CN115630569A - Millimeter wave radar sensor modeling method applied to automatic driving - Google Patents
Millimeter wave radar sensor modeling method applied to automatic driving Download PDFInfo
- Publication number
- CN115630569A CN115630569A CN202211227758.XA CN202211227758A CN115630569A CN 115630569 A CN115630569 A CN 115630569A CN 202211227758 A CN202211227758 A CN 202211227758A CN 115630569 A CN115630569 A CN 115630569A
- Authority
- CN
- China
- Prior art keywords
- target
- millimeter wave
- model
- wave radar
- data
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/08—Probabilistic or stochastic CAD
Abstract
The invention relates to the technical field of millimeter wave radar sensors, and discloses a millimeter wave radar sensor modeling method applied to automatic driving. The method provided by the invention has the advantages that a large amount of historical data is not needed for model training, the requirement on real-time property of target identification is met, the provided model has stronger practicability and short prediction time, the parameter difference between different targets is fully understood, the identification precision of the algorithm is further improved by designing the target identification model, meanwhile, multi-target detection is constructed for multi-target comparison, the error probability is reduced, the identification precision is further improved, and the phenomenon that the low precision of target association affects the target identification is avoided.
Description
Technical Field
The invention relates to the technical field of millimeter wave radar sensors, in particular to a millimeter wave radar sensor modeling method applied to automatic driving.
Background
Automatic driving, also known as unmanned driving, computer driving or wheeled mobile robot, is a leading-edge technology that relies on computer and artificial intelligence technology to complete, safe and effective driving without artificial manipulation. In the 21 st century, the problems of congestion, safety accidents and the like faced by road traffic become more serious due to the continuous increase of automobile users. Under the support of the car networking technology and the artificial intelligence technology, the automatic driving technology can coordinate the travel route and the planning time, so that the travel efficiency is greatly improved, and the energy consumption is reduced to a certain extent. Automatic driving can also help avoiding drunk driving, potential safety hazards such as driver fatigue, reduces driver's error, promotes the security simultaneously. Autonomous driving has therefore become a focus of recent development in various countries. As an automated vehicle, an autonomous vehicle can sense its environment and navigate without human manipulation. As a feasible automatic driving environment perception hardware, the vehicle-mounted millimeter wave radar can collect obstacle point cloud data in the driving process, and further, the state of an obstacle, such as the position, the speed and the size of multiple targets, can be analyzed based on the point cloud data.
At present, a large number of target detection databases need to be established for better identifying targets in the existing radar data target identification task, and the target association precision of the existing radar data target identification task cannot meet the automatic driving or auxiliary driving scene.
Disclosure of Invention
Technical problem to be solved
Aiming at the defects of the prior art, the invention provides a millimeter wave radar sensor modeling method applied to automatic driving, which has the advantages of quick target identification, high identification precision and the like, and solves the problems that a radar data target identification task needs to establish a large number of target detection databases to better identify targets, and the target association precision of the radar data target identification task is difficult to meet the automatic driving or auxiliary driving scene.
(II) technical scheme
In order to achieve the purpose, the invention provides the following technical scheme:
a millimeter wave radar sensor modeling method applied to automatic driving comprises the following steps:
the method comprises the steps of S1, carrying out data acquisition on traffic targets in the surrounding environment by using a millimeter wave radar sensor, then carrying out data processing on the acquired data targets, wherein the data processing comprises the steps of carrying out noise data cleaning processing on the data, then constructing data sets of different traffic targets through the data, and dividing each traffic target data set into a training set, a verification set and a test set.
S2, a multi-target detection state transfer function is built through a state model according to the data targets, the state model is a model for obtaining a multi-target prediction state set at a second moment according to the multi-target state set at a first moment, the multi-target state set is a random finite set and comprises state variables of at least two targets, and the multi-target prediction state set is a random finite set and comprises state variable prediction values of at least two targets.
And S3, constructing a likelihood function of multi-target detection according to the measurement model, wherein the measurement model is a model of a multi-target measurement set obtained according to the multi-target observation set, the multi-target observation set is a random finite set and comprises observation information acquired by the vehicle-mounted millimeter wave radar aiming at least two targets, and the multi-target measurement set is a random finite set and comprises the real information distribution probability of the at least two targets.
S4, performing correlation analysis on each attribute of the radar data, finding out the attribute of which the correlation with the target category is greater than a preset threshold value, performing statistical analysis on the obtained radar data of the attribute, constructing an empirical characteristic, training a classifier by using the empirical characteristic, the obtained attribute and a training set, obtaining a target recognition model, and constructing a multi-target detection likelihood function comprising a second error probability according to a measurement model, wherein the measurement model is an empirical model based on the parameters of the vehicle-mounted millimeter wave radar, and the second error probability is a statistical modeling of errors between observation information and real information acquired by the vehicle-mounted millimeter wave radar.
