plurality" means two or more unless specifically defined otherwise.
While the invention has been described in conjunction with the specific embodiments set forth above, it is evident that many alternatives, modifications, and variations will be apparent to those skilled in the art in light of the foregoing description. Accordingly, it is intended to embrace all such alternatives, modifications, and variations that fall within the spirit and scope of the appended claims.
Claims (9)
1. The utility model provides a track management system based on buffer space, its characterized in that, includes millimeter wave radar, camera and ECU, open up in the ECU and have buffer cache interval, the data that millimeter wave radar detected include target distance, target azimuth, target distance rate of change, and the data that the camera detected include target vehicle information, lane line information and traffic sign information, buffer cache interval includes M storage bit, status bit and the number of times of failure, and M is for being greater than 2 natural number.
2. The cache space-based track management system of claim 1, wherein M is 3.
3. A track management method, according to the track management system based on the cache space of claim 1 or 2, specifically comprising:
step one, setting a specific numerical value of a failure threshold;
reading data detected by the millimeter wave radar;
judging whether the data detected by the millimeter wave radar is valid or not, and distinguishing the valid data from invalid data;
if the data is valid data, storing the valid data into a storage bit which is not written with the data in the buffer cache interval to form current track information; if the data is invalid data, discarding the data, and increasing the number of failures by 1;
judging the relationship between the failure times and the failure threshold value; when the failure times are larger than the failure threshold value, destroying the current track information, and returning to the step two to carry out the next group of circulation; and when the failure times are less than or equal to the failure threshold value, keeping the current track information, and returning to the step two to carry out the next group of circulation.
4. The track management method according to claim 3, wherein in the third step, whether the data detected by the millimeter wave radar is valid is determined according to states of state bits, and the state bits include four states of 0, 1, 2, and 3.
5. The track management method according to claim 4, wherein when the status bit is 0, there are two cases according to whether the data detected by the camera is valid: 1. when data detected by the camera fails, the validity of the data detected by the millimeter wave radar is judged through the ROI, when the data detected by the millimeter wave radar falls into the ROI, the data detected by the millimeter wave radar is considered to be valid, otherwise, the data detected by the millimeter wave radar is considered to be invalid and discarded; 2. when the data detected by the camera is valid, the data detected by the millimeter wave radar is subjected to validity judgment through the ROI (region of interest), when the data detected by the millimeter wave radar falls into the ROI, the data detected by the millimeter wave radar is considered to be valid and is matched with the data detected by the camera, if the matching is successful, the data detected by the millimeter wave radar is valid, otherwise, the data detected by the millimeter wave radar is invalid and discarded; if the data detected by the millimeter wave radar is effective, filling the data detected by the millimeter wave radar into a first storage position of a buffer cache interval, modifying the state bit to be 1, and resetting the failure times to be 0; if the data detected by the millimeter wave radar is judged to be invalid, the data detected by the millimeter wave radar is not filled into the buffer cache interval, the state bit value is kept unchanged, and the failure times are increased by 1; and when the failure times are larger than the failure threshold value, resetting all the storage bits, the state bits and the failure times to be empty.
6. The track management method according to claim 4, wherein when the status bit is 1, there are two cases according to whether the data detected by the camera is valid: 1. when data detected by the camera fails, the validity of the data detected by the millimeter wave radar is judged through the ROI, when the data detected by the millimeter wave radar falls into the ROI, the data detected by the millimeter wave radar is considered to be valid, otherwise, the data detected by the millimeter wave radar is considered to be invalid and discarded; 2. when the data detected by the camera is valid, the data detected by the millimeter wave radar is subjected to validity judgment through the ROI (region of interest), when the data detected by the millimeter wave radar falls into the ROI, the data detected by the millimeter wave radar is considered to be valid and is matched with the data detected by the camera, if the matching is successful, the data detected by the millimeter wave radar is valid, otherwise, the data detected by the millimeter wave radar is invalid and discarded; if the data detected by the millimeter wave radar is effective, filling the data detected by the millimeter wave radar into a second storage position of the buffer cache interval, modifying the state bit to be 2, and resetting the failure times to be 0; if the data detected by the millimeter wave radar is judged to be invalid, the data detected by the millimeter wave radar is not filled into the buffer cache interval, the state bit value is kept unchanged, and the failure times are increased by 1; and when the failure times are larger than the failure threshold value, resetting all the storage bits, the state bits and the failure times to be empty.
