CN115183786A - Training method and device of sensor error prediction model for automatic driving - Google Patents

Training method and device of sensor error prediction model for automatic driving Download PDF

Info

Publication number
CN115183786A
CN115183786A CN202210802317.1A CN202210802317A CN115183786A CN 115183786 A CN115183786 A CN 115183786A CN 202210802317 A CN202210802317 A CN 202210802317A CN 115183786 A CN115183786 A CN 115183786A
Authority
CN
China
Prior art keywords
information
error
positioning
sensor
prediction model
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
Application number
CN202210802317.1A
Other languages
Chinese (zh)
Inventor
费再慧
李岩
张海强
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhidao Network Technology Beijing Co Ltd
Original Assignee
Zhidao Network Technology Beijing Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Zhidao Network Technology Beijing Co Ltd filed Critical Zhidao Network Technology Beijing Co Ltd
Priority to CN202210802317.1A priority Critical patent/CN115183786A/en
Publication of CN115183786A publication Critical patent/CN115183786A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
    • G01C21/28—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network with correlation of data from several navigational instruments
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/20—Instruments for performing navigational calculations
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
    • G01C21/34—Route searching; Route guidance
    • G01C21/3407—Route searching; Route guidance specially adapted for specific applications
    • 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
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/10—Internal combustion engine [ICE] based vehicles
    • Y02T10/40—Engine management systems

Landscapes

  • Engineering & Computer Science (AREA)
  • Remote Sensing (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Automation & Control Theory (AREA)
  • General Health & Medical Sciences (AREA)
  • Biophysics (AREA)
  • Evolutionary Computation (AREA)
  • Computational Linguistics (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Navigation (AREA)

Abstract

The application discloses a training method and a device of a sensor error prediction model for automatic driving, wherein the method comprises the following steps: acquiring high-precision positioning information of an automatic driving vehicle; acquiring target sensor information of the automatic driving vehicle under the condition that the high-precision positioning information meets a preset training condition, wherein the target sensor information comprises positioning information and original positioning error information of a target sensor; determining the real correction error of the target sensor according to the high-precision positioning information and the positioning information of the target sensor; and training a sensor error prediction model by using the original positioning error information and the real correction error of the target sensor to obtain the trained sensor error prediction model. The real correction error of the sensor is determined based on the high-precision positioning information meeting the preset training condition, and the sensor error prediction model is trained based on the real correction error, so that the prediction accuracy of the positioning error between the sensors is improved, and powerful support is provided for fusion positioning of the automatic driving vehicle.