And S5, collecting radar data of the target to be detected, inputting the data into a target identification model, and outputting the category of the target, so that the purpose of identifying the category of the target through modeling is achieved.
Preferably, the data target in step S1 includes collecting data of vehicles, pedestrians, and non-motor vehicles in a stationary state, a moving state, a turning state, and a surrounding environment by using a millimeter wave radar.
Preferably, the step of constructing a multi-target detection state transition function by the state model in step S2 includes constructing a multi-target detection state transition function including a first error probability according to the state model.
Preferably, the state model is an empirical model based on uniform linear motion, and the first error probability is a prior error probability of predicting the multi-target motion state transition based on the uniform linear motion.
Preferably, the attributes related to the target category and greater than the preset threshold in step S4 include the distance, the speed, and the radar reflected energy RCS value of the target.
Preferably, the statistical analysis in step S4 is specifically to analyze the variance, the mean, and the distribution of the radar data.
Preferably, the classifier in step S4 includes a support vector machine SVM or a long-short-term neural network LSTM.
Preferably, the step of constructing a multi-target detection likelihood function according to the measurement model in step S4 includes constructing a multi-target detection likelihood function including a second error probability according to the measurement model, where the measurement model is an empirical model based on parameters of the vehicle-mounted millimeter wave radar, and the second error probability is a statistical modeling of an error between observation information and real information acquired by the vehicle-mounted millimeter wave radar.
(III) advantageous effects
Compared with the prior art, the invention provides a millimeter wave radar sensor modeling method applied to automatic driving, which has the following beneficial effects:
the millimeter wave radar sensor modeling method applied to automatic driving meets the requirement of real-time property of target identification by not needing a large amount of historical data for model training, the provided model has stronger practicability and short prediction time, the difference of parameters among different targets is fully understood, the identification precision of an algorithm is further improved by designing the target identification model, multi-target detection is constructed at the same time, multi-target comparison can be carried out, the error probability is reduced, the identification precision is further improved, and the phenomenon that the low precision of target association affects the target identification is avoided.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention provides a technical scheme that: a millimeter wave radar sensor modeling method applied to automatic driving comprises the following steps:
the method comprises the steps of S1, carrying out data acquisition on traffic targets in the surrounding environment by using a millimeter wave radar sensor, carrying out data processing on the acquired data targets, wherein the data processing comprises the steps of carrying out noise data cleaning processing on the data, then constructing data sets of different traffic targets through the data, and dividing each traffic target data set into a training set, a verification set and a test set, wherein the data targets comprise the steps of utilizing the millimeter wave radar to acquire data of vehicles, pedestrians and non-motor vehicles in a static state, a moving state and a turning state in the surrounding environment.
S2, a state transition function of multi-target detection is built through a state model according to the data target, the state model is a model for obtaining a multi-target prediction state set at a second moment according to the multi-target state set at a first moment, the multi-target state set is a random finite set and comprises state variables of at least two targets, the multi-target prediction state set is a random finite set and comprises state variable prediction values of the at least two targets, the step of building the state transition function of the multi-target detection through the state model comprises the step of building the multi-target detection state transition function comprising a first error probability according to the state model, the state model is an empirical model based on uniform linear motion, and the first error probability is a priori error probability based on the uniform linear motion prediction multi-target motion state transition.
And S3, constructing a likelihood function of multi-target detection according to the measurement model, wherein the measurement model is a model of a multi-target measurement set obtained according to the multi-target observation set, the multi-target observation set is a random finite set and comprises observation information acquired by the vehicle-mounted millimeter wave radar aiming at least two targets, and the multi-target measurement set is a random finite set and comprises the real information distribution probability of the at least two targets.
S4, performing correlation analysis on each attribute of the radar data, finding out the attribute of which the correlation with the target class is greater than a preset threshold value, performing statistical analysis on the radar data of the obtained attribute, constructing an empirical characteristic, training a classifier by using the empirical characteristic, the obtained attribute and a training set to obtain a target recognition model, constructing a multi-target detection likelihood function comprising a second error probability according to the measurement model, wherein the measurement model is an empirical model based on vehicle-mounted millimeter wave radar parameters, the second error probability is statistical modeling of errors between observation information and real information acquired by a vehicle-mounted millimeter wave radar, the attribute of which the correlation with the target class is greater than the preset threshold value comprises the distance, the speed and radar reflection energy RCS values of the target, performing statistical analysis to specifically analyze the variance, the mean and the distribution of the radar data, the classifier comprises a support vector machine SVM or a neural network LSTM when the length of the radar is large and small, and the step of constructing the likelihood function of multi-target detection according to the measurement model comprises constructing a likelihood detection function comprising the second error probability according to the measurement model, the multi-target model is a multi-target model based on vehicle-mounted radar parameters, and the empirical model of which the second error probability is the statistical modeling of the observation information acquired by the millimeter wave radar and the real information acquired by the radar.