7. The track management method according to claim 4, wherein when the status bit is 2, there are two cases according to whether the data detected by the camera is valid: 1. when data detected by the camera fails, the validity of the data detected by the millimeter wave radar is judged through the ROI, when the data detected by the millimeter wave radar falls in the ROI, oval wave gate calculation is carried out on the data detected by the millimeter wave radar, if the data detected by the millimeter wave radar falls in the ROI, the data detected by the millimeter wave radar is considered to be valid, otherwise, the data detected by the millimeter wave radar is considered to be invalid and discarded; 2. when the data detected by the camera is valid, the data detected by the millimeter wave radar is subjected to validity judgment through the ROI (region of interest), when the data detected by the millimeter wave radar falls into the ROI, the data detected by the millimeter wave radar is considered to be valid and is matched with the data detected by the camera, if the matching is successful, the data detected by the millimeter wave radar is valid, otherwise, the data detected by the millimeter wave radar is invalid and discarded; if the data detected by the millimeter wave radar is effective, filling the data detected by the millimeter wave radar into a third storage position of the buffer cache interval, modifying the state bit to be 3, and resetting the failure times to be 0; if the data detected by the millimeter wave radar is judged to be invalid, the data detected by the millimeter wave radar is not filled into the buffer cache interval, the state bit value is kept unchanged, and the failure times are increased by 1; and when the failure times are larger than the failure threshold value, resetting all the storage bits, the state bits and the failure times to be empty.
8. The track management method according to claim 4, wherein when the status bit is 3, there are two cases according to whether the data detected by the camera is valid: 1. when data detected by the camera fails, the validity of the data detected by the millimeter wave radar is judged through the ROI, when the data detected by the millimeter wave radar falls in the ROI, oval wave gate calculation is carried out on the data detected by the millimeter wave radar, if the data detected by the millimeter wave radar falls in the ROI, the data detected by the millimeter wave radar is considered to be valid, otherwise, the data detected by the millimeter wave radar is considered to be invalid and discarded; 2. when the data detected by the camera is valid, the data detected by the millimeter wave radar is subjected to validity judgment through the ROI (region of interest), when the data detected by the millimeter wave radar falls into the ROI, the data detected by the millimeter wave radar is considered to be valid and is matched with the data detected by the camera, if the matching is successful, the data detected by the millimeter wave radar is valid, otherwise, the data detected by the millimeter wave radar is invalid and discarded; if the data detected by the millimeter wave radar is effective, covering the data on the second storage position of the buffer cache interval to the first storage position, covering the data on the third storage position to the second storage position, filling the data detected by the millimeter wave radar into the third storage position of the buffer cache interval, keeping the state position to be 3, and resetting the failure times to be 0; if the data detected by the millimeter wave radar is judged to be invalid, the data detected by the millimeter wave radar is not filled into the buffer cache interval, the state bit value is kept unchanged, and the failure times are increased by 1; and when the failure times are larger than the failure threshold value, resetting all the storage bits, the state bits and the failure times to be empty.
9. The track management method according to claim 7 or 8, wherein the elliptical wave gate calculation is specifically as follows:
a, reading data of a first storage bit and a second storage bit in a buffer interval, and linearly predicting a prediction value Z of the second storage bitpredictWherein the length of the prediction time is determined by the number of times n of failure between the second memory bit and the third memory bitfailureAnd is increased;
step b, calculating observation data Z detected by the millimeter wave radar to be written into the third storage bit of the buffer cache intervalobserveAnd the predicted value ZpredictThe specific formula of the residual d between the two is as follows: d = | Zobserve-ZpredictAnd finally, calculating the mahalanobis distance D between the observed value and the predicted value, wherein the specific formula is as follows:
S=H*Ppre*H’+R
D=d*inv(S)*d
in the above formula, H is an observation matrix, and R is observation noise;
step c, comparing the Mahalanobis distance D with a set threshold, if the Mahalanobis distance D is smaller than the set threshold, determining that the observation data falls in a 'wave gate', writing the observation data into a third storage position in a buffer cache interval, setting the state bit to be 3, and resetting the failure times to be 0; at the same time, for the predicted value ZpredictAnd observation data ZobservePerforming Kalman filtering tracking to obtain an estimated value as a forward valueEntering an initial value of a 'continuous tracking state';