Description

Training method and device for sensor error prediction model for automatic driving
Technical Field
The application relates to the technical field of automatic driving, in particular to a training method and a training device for a sensor error prediction model for automatic driving.
Background
Under the automatic driving scene, data of a plurality of sensors are often required to be fused to obtain fusion positioning information, and the positioning accuracy of the automatic driving vehicle under various urban complex road conditions can be ensured.
However, due to factors such as calibration and time delay among the sensors, a large error exists between the positioning information of the sensors acquired during fusion positioning, and a larger error is caused instead by directly performing fusion positioning.
Disclosure of Invention
The embodiment of the application provides a training method and a training device for a sensor error prediction model for automatic driving, so that the prediction accuracy of positioning errors among sensors is improved.
The embodiment of the application adopts the following technical scheme:
in a first aspect, an embodiment of the present application provides a method for training a sensor error prediction model for automatic driving, where the method includes:
acquiring high-precision positioning information of an automatic driving vehicle;
acquiring target sensor information of the automatic driving vehicle under the condition that the high-precision positioning information meets a preset training condition, wherein the target sensor information comprises positioning information and original positioning error information of a target sensor;
determining a real correction error of the target sensor according to the high-precision positioning information and the positioning information of the target sensor;
and training a sensor error prediction model by using the original positioning error information and the real correction error of the target sensor to obtain a trained sensor error prediction model.
Optionally, the obtaining target sensor information of the autonomous vehicle when the high-precision positioning information meets a preset training condition includes:
predicting the confidence of the high-precision positioning information by using a positioning confidence prediction model;
if the confidence coefficient of the high-precision positioning information is greater than a preset confidence coefficient threshold value, determining that the high-precision positioning information meets the preset training condition;
otherwise, determining that the high-precision positioning information does not meet the preset training condition.
Optionally, the original positioning error information includes a SLAM covariance and an original correction amount, the training a sensor error prediction model by using the original positioning error information and a real correction error of the target sensor, and obtaining the trained sensor error prediction model includes:
and taking the SLAM covariance and the original correction quantity as the input of the sensor error prediction model, taking the real correction error of the target sensor as a supervision signal, training the sensor error prediction model, and obtaining the trained sensor error prediction model.
Optionally, after determining the real correction error of the target sensor according to the high-precision positioning information and the positioning information of the target sensor, the method further includes:
based on the original positioning error information and the real correction error of the target sensor, fitting optimization is carried out by utilizing a nonlinear optimization algorithm to obtain the corresponding relation between the original positioning error information and the real correction error of the target sensor.
Optionally, the sensor error prediction model employs an LSTM long-short term memory network.
In a second aspect, the present application further provides a fusion positioning method for an autonomous vehicle, where the method includes:
acquiring current target sensor information, wherein the current target sensor information comprises current positioning information and current original positioning error information;
predicting a real correction error of the target sensor by using a sensor error prediction model based on the current target sensor information;
correcting the current positioning information by using the real correction error of the target sensor to obtain corrected positioning information;
performing fusion positioning based on the corrected positioning information to obtain a fusion positioning result of the automatic driving vehicle;
the sensor error prediction model is obtained by training based on any one of the training methods of the sensor error prediction model.
In a third aspect, an embodiment of the present application further provides a training apparatus for a sensor error prediction model for automatic driving, where the apparatus includes:
a first acquisition unit for acquiring high-precision positioning information of an autonomous vehicle;
the second acquisition unit is used for acquiring target sensor information of the automatic driving vehicle under the condition that the high-precision positioning information meets a preset training condition, wherein the target sensor information comprises positioning information of a target sensor and original positioning error information;
the determining unit is used for determining the real correction error of the target sensor according to the high-precision positioning information and the positioning information of the target sensor;
and the training unit is used for training a sensor error prediction model by utilizing the original positioning error information and the real correction error of the target sensor to obtain the trained sensor error prediction model.
In a fourth aspect, the present application further provides a fusion positioning apparatus for an autonomous vehicle, where the apparatus includes:
a third obtaining unit, configured to obtain current target sensor information, where the current target sensor information includes current positioning information and current original positioning error information;
a prediction unit for predicting a true correction error of the target sensor using a sensor error prediction model based on the current target sensor information;
the correction unit is used for correcting the current positioning information by using the real correction error of the target sensor to obtain corrected positioning information;
the fusion positioning unit is used for carrying out fusion positioning on the basis of the corrected positioning information to obtain a fusion positioning result of the automatic driving vehicle;
the sensor error prediction model is obtained by training based on the training device of the sensor error prediction model.
In a fifth aspect, an embodiment of the present application further provides an electronic device, including:
a processor; and
a memory arranged to store computer executable instructions that when executed cause the processor to perform any of the aforementioned methods of training a sensor error prediction model for autonomous driving, or the aforementioned methods of fusion localization of autonomous driving vehicles.
In a sixth aspect, embodiments of the present application further provide a computer readable storage medium storing one or more programs, which when executed by an electronic device including a plurality of application programs, cause the electronic device to perform any one of the methods for training a sensor error prediction model for autonomous driving or the method for fusion location of an autonomous driving vehicle.
The embodiment of the application adopts at least one technical scheme which can achieve the following beneficial effects: according to the training method of the sensor error prediction model for automatic driving, high-precision positioning information of an automatic driving vehicle is obtained; then, under the condition that the high-precision positioning information meets the preset training condition, acquiring target sensor information of the automatic driving vehicle, wherein the target sensor information comprises positioning information and original positioning error information of a target sensor; then, determining the real correction error of the target sensor according to the high-precision positioning information and the positioning information of the target sensor; and finally, training a sensor error prediction model by using the original positioning error information and the real correction error of the target sensor to obtain the trained sensor error prediction model. The training method for the sensor error prediction model for automatic driving determines the real correction error of the sensor based on the high-precision positioning information meeting the preset training condition, trains the sensor error prediction model based on the real correction error, improves the prediction accuracy of the positioning error between the sensors, and provides powerful support for the fusion positioning of subsequent automatic driving vehicles.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the application and together with the description serve to explain the application and not to limit the application. In the drawings:
FIG. 1 is a schematic flow chart illustrating a method for training a sensor error prediction model for autonomous driving according to an embodiment of the present disclosure;
FIG. 2 is a schematic flow chart illustrating a fusion positioning method for an autonomous vehicle according to an embodiment of the present disclosure;
FIG. 3 is a schematic structural diagram of a training apparatus for a sensor error prediction model for automatic driving according to an embodiment of the present application;
FIG. 4 is a schematic structural diagram of a fusion positioning device of an autonomous vehicle according to an embodiment of the present disclosure;
fig. 5 is a schematic structural diagram of an electronic device in an embodiment of the present application.
Detailed Description
To make the objects, technical solutions and advantages of the present application more clear, the technical solutions of the present application will be clearly and completely described below with reference to specific embodiments of the present application and the accompanying drawings. It should be apparent that the described embodiments are only a few embodiments of the present application, and not all 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 application.
The technical solutions provided by the embodiments of the present application are described in detail below with reference to the accompanying drawings.
The embodiment of the present application provides a method for training a sensor error prediction model for automatic driving, and as shown in fig. 1, provides a flow chart of the method for training a sensor error prediction model for automatic driving in the embodiment of the present application, where the method at least includes the following steps S110 to S140:
step S110, high-precision positioning information of the autonomous vehicle is acquired.
When the sensor error prediction model is trained, high-precision positioning information of an automatic driving vehicle needs to be acquired first, the high-precision positioning information can be positioning information output by Inertial navigation RTK (Real-time kinematic), the Inertial navigation RTK is positioning information output after IMU (Inertial Measurement Unit) positioning data acquired by Inertial navigation equipment and RTK (Real-time kinematic) differential algorithm are fused, and the positioning information has higher positioning precision compared with simple Inertial navigation positioning data and RTK positioning data, so that the positioning information can be used as a basis for subsequently measuring the positioning error of the sensor.
And step S120, acquiring target sensor information of the automatic driving vehicle under the condition that the high-precision positioning information meets a preset training condition, wherein the target sensor information comprises positioning information of a target sensor and original positioning error information.
The "high-precision positioning information" in the embodiment of the present application mainly refers to positioning information obtained based on inertial navigation RTK, but does not represent that the positioning information always has higher precision, for example, the positioning information may be affected by satellite positioning signal quality to reduce precision, and the precision of the high-precision positioning information may be approximately reflected in the RTK positioning state to a certain extent, but the precision of the high-precision positioning information determined by solely depending on the RTK state cannot meet the requirement of automatic driving, for example, in an urban complex road scene, the positioning error of the high-precision positioning information may still be larger in the case that the RTK positioning state is a fixed solution.
Based on this, the embodiment of the application needs to determine whether the precision of the high-precision positioning information meets the preset training condition through a certain strategy, and only when the preset training condition is met, the high-precision positioning information can be used as a basis for measuring the error of the sensor, so that the precision and the effect of model training are improved.
If the current high-precision positioning information meets the preset training condition, target sensor information of the autonomous vehicle can be further obtained, the target sensor can be understood as a target object needing to build a sensor error prediction model, such as a laser radar, a camera and the like, and the target sensor information specifically can include positioning information and original positioning error information output by the target sensor. The original positioning error information can be regarded as a preliminary positioning error information calculated by the target sensor based on a self positioning algorithm and a strategy, and the positioning error information is generally low in precision, so that positioning correction cannot be directly performed based on the positioning error information, which is also a main reason for constructing a sensor error prediction model in the embodiment of the application.
Step S130, determining the real correction error of the target sensor according to the high-precision positioning information and the positioning information of the target sensor.
As described above, the high-precision positioning information can be used as a basis for measuring the positioning error of the target sensor when the preset training condition is satisfied, and therefore, the positioning deviation between the positioning information of the target sensor and the high-precision positioning information can be calculated here as the real correction error. The ultimate goal of model training is to predict the correct correction error as the basis for correcting the positioning information.
Therefore, the embodiment of the application is equivalent to that in the training phase, the inertial navigation RTK equipment and the positioning algorithm based on the target sensor are started simultaneously, so that the real correction error is determined.
And step S140, training a sensor error prediction model by using the original positioning error information and the real correction error of the target sensor to obtain a trained sensor error prediction model.
Although the original positioning error information cannot be directly used for correcting the positioning error, the original positioning error information can be used as input information in a model training stage, the actual correction error of the target sensor is the actual positioning error, and the accuracy is high, so that the training process of the model can be constrained, and the sensor error prediction model can be trained on the basis of the information of the two dimensions, so that the trained sensor error prediction model is obtained.
The training method for the sensor error prediction model for automatic driving determines the real correction error of the sensor based on the high-precision positioning information meeting the preset training condition, trains the sensor error prediction model based on the real correction error, improves the prediction accuracy of the positioning error between the sensors, and provides powerful support for the fusion positioning of subsequent automatic driving vehicles.
In some embodiments of the application, the sensor error prediction model of the embodiment of the application can realize online training based on interaction between a vehicle end and a cloud end, for example, the vehicle end of an automatic driving vehicle can acquire inertial RTK positioning information in real time and judge the precision of the inertial RTK positioning information, and when the precision meets a preset training condition, the inertial RTK positioning information and target sensor information are sent to the cloud end together for online training, so that the vehicle end operation load can be reduced, and the real-time requirement is met. Of course, an off-line training mode, which is specifically adopted, may also be adopted, and those skilled in the art may flexibly select according to actual requirements.
In some embodiments of the application, the high-precision positioning information of the embodiment of the application can be acquired through high-precision inertial navigation equipment installed on a vehicle, the positioning effect of the high-precision inertial navigation equipment is slightly influenced by external factors, and the high-precision inertial navigation equipment has higher positioning precision compared with general inertial navigation equipment.
In some embodiments of the present application, the obtaining target sensor information of the autonomous vehicle in a case where the high-precision positioning information satisfies a preset training condition includes: predicting the confidence of the high-precision positioning information by using a positioning confidence prediction model; if the confidence of the high-precision positioning information is greater than a preset confidence threshold, determining that the high-precision positioning information meets the preset training condition; otherwise, determining that the high-precision positioning information does not meet the preset training condition.
When determining whether the high-precision positioning information meets the preset training condition, the confidence degree of the currently acquired high-precision positioning information, namely the credibility of the positioning result, can be determined, the confidence degree of the high-precision positioning information can be predicted by using a positioning confidence degree prediction model trained in advance, then the confidence degree of the high-precision positioning information is compared with a preset confidence degree threshold, if the confidence degree of the high-precision positioning information is greater than the preset confidence degree threshold, the precision of the high-precision positioning information is considered to meet the requirement of subsequent training, otherwise, the high-precision positioning information cannot be used for the subsequent training, and the precision of model training is ensured.
The size of the preset confidence threshold may be flexibly set according to actual training requirements, for example, may be set to 0.95, and when the confidence of the high-precision positioning information exceeds 0.95, it may be considered that the precision is higher, and the high-precision positioning information may be used for subsequent training.
The positioning confidence prediction model of the embodiment of the application can be obtained by training in the following way: acquiring inertial navigation RTK information and corresponding high-precision positioning information, wherein the inertial navigation RTK information comprises inertial navigation RTK positioning information, absolute time of the inertial navigation RTK information, an RTK positioning state, a horizontal position precision factor and satellite number; determining a positioning error of inertial navigation RTK positioning information according to the high-precision positioning information; determining the confidence coefficient of the inertial navigation RTK positioning information according to the positioning error of the inertial navigation RTK positioning information; and training a position confidence degree prediction model by using the inertial navigation RTK information and the confidence degree of the inertial navigation RTK positioning information to obtain the trained position confidence degree prediction model.
When determining the confidence of the inertial navigation RTK positioning information according to the positioning error of the inertial navigation RTK positioning information, the embodiment of the application can specifically adopt the following mode: if the positioning error of the inertial navigation RTK positioning information is not larger than a first preset error threshold, determining the confidence coefficient of the inertial navigation RTK positioning information as a first confidence coefficient; if the positioning error of the inertial navigation RTK positioning information is not smaller than a second preset error threshold, determining the confidence coefficient of the inertial navigation RTK positioning information as a second confidence coefficient; if the positioning error of the inertial navigation RTK positioning information is larger than a first preset error threshold and smaller than a second preset error threshold, determining the confidence coefficient of the inertial navigation RTK positioning information as a third confidence coefficient; the first preset error threshold is smaller than the second preset error threshold, the first confidence coefficient is larger than the third confidence coefficient, and the third confidence coefficient is larger than the second confidence coefficient.
In some embodiments of the present application, the raw positioning error information includes SLAM covariance and raw correction amount, and the training a sensor error prediction model using the raw positioning error information and a real correction error of the target sensor to obtain a trained sensor error prediction model includes: and taking the SLAM covariance and the original correction quantity as the input of the sensor error prediction model, taking the real correction error of the target sensor as a supervision signal, training the sensor error prediction model, and obtaining the trained sensor error prediction model.
As described above, the target sensor in the embodiment of the present application may refer to a laser radar, a camera, and the like, and a positioning scheme based on a laser SLAM (Simultaneous Localization And Mapping) may be implemented based on data acquired by the laser radar, and a positioning scheme based on a visual SLAM may be implemented based on data acquired by the camera.
The main function of the laser SLAM or the visual SLAM is to provide an auxiliary positioning function when high-precision positioning signals such as RTK and the like are interfered or positioning information with high precision cannot be provided, so that the final fusion positioning precision of the autonomous vehicle is also influenced by the positioning deviation between the laser SLAM or the visual SLAM and other sensors. Based on this, the purpose of training the sensor error prediction model in the embodiments of the present application is to accurately predict the positioning deviation between sensors such as a laser SLAM or a visual SLAM, so as to perform positioning correction, thereby improving the fusion positioning accuracy.
The positioning algorithm based on the laser SLAM/visual SLAM can output an SLAM covariance and a corresponding original correction quantity, the SLAM covariance represents the positioning deviation between the laser SLAM/visual SLAM and other sensors, the original correction quantity is an initial correction value calculated by the positioning algorithm based on the covariance, the correction value is generally low in precision, and the sensitivity is high only when the positioning error is large.
Therefore, according to the embodiment of the application, the SLAM covariance and the corresponding original correction quantity can be used as the input of the sensor error prediction model, although the original correction quantity has a certain error, the original correction quantity can be used as reference information input to improve the training efficiency of the model, the real correction error of the target sensor is used as the output of the sensor error prediction model, so that the training process of the model is supervised and constrained, when the model prediction accuracy meets the preset accuracy requirement, the training is ended, and the trained sensor error prediction model is output.
In some embodiments of the present application, after determining the true correction error of the target sensor from the high-precision positioning information and the positioning information of the target sensor, the method further comprises: and fitting and optimizing by utilizing a nonlinear optimization algorithm based on the original positioning error information and the real correction error of the target sensor to obtain the corresponding relation between the original positioning error information and the real correction error of the target sensor.
In addition to the fact that the correction error of the sensor can be predicted by training the sensor error prediction model in a self-adaptive training mode, the corresponding relation between the original positioning error information and the real correction error of the target sensor can be fitted and optimized by adopting a certain nonlinear optimization algorithm such as least square and LM (Levenberg-Marquardt ) and the like, and then the fitting optimization result can be used as the basis for predicting the real correction error of the target sensor subsequently.
In some embodiments of the present application, the sensor error prediction model employs an LSTM long-short term memory network.
The sensor error prediction model of the embodiment of the application can be trained by adopting the structure of an LSTM (Long Short-Term Memory) Long-Short Term Memory network. LSTM is a special recurrent neural network with the ability to learn long-term dependencies, and LSTM can remember that long-term information is actually their own attributes, rather than the ability to be gained through learning or training. In the embodiment of the application, the SLAM covariance, the corresponding original correction amount and the like are parameters related to a time sequence, so that the requirement of an actual scene is better met by training by adopting an LSTM network structure.
It should be noted that, besides the LSTM network may be used to train the sensor error prediction model, a traditional BP (Back Propagation) neural network may also be used to train the sensor error prediction model, and those skilled in the art may flexibly select the model according to actual requirements, which is not listed here.
The embodiment of the present application further provides a fusion positioning method for an autonomous vehicle, and as shown in fig. 2, a flow diagram of the fusion positioning method for an autonomous vehicle in the embodiment of the present application is provided, where the method at least includes the following steps S210 to S240:
step S210, obtaining current target sensor information, wherein the current target sensor information comprises current positioning information and current original positioning error information;
step S220, based on the current target sensor information, predicting the real correction error of the target sensor by using a sensor error prediction model;
step S230, correcting the current positioning information by using the real correction error of the target sensor to obtain corrected positioning information;
step S240, carrying out fusion positioning based on the corrected positioning information to obtain a fusion positioning result of the automatic driving vehicle;
the sensor error prediction model is obtained by training based on any one of the training methods of the sensor error prediction model.
In an actual fusion positioning scenario, current target sensor information, specifically including current positioning information such as positioning information of laser SLAM/visual SLAM, and current original positioning error information such as SLAM covariance and original correction amount, needs to be acquired.
The above information is input into the trained sensor error prediction model in the foregoing embodiment, a real correction error (dx ', dy') of the target sensor is predicted, and the current positioning information (dx, dy) is corrected by using the real correction error, so as to obtain corrected positioning information (dx 1, dy 1), which may be expressed as follows, for example:
dx1＝dx+dx’
dy1＝dy+dy’
and finally, the corrected positioning information can be used as observation information and is input into a Kalman filter or an extended Kalman filter together with positioning data of other sensors, such as IMU positioning data, wheel speed data and the like, so as to perform fusion positioning, and a fusion positioning result of the automatic driving vehicle is obtained.
The embodiment of the present application further provides a training apparatus 300 for a sensor error prediction model for automatic driving, and as shown in fig. 3, a schematic structural diagram of the training apparatus for a sensor error prediction model for automatic driving in the embodiment of the present application is provided, where the apparatus 300 includes: a first obtaining unit 310, a second obtaining unit 320, a determining unit 330, and a training unit 340, wherein:
a first acquisition unit 310 for acquiring high-precision positioning information of an autonomous vehicle;
a second obtaining unit 320, configured to obtain target sensor information of the autonomous vehicle when the high-precision positioning information meets a preset training condition, where the target sensor information includes positioning information of a target sensor and original positioning error information;
a determining unit 330, configured to determine a real correction error of the target sensor according to the high-precision positioning information and the positioning information of the target sensor;
and the training unit 340 is configured to train a sensor error prediction model by using the original positioning error information and the real correction error of the target sensor, so as to obtain a trained sensor error prediction model.
In some embodiments of the present application, the second obtaining unit 320 is specifically configured to: predicting the confidence of the high-precision positioning information by using a positioning confidence prediction model; if the confidence of the high-precision positioning information is greater than a preset confidence threshold, determining that the high-precision positioning information meets the preset training condition; otherwise, determining that the high-precision positioning information does not meet the preset training condition.
In some embodiments of the present application, the raw positioning error information includes SLAM covariance and raw correction, and the training unit 340 is specifically configured to: and taking the SLAM covariance and the original correction quantity as the input of the sensor error prediction model, taking the real correction error of the target sensor as a supervision signal, training the sensor error prediction model, and obtaining the trained sensor error prediction model.
In some embodiments of the present application, the apparatus further comprises: and the fitting unit is used for performing fitting optimization by using a nonlinear optimization algorithm based on the original positioning error information and the real correction error of the target sensor to obtain the corresponding relation between the original positioning error information and the real correction error of the target sensor.
In some embodiments of the present application, the sensor error prediction model employs an LSTM long-short term memory network.
It can be understood that the above-mentioned training apparatus for a sensor error prediction model for automatic driving can implement the steps of the training method for a sensor error prediction model for automatic driving provided in the foregoing embodiments, and the explanations regarding the training method for a sensor error prediction model for automatic driving are applicable to the training apparatus for a sensor error prediction model for automatic driving, and are not repeated herein.
The embodiment of the present application further provides a fusion positioning device 400 of an autonomous vehicle, as shown in fig. 4, which provides a schematic structural diagram of a fusion positioning device of an autonomous vehicle in the embodiment of the present application, where the device 400 includes: a third obtaining unit 410, a prediction unit 420, a correction unit 430, and a fusion positioning unit 440, wherein:
a third obtaining unit 410, configured to obtain current target sensor information, where the current target sensor information includes current positioning information and current raw positioning error information;
a prediction unit 420 for predicting a true correction error of the target sensor using a sensor error prediction model based on the current target sensor information;
a correcting unit 430, configured to correct the current positioning information by using the real correction error of the target sensor, so as to obtain corrected positioning information;
a fusion positioning unit 440, configured to perform fusion positioning based on the corrected positioning information to obtain a fusion positioning result of the autonomous vehicle;
the sensor error prediction model is obtained by training based on the training device of the sensor error prediction model.
It can be understood that the fusion positioning device for an autonomous vehicle can implement the steps of the fusion positioning method for an autonomous vehicle provided in the foregoing embodiment, and the explanations related to the fusion positioning method for an autonomous vehicle are applicable to the fusion positioning device for an autonomous vehicle, and are not repeated herein.
Fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present application. Referring to fig. 5, at a hardware level, the electronic device includes a processor, and optionally further includes an internal bus, a network interface, and a memory. The Memory may include a Memory, such as a Random-Access Memory (RAM), and may further include a non-volatile Memory, such as at least 1 disk Memory. Of course, the electronic device may also include hardware required for other services.
The processor, the network interface, and the memory may be connected to each other via an internal bus, which may be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, an EISA (Extended Industry Standard Architecture) bus, or the like. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one double-headed arrow is shown in FIG. 5, but this does not indicate only one bus or one type of bus.
And the memory is used for storing programs. In particular, the program may include program code comprising computer operating instructions. The memory may include both memory and non-volatile storage and provides instructions and data to the processor.
The processor reads a corresponding computer program from the nonvolatile memory into the memory and then runs the computer program to form a training device of a sensor error prediction model for automatic driving on a logic level. The processor is used for executing the program stored in the memory and is specifically used for executing the following operations:
acquiring high-precision positioning information of an automatic driving vehicle;
acquiring target sensor information of the automatic driving vehicle under the condition that the high-precision positioning information meets a preset training condition, wherein the target sensor information comprises positioning information and original positioning error information of a target sensor;
determining a real correction error of the target sensor according to the high-precision positioning information and the positioning information of the target sensor;
and training a sensor error prediction model by using the original positioning error information and the real correction error of the target sensor to obtain a trained sensor error prediction model.
The method performed by the training apparatus for the sensor error prediction model for automatic driving disclosed in the embodiment of fig. 1 of the present application may be applied to or implemented by a processor. The processor may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuits of hardware in a processor or by instructions in the form of software. The Processor may be a general-purpose Processor, including a Central Processing Unit (CPU), a Network Processor (NP), and the like; but also Digital Signal Processors (DSPs), application Specific Integrated Circuits (ASICs), field Programmable Gate Arrays (FPGAs) or other Programmable logic devices, discrete Gate or transistor logic devices, discrete hardware components. The various methods, steps, and logic blocks disclosed in the embodiments of the present application may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present application may be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software module may be located in ram, flash memory, rom, prom, or eprom, registers, etc. storage media as is well known in the art. The storage medium is located in a memory, and a processor reads information in the memory and completes the steps of the method in combination with hardware of the processor.
The electronic device may further execute the method executed by the training apparatus for the sensor error prediction model for automatic driving in fig. 1, and implement the functions of the training apparatus for the sensor error prediction model for automatic driving in the embodiment shown in fig. 1, which are not described herein again.
Embodiments of the present application further propose a computer-readable storage medium storing one or more programs, the one or more programs including instructions, which when executed by an electronic device including a plurality of application programs, enable the electronic device to perform the method performed by the training apparatus for a sensor error prediction model for automatic driving in the embodiment shown in fig. 1, and in particular to perform:
acquiring high-precision positioning information of an automatic driving vehicle;
acquiring target sensor information of the automatic driving vehicle under the condition that the high-precision positioning information meets a preset training condition, wherein the target sensor information comprises positioning information and original positioning error information of a target sensor;
determining a real correction error of the target sensor according to the high-precision positioning information and the positioning information of the target sensor;
and training a sensor error prediction model by using the original positioning error information and the real correction error of the target sensor to obtain the trained sensor error prediction model.
It should be noted that the electronic device according to the embodiment of the present application may also be used to execute the method executed by the fusion positioning apparatus of the autonomous vehicle disclosed in the embodiment shown in fig. 2, and details are not repeated.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include forms of volatile memory in a computer readable medium, random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.
Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.
It should also be noted that 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. Without further limitation, an element defined by the phrases "comprising a," "8230," "8230," or "comprising" does not exclude the presence of other like elements in a process, method, article, or apparatus comprising the element.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The above description is only an example of the present application and is not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims (10)

1. A method of training a sensor error prediction model for autonomous driving, wherein the method comprises:
acquiring high-precision positioning information of an automatic driving vehicle;
acquiring target sensor information of the automatic driving vehicle under the condition that the high-precision positioning information meets a preset training condition, wherein the target sensor information comprises positioning information and original positioning error information of a target sensor;
determining a real correction error of the target sensor according to the high-precision positioning information and the positioning information of the target sensor;
and training a sensor error prediction model by using the original positioning error information and the real correction error of the target sensor to obtain the trained sensor error prediction model.
2. The method of claim 1, wherein said obtaining target sensor information of the autonomous vehicle in the case that the high-precision positioning information satisfies a preset training condition comprises:
predicting the confidence of the high-precision positioning information by using a positioning confidence prediction model;
if the confidence of the high-precision positioning information is greater than a preset confidence threshold, determining that the high-precision positioning information meets the preset training condition;
otherwise, determining that the high-precision positioning information does not meet the preset training condition.
3. The method of claim 1, wherein the raw positioning error information includes SLAM covariance and raw correction, and wherein training a sensor error prediction model using the raw positioning error information and actual correction error of the target sensor to obtain a trained sensor error prediction model comprises:
and taking the SLAM covariance and the original correction quantity as the input of the sensor error prediction model, taking the real correction error of the target sensor as a supervision signal, training the sensor error prediction model, and obtaining the trained sensor error prediction model.
4. The method of claim 1, wherein after determining the true correction error of the target sensor based on the high accuracy positioning information and the positioning information of the target sensor, the method further comprises:
and fitting and optimizing by utilizing a nonlinear optimization algorithm based on the original positioning error information and the real correction error of the target sensor to obtain the corresponding relation between the original positioning error information and the real correction error of the target sensor.
5. The method of any one of claims 1 to 4, wherein the sensor error prediction model employs an LSTM long and short term memory network.
6. A fusion localization method of an autonomous vehicle, wherein the method comprises:
acquiring current target sensor information, wherein the current target sensor information comprises current positioning information and current original positioning error information;
predicting a true correction error of the target sensor based on the current target sensor information by using a sensor error prediction model;
correcting the current positioning information by using the real correction error of the target sensor to obtain corrected positioning information;
performing fusion positioning based on the corrected positioning information to obtain a fusion positioning result of the automatic driving vehicle;
wherein the sensor error prediction model is obtained by training based on the training method of the sensor error prediction model according to any one of claims 1 to 5.
7. A training apparatus of a sensor error prediction model for autonomous driving, wherein the apparatus comprises:
a first acquisition unit for acquiring high-precision positioning information of an autonomous vehicle;
the second acquisition unit is used for acquiring target sensor information of the automatic driving vehicle under the condition that the high-precision positioning information meets a preset training condition, wherein the target sensor information comprises positioning information of a target sensor and original positioning error information;
the determining unit is used for determining the real correction error of the target sensor according to the high-precision positioning information and the positioning information of the target sensor;
and the training unit is used for training a sensor error prediction model by utilizing the original positioning error information and the real correction error of the target sensor to obtain the trained sensor error prediction model.
8. A fusion positioning apparatus of an autonomous vehicle, wherein the apparatus comprises:
a third obtaining unit, configured to obtain current target sensor information, where the current target sensor information includes current positioning information and current original positioning error information;
a prediction unit for predicting a true correction error of the target sensor using a sensor error prediction model based on the current target sensor information;
the correction unit is used for correcting the current positioning information by using the real correction error of the target sensor to obtain corrected positioning information;
the fusion positioning unit is used for carrying out fusion positioning on the basis of the corrected positioning information to obtain a fusion positioning result of the automatic driving vehicle;
wherein the sensor error prediction model is trained based on the training device of the sensor error prediction model of claim 7.
9. An electronic device, comprising:
a processor; and
a memory arranged to store computer executable instructions that when executed cause the processor to perform a method of training a sensor error prediction model for autonomous driving as claimed in any one of claims 1 to 5, or a method of fusion localization of autonomous vehicles as claimed in claim 6.
10. A computer readable storage medium storing one or more programs which, when executed by an electronic device including a plurality of application programs, cause the electronic device to perform a method of training a sensor error prediction model for autonomous driving according to any one of claims 1 to 5, or a method of fusion localization of an autonomous vehicle according to claim 6.
CN202210802317.1A 2022-07-07 2022-07-07 Training method and device of sensor error prediction model for automatic driving Pending CN115183786A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210802317.1A CN115183786A (en) 2022-07-07 2022-07-07 Training method and device of sensor error prediction model for automatic driving

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210802317.1A CN115183786A (en) 2022-07-07 2022-07-07 Training method and device of sensor error prediction model for automatic driving

Publications (1)

Publication Number Publication Date
CN115183786A true CN115183786A (en) 2022-10-14

Family

ID=83517937

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210802317.1A Pending CN115183786A (en) 2022-07-07 2022-07-07 Training method and device of sensor error prediction model for automatic driving

Country Status (1)

Country Link
CN (1) CN115183786A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117706583A (en) * 2023-12-29 2024-03-15 无锡物联网创新中心有限公司 High-precision positioning method and system

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117706583A (en) * 2023-12-29 2024-03-15 无锡物联网创新中心有限公司 High-precision positioning method and system

Similar Documents

Publication Publication Date Title
CN111077555B (en) Positioning method and device
CN113899374B (en) Automatic driving vehicle positioning method and device, electronic equipment and storage medium
CN113791435B (en) GNSS signal abnormal value detection method and device, electronic equipment and storage medium
CN115390086B (en) Fusion positioning method and device for automatic driving, electronic equipment and storage medium
CN114894214A (en) Fusion positioning method and device for automatic driving vehicle and electronic equipment
CN114279453B (en) Automatic driving vehicle positioning method and device based on vehicle-road cooperation and electronic equipment
CN113360795A (en) Method and device for correcting vehicle course angle, electronic equipment and storage medium
CN111127551A (en) Target detection method and device
CN114973198A (en) Course angle prediction method and device of target vehicle, electronic equipment and storage medium
CN115183786A (en) Training method and device of sensor error prediction model for automatic driving
CN114877900A (en) Automatic driving vehicle fusion positioning method for tunnel and related device
CN114114369B (en) Autonomous vehicle positioning method and apparatus, electronic device, and storage medium
CN115962774A (en) Point cloud map updating method and device, electronic equipment and storage medium
CN114993333A (en) Fusion positioning method and device for automatic driving vehicle and electronic equipment
CN115056801A (en) Multipath recognition method and device for automatic driving, electronic equipment and storage medium
CN115856976B (en) Fusion positioning method and device for automatic driving vehicle and electronic equipment
CN115856979B (en) Positioning method and device for automatic driving vehicle, electronic equipment and storage medium
CN115950441A (en) Fusion positioning method and device for automatic driving vehicle and electronic equipment
CN115014332A (en) Laser SLAM mapping method and device, electronic equipment and computer readable storage medium
CN116148821A (en) Laser radar external parameter correction method and device, electronic equipment and storage medium
CN113932835B (en) Calibration method and device for positioning lever arm of automatic driving vehicle and electronic equipment
US20240011792A1 (en) Method and apparatus for updating confidence of high-precision map
CN116106869A (en) Positioning evaluation method and device for automatic driving vehicle and electronic equipment
CN114739416A (en) Automatic driving vehicle positioning method and device, electronic equipment and storage medium
CN116222586A (en) Fusion positioning method and device for automatic driving vehicle and electronic equipment

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