And S5, collecting radar data of the target to be detected, inputting the data into a target identification model, and outputting the category of the target, so that the purpose of identifying the category of the target through modeling is achieved.
The method has the advantages that a large amount of historical data is not needed for model training, the requirement of real-time property of target identification is met, the provided model is high in practicability and short in prediction time, the difference of parameters among different targets is fully understood, the identification precision of an algorithm is further improved by designing the target identification model, multi-target detection is constructed at the same time, multi-target comparison can be carried out, the error probability is reduced, the identification precision is further improved, and the phenomenon that the low precision of target association affects target identification is avoided.
The invention has the beneficial effects that: the method has the advantages that a large amount of historical data is not needed for model training, the requirement of real-time property of target identification is met, the provided model is high in practicability and short in prediction time, the difference of parameters among different targets is fully understood, the identification precision of an algorithm is further improved by designing the target identification model, multi-target detection is constructed at the same time, multi-target comparison can be carried out, the error probability is reduced, the identification precision is further improved, and the phenomenon that the low precision of target association affects target identification is avoided.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. The term "comprising", without further limitation, means that the element so defined is not excluded from the group consisting of additional identical elements in the process, method, article, or apparatus that comprises the element.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (8)
1. A millimeter wave radar sensor modeling method applied to automatic driving is characterized by comprising the following steps:
s1, carrying out data acquisition on traffic targets in the surrounding environment by using a millimeter wave radar sensor, and then carrying out data processing on the acquired data targets, wherein the data processing comprises the steps of carrying out noise data cleaning processing on the data, then constructing data sets of different traffic targets through the data, and dividing each traffic target data set into a training set, a verification set and a test set;
s2, constructing a multi-target detection state transfer function through a state model according to the data target, wherein the state model is a model for obtaining a multi-target prediction state set at a second moment according to the multi-target state set at a first moment, the multi-target state set is a random finite set and comprises state variables of at least two targets, and the multi-target prediction state set is a random finite set and comprises state variable prediction values of at least two targets;
s3, constructing a likelihood function of multi-target detection according to a measurement model, wherein the measurement model is a model for obtaining a multi-target measurement set according to a multi-target observation set, the multi-target observation set is a random finite set and comprises observation information acquired by the vehicle-mounted millimeter wave radar aiming at least two targets, and the multi-target measurement set is a random finite set and comprises the real information distribution probability of the at least two targets;
s4, performing correlation analysis on each attribute of radar data, finding out an attribute of which the correlation with a target category is greater than a preset threshold value, performing statistical analysis on the obtained radar data of the attribute, constructing an empirical characteristic, training a classifier by using the empirical characteristic, the obtained attribute and a training set to obtain a target identification model, and constructing a multi-target detection likelihood function comprising a second error probability according to a measurement model, wherein the measurement model is an empirical model based on vehicle-mounted millimeter wave radar parameters, and the second error probability is statistical modeling of errors between observation information and real information acquired by the vehicle-mounted millimeter wave radar;
and S5, collecting radar data of the target to be detected, inputting the data into a target identification model, and outputting the category of the target, so that the purpose of identifying the category of the target through modeling is achieved.
2. The millimeter wave radar sensor modeling method applied to automatic driving according to claim 1, wherein: the data target in the step S1 comprises the step of collecting data of vehicles, pedestrians and non-motor vehicles in a static state, a moving state and a turning state and in the surrounding environment by using a millimeter wave radar.
3. The millimeter wave radar sensor modeling method applied to autopilot as set forth in claim 1, wherein: the step of constructing the multi-target detection state transfer function by the state model in the step S2 comprises constructing the multi-target detection state transfer function comprising the first error probability according to the state model.
4. The millimeter wave radar sensor modeling method applied to autopilot according to claim 3, characterized in that: the state model is an empirical model based on uniform linear motion, and the first error probability is a priori error probability of predicting multi-target motion state transition based on the uniform linear motion.
5. The millimeter wave radar sensor modeling method applied to automatic driving according to claim 1, wherein: the attributes in step S4 whose correlation with the target category is greater than the preset threshold include the distance, speed, and radar reflection energy RCS value of the target.
6. The millimeter wave radar sensor modeling method applied to automatic driving according to claim 1, wherein: the step S4 of performing statistical analysis specifically includes analyzing variance, mean, and distribution of the radar data.
7. The millimeter wave radar sensor modeling method applied to automatic driving according to claim 1, wherein: and the classifier in the step S4 comprises a Support Vector Machine (SVM) or a long-time and short-time neural network (LSTM).
8. The millimeter wave radar sensor modeling method applied to autopilot as set forth in claim 1, wherein: the step of constructing the multi-target detection likelihood function according to the measurement model in the step S4 includes constructing the multi-target detection likelihood function including a second error probability according to the measurement model, where the measurement model is an empirical model based on parameters of the vehicle-mounted millimeter wave radar, and the second error probability is a statistical modeling of an error between observation information and real information acquired by the vehicle-mounted millimeter wave radar.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211227758.XA CN115630569A (en) | 2022-10-09 | 2022-10-09 | Millimeter wave radar sensor modeling method applied to automatic driving |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211227758.XA CN115630569A (en) | 2022-10-09 | 2022-10-09 | Millimeter wave radar sensor modeling method applied to automatic driving |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115630569A true CN115630569A (en) | 2023-01-20 |
Family
ID=84903938
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211227758.XA Pending CN115630569A (en) | 2022-10-09 | 2022-10-09 | Millimeter wave radar sensor modeling method applied to automatic driving |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115630569A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116699590A (en) * | 2023-02-15 | 2023-09-05 | 深圳觅感科技有限公司 | FMCW multi-target ranging method and system based on 5.8G microwave radar |
| CN117609750A (en) * | 2024-01-19 | 2024-02-27 | 中国电子科技集团公司第五十四研究所 | Method for calculating target recognition rate interval based on electric digital data processing technology |
-
2022
- 2022-10-09 CN CN202211227758.XA patent/CN115630569A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116699590A (en) * | 2023-02-15 | 2023-09-05 | 深圳觅感科技有限公司 | FMCW multi-target ranging method and system based on 5.8G microwave radar |
| CN117609750A (en) * | 2024-01-19 | 2024-02-27 | 中国电子科技集团公司第五十四研究所 | Method for calculating target recognition rate interval based on electric digital data processing technology |
| CN117609750B (en) * | 2024-01-19 | 2024-04-09 | 中国电子科技集团公司第五十四研究所 | Method for calculating target recognition rate interval based on electric digital data processing technology |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhao et al. | Detection and tracking of pedestrians and vehicles using roadside LiDAR sensors | |
| EP3633615A1 (en) | Deep learning network and average drift-based automatic vessel tracking method and system | |
| CN115630569A (en) | Millimeter wave radar sensor modeling method applied to automatic driving | |
| CN102463990B (en) | For the system and method for tracking object | |
| Son et al. | Robust multi-lane detection and tracking using adaptive threshold and lane classification | |
| CN107193012A (en) | Intelligent vehicle laser radar multiple-moving target tracking method based on IMM MHT algorithms | |
| Jeong et al. | Bidirectional long shot-term memory-based interactive motion prediction of cut-in vehicles in urban environments | |
| CN112487905B (en) | Method and system for predicting danger level of pedestrian around vehicle | |
| Li et al. | An adaptive 3D grid-based clustering algorithm for automotive high resolution radar sensor | |
| Hafeez et al. | Insights and strategies for an autonomous vehicle with a sensor fusion innovation: A fictional outlook | |
| CN114488194A (en) | Method for detecting and identifying targets under structured road of intelligent driving vehicle | |
| Zhang et al. | A framework for turning behavior classification at intersections using 3D LIDAR | |
| CN114998276B (en) | Robot dynamic obstacle real-time detection method based on three-dimensional point cloud | |
| CN114879192A (en) | Decision tree vehicle type classification method based on road side millimeter wave radar and electronic equipment | |
| CN114359876B (en) | Vehicle target identification method and storage medium | |
| CN115761534A (en) | Method for detecting and tracking small target of infrared unmanned aerial vehicle under air background | |
| Qing et al. | A novel particle filter implementation for a multiple-vehicle detection and tracking system using tail light segmentation | |
| Bouain et al. | Multi-sensor fusion for obstacle detection and recognition: A belief-based approach | |
| CN113432615B (en) | Detection method and system based on multi-sensor fusion drivable area and vehicle | |
| Lin et al. | Density variation-based background filtering algorithm for low-channel roadside lidar data | |
| Kim et al. | Semantic point cloud-based adaptive multiple object detection and tracking for autonomous vehicles | |
| Lindenmaier et al. | GM-PHD filter based sensor data fusion for automotive frontal perception system | |
| Li et al. | Obstacle detection and tracking algorithm based on multi‐lidar fusion in urban environment | |
| CN115083199A (en) | Parking space information determination method and related equipment thereof | |
| Liu et al. | Research on security of key algorithms in intelligent driving system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication |