A high-precision synchronous positioning and mapping method for air-ground collaboration

Through the high-precision synchronous positioning and mapping method of air-ground collaboration, the tightly coupled observation of drones and vehicle-mounted INS solves the problems of reduced positioning accuracy and poor flexibility in the existing technology, and achieves high-precision and high-flexibility positioning and image creation effect.

CN114923494BActive Publication Date: 2024-11-29SOUTHEAST UNIV
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202210532663.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-05-13
Publication Date
2024-11-29
Estimated Expiration
2042-05-13

AI Technical Summary

Technical Problem

The existing synchronous positioning and mapping methods fail when facing a semi-occluded or fully occluded environment, and rely on loopback detection of repeated motions to reduce cumulative errors, resulting in reduced positioning accuracy and poor working flexibility.

Method used

Using the high-precision synchronous positioning and mapping method of air-ground collaboration, the drone is equipped with a wireless sensing module as a 'dynamic base station' and conducts tight coupling observation with the on-board INS, providing accurate absolute positioning to assist SLAM loopback and reducing cumulative errors , and improves the flexibility of SLAM work.

Benefits of technology

Effectively suppress the impact of cumulative errors, improve positioning accuracy and flexibility, break the constraints of relying on fixed paths, and promote the further implementation of SLAM technology.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114923494B_ABST
    Figure CN114923494B_ABST
Patent Text Reader

Abstract

The present invention discloses a high-precision synchronous positioning and mapping method for air-ground collaboration. First, an unmanned aerial vehicle system and a ground unmanned vehicle system are constructed. Then, the position of the ground unmanned vehicle is recursively calculated in real time using the synchronous positioning and mapping method. Finally, air-ground collaborative multi-sensor tightly coupled observation is used to provide loop optimization to reduce cumulative errors. The air-ground collaborative high-precision synchronous positioning and mapping method disclosed by the present invention effectively overcomes the defects of the existing method that it is easily affected by cumulative errors and has an inflexible working mode.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of ground unmanned vehicle positioning, and in particular relates to a high-precision synchronous positioning and mapping method for air-ground collaboration. Background Art

[0002] The development of cutting-edge technologies such as information fusion and artificial intelligence has promoted the intelligent upgrade of ground unmanned vehicles represented by self-driving cars and robots, and has gradually enabled ground unmanned vehicles to be applied to industries such as post-disaster rescue, smart transportation and even military detection, playing an important role in improving safety, efficiency, energy conservation and emission reduction. Among them, achieving accurate global positioning is an important part of the intelligence of ground unmanned vehicles and is the basis for its intelligent functions such as path planning and decision-making control. Although the Global Navigation Satellite System (GNSS) can already achieve accurate positioning of ground unmanned vehicles in unobstructed conditions, it will fail when it faces a semi-obstructed or fully obstructed environment due to the inability to receive enough satellite signals. In addition, although the Inertial Navigation System (INS) is not affected by the environment and can provide location information in a short time, its cumulative error will increase rapidly over time and lead to failure.

[0003] Simultaneous Localization and Mapping (SLAM) methods can be divided into visual SLAM and lidar SLAM according to the different sensors used. They can achieve continuous, accurate and reliable positioning when GNSS fails. The detailed principle can be found in the reference (Ji Zhang, Sanjiv Singh. Low-drift and real-time lidar SLAM [J]. Autonomous Robots, 2017.). However, both visual SLAM and lidar SLAM will produce cumulative errors that grow over time due to the incremental positioning method. The loop detection method in SLAM can be used to reduce or even eliminate the impact of cumulative errors, but loop detection relies on repeated movement in the actual physical space, that is, it must return to a certain place that has been passed before, which brings great constraints to the practical application of SLAM. Therefore, in practical applications, the existing simultaneous localization and mapping methods either have the defect of reducing the positioning and mapping accuracy due to cumulative errors, or have the problem of poor working flexibility due to relying on fixed path loops. Summary of the invention

[0004] The present invention proposes a high-precision synchronous positioning and mapping method based on air-ground collaboration, which effectively solves the problem that the existing synchronous positioning and mapping methods rely on repeated movements in the actual physical space to form a loop to reduce the cumulative error. Both visual SLAM and lidar SLAM can be used, and its specific features are: (1) High precision. The UAV is equipped with a wireless sensor module as a "dynamic base station" to enrich the observation values ​​of visible stars, and then tightly coupled with the vehicle-mounted INS for observation, and then provides accurate absolute position to assist SLAM loop, thereby overcoming the influence of cumulative errors, especially cumulative errors at large scales; (2) High flexibility. The wireless auxiliary observation source does not need to be set in advance, and the position of the UAV as a "dynamic base station" can be obtained in real time, which greatly improves the flexibility of SLAM work.

[0005] The high-precision synchronous positioning and mapping method for air-ground collaboration proposed in the present invention can effectively suppress the influence of cumulative errors, break the limitation of poor flexibility caused by existing methods relying on fixed paths to reduce cumulative errors, and can effectively promote the further implementation and application of SLAM technology.

[0006] The idea of ​​the present invention is further described below:

[0007] Step 1: Build UAV system and ground unmanned vehicle system

[0008] Sub-step 1: Build the drone system

[0009] In the present invention, the sensors installed on the drone include but are not limited to: a GNSS receiver, a wireless signal transmitter and a communication module, wherein the GNSS receiver is used to receive satellite signals to determine the drone's own position, the wireless signal transmitter is used to measure the distance between the drone itself and the ground unmanned system, and the communication module is used to transmit the drone's own position information to the ground unmanned vehicle;

[0010] Sub-step 2: Build a ground unmanned vehicle system

[0011] In the present invention, the sensors installed on the ground unmanned vehicle include but are not limited to: a GNSS receiver, an INS, a visual sensor or a laser radar sensor, a wireless signal receiver, and a communication module. Among them, the GNSS receiver is used to receive satellite signals, the INS is used to provide acceleration and angular velocity information, the visual or laser radar is used for SLAM positioning, the wireless signal receiver is used to receive wireless signals transmitted by the drone, and the communication module is used to obtain the location information of the drone;

[0012] Step 2: Use SLAM to perform position recursion

[0013] The position of the ground unmanned vehicle is obtained by recursion using visual SLAM or lidar SLAM. The position of the ground unmanned vehicle at time k-1 is (xk-1 ,y k-1 ,z k-1 ), the displacement obtained by the SLAM method at time k is Δx k ,Δy k ,Δz k , then the ground unmanned vehicle is obtained through SLAM at time k The location is:

[0014]

[0015] Step 3: Using global position loop closure optimization

[0016] However, the position obtained by SLAM will have a certain cumulative error, which needs to be optimized using the idea of ​​loop closure. The specific steps are as follows:

[0017] Sub-step 1: Develop an optimization strategy

[0018] In order to reduce the computing pressure of the computing unit on the ground unmanned vehicle, the loop mechanism is triggered only when certain conditions are met. In the present invention, the premise of the loop is that the number of visible satellites plus the number of drones is greater than a certain threshold. The loop strategy is as follows:

[0019] (1) When the unmanned vehicle on the ground operates for more than 60 seconds;

[0020] (2) When the operating distance of the unmanned ground vehicle exceeds 200 meters;

[0021] (3) When the number of visible stars drops sharply;

[0022] Under the premise of the loopback of the present invention, when one or more of the above conditions are met, the loopback is triggered and the evaluation starts again according to the above conditions;

[0023] Sub-step 2: Constructing a tightly coupled observation system between visible satellites and inertial navigation

[0024] When there are enough visible satellites, the tightly coupled observation equation of visible satellite / INS is constructed to solve the global position (x k ,y k ,z k ). Its tightly coupled observation equation can be found in the reference (Wang Xinlong. SINS / GPS integrated navigation technology [M]. Beijing University of Aeronautics and Astronautics Press, 2015.);

[0025] Sub-step 3: Build airborne wireless sensor module / visible satellite / inertial navigation tightly coupled observation

[0026] If the number of visible stars is insufficient, the airborne UAV is used as a supplementary information source to establish the following tightly coupled observation equation:

[0027]

[0028] in, represents the position of the satellite at time k, the superscript m represents the number of the satellite, the subscript k represents the time, and Δt represents the satellite clock error. represents the position of the drone at time k, (x k ,y k ,z k ) represents the position of the ground unmanned vehicle at time k. The wireless sensor module on the unmanned vehicle synchronizes time through the network, thereby calculating the position of the ground unmanned vehicle;

[0029] Sub-step 4: Loop Optimization Using Factor Graph

[0030] Under the loopback premise of the present invention, the time of triggering the loopback is recorded as p, and the global position calculated by the above steps is X p , the last time the loop was triggered was at time q, and its position was at X q , adding a loop factor between the two, the loss function is:

[0031] f Loop (X p ,X q )=d(Zh(X p ,X q ))#(6)

[0032] Among them, Z represents the loop factor observation, that is, the difference between the tightly coupled observation results at two moments, and h() is the observation equation. The factor graph is used to optimize this interval. The detailed principle of the factor graph can be found in the reference (Ma Xiaoshuang, Liu Xixiang, Zhang Tongwei, et al. AUV multi-sensor integrated navigation algorithm based on factor graph [J]. Journal of Chinese Inertial Technology, 2019, 027(004): 454-459.), thereby reducing or even eliminating the cumulative error and improving the accuracy of positioning and mapping without repeating the past path.

[0033] Compared with the prior art, the beneficial effects of the present invention are: breaking the defect of the existing simultaneous positioning and mapping method that relies on repeated path movement to reduce cumulative errors, greatly improving the accuracy and flexibility, and strongly promoting the further development of the simultaneous positioning and mapping method. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a method framework diagram; DETAILED DESCRIPTION

[0035] The development of cutting-edge technologies such as information fusion and artificial intelligence has promoted the intelligent upgrade of ground unmanned vehicles represented by self-driving cars and robots, and has gradually enabled ground unmanned vehicles to be applied to industries such as post-disaster rescue, smart transportation and even military detection, playing an important role in improving safety, efficiency, energy conservation and emission reduction. Among them, achieving accurate global positioning is an important part of the intelligence of ground unmanned vehicles and is the basis for its intelligent functions such as path planning and decision-making control. Although the Global Navigation Satellite System (GNSS) can already achieve accurate positioning of ground unmanned vehicles in unobstructed conditions, it will fail when it faces a semi-obstructed or fully obstructed environment due to the inability to receive enough satellite signals. In addition, although the Inertial Navigation System (INS) is not affected by the environment and can provide location information in a short time, its cumulative error will increase rapidly over time and lead to failure.

[0036] The simultaneous localization and mapping (SLAM) method can be divided into visual SLAM and lidar SLAM according to the different sensors used. It can achieve continuous, accurate and reliable positioning when GNSS fails. The detailed principle can be found in the reference (Ji Zhang, Sanjiv Singh. Low-drift and real-time lidar SLAM [J]. Autonomous Robots, 2017.). However, both visual SLAM and lidar SLAM will produce cumulative errors that grow over time due to the incremental positioning method. The loop detection method in SLAM can be used to reduce or even eliminate the impact of cumulative errors, but loop detection relies on repeated movement in the actual physical space, that is, it must return to a certain place that has been passed before, which brings great constraints to the practical application of SLAM. Therefore, in practical applications, the existing simultaneous localization and mapping methods either have the defect of reducing the positioning and mapping accuracy due to cumulative errors, or have the problem of poor working flexibility due to fixed path loops.

[0037] In view of the above problems, the present invention proposes a high-precision synchronous positioning and mapping method based on air-ground collaboration, which effectively solves the problem that the existing synchronous positioning and mapping methods rely on repeated movements in the actual physical space to form loops to reduce cumulative errors. Both visual SLAM and lidar SLAM can be used, and its specific features are: (1) High precision. The UAV is equipped with a wireless sensor module as a "dynamic base station" to enrich the observation values ​​of visible stars, and then tightly coupled with the vehicle-mounted INS for observation, and then provides accurate absolute position to assist SLAM loop, thereby overcoming the influence of cumulative errors, especially cumulative errors at large scales; (2) High flexibility. The wireless auxiliary observation source does not need to be set in advance, and the position of the UAV as a "dynamic base station" can be obtained in real time, which greatly improves the flexibility of SLAM work.

[0038] The high-precision synchronous positioning and mapping method for air-ground collaboration proposed in the present invention can effectively suppress the influence of cumulative errors, break the limitation of poor flexibility caused by existing methods relying on fixed paths to reduce cumulative errors, and can effectively promote the further implementation and application of SLAM technology.

[0039] The idea of ​​the present invention is further described below:

[0040] Step 1: Build UAV system and ground unmanned vehicle system

[0041] Sub-step 1: Build the drone system

[0042] In the present invention, the sensors installed on the drone include but are not limited to: a GNSS receiver, a wireless signal transmitter and a communication module, wherein the GNSS receiver is used to receive satellite signals to determine the drone's own position, the wireless signal transmitter is used to measure the distance between the drone itself and the ground unmanned system, and the communication module is used to transmit the drone's own position information to the ground unmanned vehicle;

[0043] Sub-step 2: Build a ground unmanned vehicle system

[0044] In the present invention, the sensors installed on the ground unmanned vehicle include but are not limited to: a GNSS receiver, an INS, a visual sensor or a laser radar sensor, a wireless signal receiver, and a communication module. Among them, the GNSS receiver is used to receive satellite signals, the INS is used to provide acceleration and angular velocity information, the visual or laser radar is used for SLAM positioning, the wireless signal receiver is used to receive wireless signals transmitted by the drone, and the communication module is used to obtain the location information of the drone;

[0045] Step 2: Use SLAM to perform position recursion

[0046] The position of the ground unmanned vehicle is obtained by recursion using visual SLAM or lidar SLAM. The position of the ground unmanned vehicle at time k-1 is (x k-1 ,y k-1 ,z k-1 ), the displacement obtained by the SLAM method at time k is Δx k ,Δy k ,Δz k , then the ground unmanned vehicle is obtained through SLAM at time k The location is:

[0047]

[0048] Step 3: Using global position loop closure optimization

[0049] However, the position obtained by SLAM will have a certain cumulative error, which needs to be optimized using the idea of ​​loop closure. The specific steps are as follows:

[0050] Sub-step 1: Develop an optimization strategy

[0051] In order to reduce the computing pressure of the computing unit on the ground unmanned vehicle, the loop mechanism is triggered only when certain conditions are met. In the present invention, the premise of the loop is that the number of visible satellites plus the number of drones is greater than a certain threshold. The loop strategy is as follows:

[0052] (1) When the unmanned vehicle on the ground operates for more than 60 seconds;

[0053] (2) When the operating distance of the unmanned ground vehicle exceeds 200 meters;

[0054] (3) When the number of visible stars drops sharply;

[0055] Under the premise of the loopback of the present invention, when one or more of the above conditions are met, the loopback is triggered and the evaluation starts again according to the above conditions;

[0056] Sub-step 2: Constructing a tightly coupled observation system between visible satellites and inertial navigation

[0057] When there are enough visible satellites, the tightly coupled observation equation of visible satellite / INS is constructed to solve the global position (x k ,y k ,z k ). Its tightly coupled observation equation can be found in the reference (Wang Xinlong. SINS / GPS integrated navigation technology [M]. Beijing University of Aeronautics and Astronautics Press, 2015.);

[0058] Sub-step 3: Build airborne wireless sensor module / visible satellite / inertial navigation tightly coupled observation

[0059] If the number of visible stars is insufficient, the airborne UAV is used as a supplementary information source to establish the following tightly coupled observation equation:

[0060]

[0061] in, represents the position of the satellite at time k, the superscript m represents the number of the satellite, the subscript k represents the time, and Δt represents the satellite clock error. represents the position of the drone at time k, (x k ,y k ,z k ) represents the position of the ground unmanned vehicle at time k. The wireless sensor module on the unmanned vehicle synchronizes time through the network, thereby calculating the position of the ground unmanned vehicle;

[0062] Sub-step 4: Loop Optimization Using Factor Graph

[0063] Under the loopback premise of the present invention, the time of triggering the loopback is recorded as p, and the global position calculated by the above steps is X p , the last time the loop was triggered was at q, and its position was X q , adding a loop factor between the two, the loss function is:

[0064] f Loop (X p ,X q )=d(Zh(X p ,X q ))#(9)

[0065] Among them, Z represents the loop factor observation, that is, the difference between the tightly coupled observation results at two moments, and h() is the observation equation. The factor graph is used to optimize this interval. The detailed principle of the factor graph can be found in the reference (Ma Xiaoshuang, Liu Xixiang, Zhang Tongwei, et al. AUV multi-sensor integrated navigation algorithm based on factor graph [J]. Journal of Chinese Inertial Technology, 2019, 027(004): 454-459.), thereby reducing or even eliminating the cumulative error and improving the accuracy of positioning and mapping without repeating the past path.

Claims

1. A high-precision synchronous positioning and mapping method for air-ground collaboration, characterized in that: The specific steps include: Step 1: Build UAV system and ground unmanned vehicle system Sub-step 1: Build the drone system The sensors installed on the drone include but are not limited to: GNSS receiver, wireless signal transmitter and communication module. The GNSS receiver is used to receive satellite signals to determine the drone's own position, the wireless signal transmitter is used to measure the distance between the drone itself and the ground unmanned system, and the communication module is used to transmit the drone's own position information to the ground unmanned vehicle; Sub-step 2: Build a ground unmanned vehicle system The sensors installed on the ground unmanned vehicle include but are not limited to: GNSS receiver, INS, visual sensor or laser radar sensor, wireless signal receiver and communication module, wherein the GNSS receiver is used to receive satellite signals, the INS is used to provide acceleration and angular velocity information, the visual or laser radar is used for SLAM positioning, the wireless signal receiver is used to receive wireless signals transmitted by the drone, and the communication module is used to obtain the location information of the drone; Step 2: Use SLAM to perform position recursion The position of the ground unmanned vehicle is obtained by recursion through visual SLAM or lidar SLAM. The position of the ground unmanned vehicle at time k-1 is (x k-1 ,y k-1 ,z k-1 ), the displacement obtained by the SLAM method at time k is Δx k ,Δy k ,Δz k , then the ground unmanned vehicle is obtained through SLAM at time k The location is: Step 3: Using global position loop closure optimization The position obtained by SLAM will have a certain cumulative error, which needs to be optimized using the idea of ​​loop closure. The specific steps are as follows: Sub-step 1: Develop an optimization strategy In order to reduce the computing pressure of the computing unit on the ground unmanned vehicle, the loopback mechanism is triggered only when certain conditions are met. The premise of the loopback is that the number of visible satellites plus the number of drones is greater than a certain threshold. The loopback strategy is as follows: (1) When the unmanned vehicle on the ground operates for more than 60 seconds; (2) When the operating distance of the unmanned ground vehicle exceeds 200 meters; (3) When the number of visible stars drops sharply; Under the premise of loopback, when one or more of the above conditions are met, the loopback is triggered and the evaluation starts again according to the above conditions; Sub-step 2: Constructing a tightly coupled observation system between visible satellites and inertial navigation When there are enough visible satellites, the tightly coupled observation equation of visible satellite / INS is constructed to solve the global position (x k ,y k ,z k ); Sub-step 3: Build airborne wireless sensor module / visible satellite / inertial navigation tightly coupled observation If the number of visible stars is insufficient, the airborne UAV is used as a supplementary information source to establish the following tightly coupled observation equation: in, represents the position of the satellite at time k, the superscript m represents the number of the satellite, the subscript k represents the time, and Δt represents the satellite clock error. represents the position of the drone at time k, (x k ,y k ,z k ) represents the position of the ground unmanned vehicle at time k. The wireless sensor module on the unmanned vehicle synchronizes time through the network, thereby calculating the position of the ground unmanned vehicle; Sub-step 4: Loop Optimization Using Factor Graph Under the premise of loop closure, the time of triggering loop closure is recorded as p, and the global position calculated by the above steps is X p , the last time the loop was triggered was at q, and its position was X q , adding a loop factor between the two, the loss function is: f Loop (X p ,X q )=d(Z-h(X p ,X q ))#(3) Among them, Z represents the loop factor observation, that is, the difference between the tightly coupled observation results at two moments, h() is the observation equation, and the factor graph is used to optimize this interval to reduce or even eliminate the cumulative error, thereby improving the accuracy of positioning and mapping without repeating the past path.

Citation Information

Patent Citations

  • Factor graph-based vehicle robust positioning method in urban canyon environment

    CN113237482A

  • Robot instant localization and mapping method and system based on multiple information sources

    CN113432600A

  • Air-ground cooperative vehicle positioning and orienting method

    CN113971697A

  • Laser radar SLAM method and system based on local structure information extension

    CN114280626A

  • Drone localization

    US10429487B1