d, if the Mahalanobis distance D is larger than a set threshold value, the observation data is considered to fall outside a 'wave gate', the observation data is judged to be invalid, the failure times are increased by 1, and the observation data is discarded; meanwhile, comparing the failure times with a failure threshold, if the failure times is greater than the failure threshold, resetting and emptying all the storage bits, the state bits and the failure times, and entering a next group of observation data judgment circulation; and if the failure times are less than or equal to the failure threshold value, directly entering a loop of next group of observation data judgment.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010472466.7A CN111737384A (en) | 2020-05-29 | 2020-05-29 | Track management system and method based on cache space |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010472466.7A CN111737384A (en) | 2020-05-29 | 2020-05-29 | Track management system and method based on cache space |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111737384A true CN111737384A (en) | 2020-10-02 |
Family
ID=72647967
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010472466.7A Pending CN111737384A (en) | 2020-05-29 | 2020-05-29 | Track management system and method based on cache space |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111737384A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113611112A (en) * | 2021-07-29 | 2021-11-05 | 中国第一汽车股份有限公司 | Target association method, device, equipment and storage medium |
CN113701758A (en) * | 2021-08-23 | 2021-11-26 | 中国北方工业有限公司 | Multi-target data association method and system based on biological search algorithm |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107798870A (en) * | 2017-10-25 | 2018-03-13 | 清华大学 | A kind of the flight path management method and system, vehicle of more vehicle target tracking |
CN110750612A (en) * | 2019-10-23 | 2020-02-04 | 福建汉特云智能科技有限公司 | Laser radar-based flight path management method and system |
CN110824467A (en) * | 2019-11-15 | 2020-02-21 | 中山大学 | Multi-target tracking data association method and system |
-
2020
- 2020-05-29 CN CN202010472466.7A patent/CN111737384A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107798870A (en) * | 2017-10-25 | 2018-03-13 | 清华大学 | A kind of the flight path management method and system, vehicle of more vehicle target tracking |
CN110750612A (en) * | 2019-10-23 | 2020-02-04 | 福建汉特云智能科技有限公司 | Laser radar-based flight path management method and system |
CN110824467A (en) * | 2019-11-15 | 2020-02-21 | 中山大学 | Multi-target tracking data association method and system |
Non-Patent Citations (1)
Title |
---|
黄文奎 等: "一种 Kalman 最近邻的智能汽车传感器多目标跟踪算法" * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113611112A (en) * | 2021-07-29 | 2021-11-05 | 中国第一汽车股份有限公司 | Target association method, device, equipment and storage medium |
CN113701758A (en) * | 2021-08-23 | 2021-11-26 | 中国北方工业有限公司 | Multi-target data association method and system based on biological search algorithm |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109863500B (en) | Event driven region of interest management | |
US20190236862A1 (en) | Self-diagnosis of faults with a secondary system in an autonomous driving system | |
CN111737384A (en) | Track management system and method based on cache space | |
CN112329552A (en) | Obstacle detection method and device based on automobile | |
US8233663B2 (en) | Method for object formation | |
CN104217590A (en) | On-board traffic density estimator | |
CN108764108A (en) | A kind of Foregut fermenters method based on Bayesian inference | |
WO2023201904A1 (en) | Abnormal vehicle traveling detection method, and electronic device and storage medium | |
CN103366569A (en) | Method and system for snapshotting traffic violation vehicle in real time | |
US11142193B2 (en) | Vehicle and method for performing inter-vehicle distance control | |
CN111505624A (en) | Environment sensing method based on machine vision and millimeter wave radar data fusion | |
US11436815B2 (en) | Method for limiting object detection area in a mobile system equipped with a rotation sensor or a position sensor with an image sensor, and apparatus for performing the same | |
CN112367475B (en) | Traffic incident detection method and system and electronic equipment | |
CN112652190B (en) | Parking space identification method and system for automatic parking | |
CN112706785B (en) | Method and device for selecting cognitive target of driving environment of automatic driving vehicle and storage medium | |
CN106585670A (en) | Video-based urban rail transit ahead train detection system and method | |
US20220121213A1 (en) | Hybrid planning method in autonomous vehicle and system thereof | |
US20200406970A1 (en) | Vehicle and method for steering avoidance control | |
KR20170070580A (en) | Ecu, autonomous vehicle including the ecu, and method of controlling lane change for the same | |
CN115222779B (en) | Vehicle cut-in detection method and device and storage medium | |
CN115848412A (en) | Perception fusion method, device, equipment and storage medium | |
US20230219568A1 (en) | Diagnostic device for other-vehicle behavior prediction and diagnostic method for other-vehicle behavior prediction | |
Fekri et al. | A Forward Collision Warning System Using Deep Reinforcement Learning | |
KR100794625B1 (en) | Integrated room mirror type automatic recording system for traffic accident | |
KR100853444B1 (en) | Method of sensing car accidents |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |