Unmanned aerial vehicle indoor and outdoor seamless connection system based on external active visual guidance and cooperative control method

Abstract

The invention discloses an unmanned aerial vehicle indoor and outdoor seamless connection system based on external active visual guidance and a cooperative control method, and solves the problems of hard light failure, coordinate switching jump and unreliable communication in the prior art. The system comprises an airborne subsystem and a ground guide subsystem, an airborne end is provided with a specific wavelength active cursor and a dual-mode receiving module, and a ground end comprises a visual capture unit with a narrow-band optical filter, a dual-mode transmitting unit and a central processing unit. According to the control method, through outdoor hovering, visual capturing and filtering, coordinate handshaking and dual-mode link grading guiding, stable positioning in a strong light environment, smooth coordinate switching and millimeter-level connection precision are achieved, and indoor and outdoor seamless safe connection of the unmanned aerial vehicle is guaranteed.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) navigation, specifically relating to an indoor-outdoor seamless docking system and collaborative control method for UAVs based on external active vision guidance. Background Technology

[0002] Current indoor-outdoor drone docking technology mainly relies on onboard sensors for autonomous positioning (Inside-Out). It completes positioning and docking using the drone's own sensors without external auxiliary equipment. The core is to achieve a closed loop of "autonomous environmental perception-positioning-navigation" through sensor data fusion. Its main technological branches include GPS + inertial navigation fusion technology, LiDAR + SLAM technology, and visual SLAM technology.

[0003] In GPS+inertial navigation fusion technology, outdoor locations rely on GPS to obtain latitude and longitude coordinates. When indoor GPS signals are missing, the system switches to an onboard IMU (inertial measurement unit) to estimate the location, and in some cases, it is combined with a barometer to assist in altitude positioning.

[0004] In LiDAR+SLAM technology, an airborne LiDAR scans the environment to generate a 3D point cloud map, which is then used for real-time localization using the SLAM (Simultaneous Localization and Mapping) algorithm, without relying on external signals.

[0005] In visual SLAM technology, images are acquired using airborne monocular / binocular cameras, and environmental maps are constructed and located through feature point matching. In some cases, depth information is obtained by combining RGB-D cameras to improve accuracy.

[0006] Another common approach is based on external passive visual guidance technology. This involves using external fixed equipment (such as a camera) to capture "passive markers" (without active light emission) on the drone to achieve positioning and docking guidance, which falls under the "Outside-In mode." Its main branches include infrared reflective ball motion capture technology and visual tag recognition technology.

[0007] In infrared reflective ball motion capture technology, infrared reflective balls are attached to the fuselage of the drone, and multiple infrared cameras are set up indoors. The three-dimensional coordinates of the drone are calculated by capturing the reflected signals of the reflective balls. Typical examples include NOKOV measurement and OptiTrack system.

[0008] In visual tag recognition technology, drones are equipped with downward-facing cameras, and visual tags such as QR codes, color blocks, and line segments are placed on the ground / walls. The drones are positioned by recognizing the location of the tags.

[0009] There are also non-visual fusion positioning technologies that do not rely on optical signals but integrate technologies such as radio frequency, 5G, and BeiDou to solve connection problems in extreme environments (such as pitch-black indoor spaces or areas with strong electromagnetic interference).

[0010] However, existing technologies have some significant drawbacks: The risk of failure in strong light environments arises when a drone enters an indoor space through a window from the outside, where there is strong natural backlight. Existing passive reflector technology is easily overwhelmed by strong background light, leading to loss of visual tracking.

[0011] The risk of abrupt changes during coordinate system switching is significant. Outdoor navigation is based on the GPS coordinate system (latitude and longitude), while indoor navigation is based on a local Cartesian coordinate system. Current technology lacks an effective coordinate system handshake mechanism. When a drone hard switches from GPS mode to indoor mode, the origins of the two coordinate systems do not coincide, which can easily cause position data jumps, leading to the drone losing control or crashing into walls.

[0012] The communication link is singular and unreliable. Traditional solutions typically use only a single communication link to send commands, which cannot meet the different needs of "long-distance data transmission" and "short-distance high-frequency real-time control". Summary of the Invention

[0013] This invention proposes a seamless indoor-outdoor docking system and collaborative control method for unmanned aerial vehicles (UAVs) based on external active vision guidance. The system includes the following components: (1) Airborne subsystems, including active cursor unit, dual-mode receiver module and airborne flight control.

[0014] The active cursor unit is a specific wavelength (e.g., 940nm) infrared LED module installed on the top of the fuselage, serving as the sole passive target tracking unit.

[0015] The dual-mode receiving module includes a first receiver (for receiving WiFi / data transmission data) and a second receiver (for receiving 2.4G analog remote control signals).

[0016] The airborne flight controller has GPS positioning capabilities and an open external control interface.

[0017] (2) Ground guidance subsystem, including anti-light vision capture unit, dual-mode transmission unit and central processing unit.

[0018] The anti-light vision capture unit consists of a fixedly installed industrial camera. Its key feature is that a narrowband filter with a center wavelength that matches the onboard LED is installed in front of the lens to filter out non-narrowband wavelength light.

[0019] The principle behind this is based on spectral selectivity. Sunlight is full-spectrum, while LED light is narrow-band. The filter only allows the LED wavelength to pass through, greatly improving the signal-to-noise ratio and enabling the system to operate stably under backlight conditions.

[0020] The dual-mode transmission unit includes a data transmission module (sending ROS messages) and an analog remote control transmission module (sending PWM / SBUS signals).

[0021] The central processing unit runs the positioning calculation and process control program.

[0022] The collaborative control method includes the following steps: S1. Outdoor Approach and Hovering: The drone uses its onboard GPS navigation to fly to the preset coordinates outside the window and hovers, waiting for instructions from the ground system.

[0023] S2. Visual capture and physical filtering: The drone enters the field of view of the ground camera. Due to the narrow-band filter, the image sensor can only sense light of a specific wavelength emitted by the onboard LED. The bright background of the window is filtered into a dark color, and the system quickly identifies and locks onto the drone cursor.

[0024] S3. Coordinate system air handshake.

[0025] To prevent the drone from jumping erratically during the switchover, the system executes the following alignment logic: S3.1 At the moment of acquisition t0, the system reads the GPS coordinates currently transmitted back by the UAV (the converted local coordinates P_gps_t0).

[0026] S3.2 Simultaneously obtain the indoor coordinates P_vis_t0 calculated by the vision system.

[0027] S3.3 Calculate the initial offset, which is expressed as follows: ; S3.4 Establishing a virtual coordinate system: At all subsequent times t, the ground system sends the target coordinates to the UAV. All are based on visual coordinates plus this bias: ; The principle is to deceive the drone into believing it is still flying in its original GPS coordinate system, when in fact its relative displacement is entirely controlled by the indoor vision system, thus achieving a "zero-jump" switch. By utilizing the invariance of relative coordinates, instead of forcibly modifying the drone's internal coordinate system, a fixed "initial deviation value" is superimposed to allow external guidance commands to adapt to the drone's current state, eliminating control abrupt changes caused by coordinate system inconsistencies.

[0028] S4. Communication link switching and hierarchical guidance, mainly including indoor approach and precise landing.

[0029] The indoor approach uses a first communication link (data link), where the ground system sends the target position coordinates (Position Setpoint) with an offset via ROS Topic. The UAV flight controller treats this as an external GPS signal and uses its own algorithm to plan a path to the charging dock.

[0030] The precise landing refers to the system switching to the second communication link (remote control link) when the visual judgment indicates that the drone is above the charging dock and the error is less than a preset value. The ground system directly sends low-level attitude (Roll / Pitch) and throttle (Throttle) analog signals through the NRF module, bypassing the navigation logic of the airborne flight control system, and directly controlling the motors to perform a precise vertical landing.

[0031] Compared with the prior art, the beneficial effects of the present invention include: With strong anti-interference capabilities, compared to traditional motion capture, this system can still achieve stable positioning under strong natural light interference through the hardware combination of "active light emission + narrowband filtering".

[0032] The transition is smooth and safe, and the unique coordinate handshake algorithm eliminates the risk of jitter when switching from GPS to indoor positioning.

[0033] With high control precision, it combines the flexibility of ROS navigation (Phase 1) with the directness of analog remote control (Phase 2), ensuring efficiency in the docking process and millimeter-level accuracy in the final landing. Attached Figure Description

[0034] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0035] Figure 1 This is a diagram of the overall hardware architecture of the system according to the method of the present invention.

[0036] Figure 2 This is a flowchart illustrating the coordinate system handshake logic of the method of the present invention.

[0037] Figure 3 This is a schematic diagram of coordinate system handshaking and transformation in the method of the present invention.

[0038] Figure 4 This is a flowchart illustrating the logic of the dual-mode hierarchical guidance and landing phase of the method of the present invention. Detailed Implementation

[0039] A warehousing and logistics center needs to achieve seamless connection between the outdoor drone material receiving point and the indoor storage shelves to automate the transfer of small spare parts. The scenario presents challenges such as window backlighting, direct midday sun illuminance at windows ≥50000 lux, lack of indoor GPS signal, and connection accuracy requirements of ±5mm. This invention's system and method achieve safe and efficient connection.

[0040] According to the present invention, a seamless indoor-outdoor docking system and collaborative control method for unmanned aerial vehicles based on external active vision guidance are described below, and the overall hardware architecture diagram of the system is as follows. Figure 1 As shown, Figure 1 (1) is a dual-mode receiving unit, (2) is a drone, (3) is an active cursor unit, (4) is a filtering element, (5) is an anti-light visual capture unit, and (6) is a dual-mode transmitting unit.

[0041] The active cursor unit in the airborne subsystem uses an Everlight 940nm infrared LED module, which is installed at the center of the top of the UAV fuselage to ensure that it can be captured from all angles. The dual-mode receiver module uses a WiFi data transmission receiver + a 2.4G analog remote control receiver to receive data link and remote control link signals respectively; The airborne flight controller uses the DJI N3, which supports GPS positioning and an open external control interface, and is compatible with ROS message reception. The drone platform uses the DJI M300RTK, equipped with the aforementioned modules, to meet both outdoor hovering stability and indoor load requirements.

[0042] The anti-glare visual capture unit in the ground guidance subsystem uses a Hikvision industrial camera + 940nm narrowband filter, which is fixedly installed on the upper inner and outer sides of the window, covering the drone hovering area to the docking point; The dual-mode transmitter unit uses a ROS data transmission module + a 2.4G analog remote control transmission module to transmit coordinate data and PWM signals; The central processing unit uses an Intel NUC12 Pro mini-PC, running Ubuntu 20.04 and ROSNoetic, and deploys location calculation and process control programs.

[0043] The coordinate system handshake logic flow for control using the aforementioned collaborative control method is as follows: Figure 2 As shown, the steps are as follows: S1. Outdoor Approach and Hovering: The drone uses its onboard GPS navigation to fly to the preset coordinates outside the window and hovers, waiting for instructions from the ground system.

[0044] The drone uses its onboard GPS for navigation to fly to the preset coordinates at the outdoor window of the warehouse center. It hovers at a height of 1.5m above the window and sends a "ready" signal to the ground system after its attitude stabilizes, waiting for instructions.

[0045] S2. Visual capture and physical filtering: The drone enters the field of view of the ground camera. Due to the narrow-band filter, the image sensor can only sense light of a specific wavelength emitted by the onboard LED. The bright background of the window is filtered into a dark color, and the system quickly identifies and locks onto the drone cursor.

[0046] The ground camera initiates data acquisition. Due to the narrowband filter, which only allows 940nm infrared light to pass through, the bright background at the window is filtered into a dark color, with only the onboard LED cursor appearing bright. The system uses the OpenCV image recognition algorithm to lock the cursor within 300ms. After continuously capturing 20 stable frames, visual acquisition is considered successful.

[0047] S3. Coordinate system handshake in mid-air, the coordinate system handshake and transformation are as follows: Figure 3 As shown, to prevent the drone from jumping erratically during the switching process, the system executes the following alignment logic: S3.1 At the moment of acquisition t0, the system reads the GPS coordinates currently transmitted back by the UAV (the converted local coordinates P_gps_t0).

[0048] S3.2 Simultaneously obtain the indoor coordinates P_vis_t0 calculated by the vision system.

[0049] S3.3 Calculate the initial offset, which is expressed as follows: ; S3.4 Establishing a virtual coordinate system: At all subsequent times t, the ground system sends the target coordinates to the UAV. All are based on visual coordinates plus this bias: ; The principle is to deceive the drone into believing it is still flying in its original GPS coordinate system, when in fact its relative displacement is entirely controlled by the indoor vision system, thus achieving a "zero-jump" switch. By utilizing the invariance of relative coordinates, instead of forcibly modifying the drone's internal coordinate system, a fixed "initial deviation value" is superimposed to allow external guidance commands to adapt to the drone's current state, eliminating control abrupt changes caused by coordinate system inconsistencies.

[0050] S4. Communication link switching and hierarchical guidance, mainly including indoor approach and precise landing, the flowchart of which is as follows: Figure 4 As shown.

[0051] The indoor approach uses a first communication link (data link), where the ground system sends the target position coordinates (Position Setpoint) with an offset via ROS Topic. The UAV flight controller treats this as an external GPS signal and uses its own algorithm to plan a path to the charging dock.

[0052] When approaching the indoor charging dock, the ground system sends the target coordinates with an offset via ROSTopic through a WiFi link. The drone's flight controller interprets this as a GPS signal and plans a path to fly to the indoor charging dock.

[0053] The precise landing refers to the system switching to the second communication link (remote control link) when the visual judgment indicates that the drone is above the charging dock and the error is less than a preset value. The ground system directly sends low-level attitude (Roll / Pitch) and throttle (Throttle) analog signals through the NRF module, bypassing the navigation logic of the airborne flight control system, and directly controlling the motors to perform a precise vertical landing.

[0054] When the vision system detects that the drone's coordinate error is ≤3mm, it switches to the 2.4G remote control link and sends Roll / Pitch attitude signals (to maintain horizontality) and Throttle throttle signals through the NRF module to directly control the motors to achieve vertical landing. The final docking error is ±2mm, completing a seamless docking.

[0055] This embodiment is based on the indoor and outdoor docking scenario of drones in a warehousing and logistics center. It strictly follows the core steps of the invention, namely outdoor hovering, visual capture and filtering, coordinate handshake, and dual-mode link hierarchical guidance. It uses mature industrial-grade hardware to build the system and successfully solves the problems of strong light interference, coordinate switching jumps, and communication reliability. Finally, it achieves millimeter-level docking accuracy, which verifies the stability, practicality, and accuracy of the invention in practical applications. It provides a feasible reference solution for seamless drone docking in similar scenarios.

Claims

1. A seamless indoor / outdoor docking system and collaborative control method for unmanned aerial vehicles (UAVs) based on external active vision guidance, characterized in that, The system includes: Airborne subsystems and ground guidance subsystems; The airborne subsystem includes an active cursor unit, a dual-mode receiver module, and an airborne flight control system. The ground guidance subsystem includes an anti-light visual capture unit, a dual-mode transmission unit, and a central processing unit.

2. The seamless indoor / outdoor docking system and collaborative control method for unmanned aerial vehicles (UAVs) based on external active vision guidance as described in claim 1, characterized in that, The active cursor unit is a specific wavelength infrared LED module installed on the top of the fuselage, serving as the sole passive target tracking unit; The dual-mode receiving module includes a first receiver and a second receiver. The first receiver is used to receive WiFi / data transmission data, and the second receiver is used to receive 2.4G analog remote control signals. The airborne flight controller has GPS positioning capabilities and an open external control interface.

3. The seamless indoor / outdoor docking system and collaborative control method for unmanned aerial vehicles (UAVs) based on external active vision guidance as described in claim 1, characterized in that, The anti-light vision capture unit consists of a fixedly installed industrial camera. The key feature is the addition of a narrowband filter in front of the lens, whose center wavelength matches that of the onboard LED, filtering out non-narrowband wavelengths of light.

4. The seamless indoor / outdoor docking system and collaborative control method for unmanned aerial vehicles (UAVs) based on external active vision guidance as described in claim 1, characterized in that, The dual-mode transmission unit includes a data transmission module and an analog remote control transmission module; The data transmission module is used to send ROS messages; The simulated remote control transmitter module is used to send PWM / SBUS signals.

5. The collaborative control method of any of the UAV indoor / outdoor seamless docking systems based on external active vision guidance as described in claims 1-4 includes the following steps: S1. Outdoor approach and hovering: The drone uses its onboard GPS navigation to fly to the preset coordinates outside the window and hovers, waiting for instructions from the ground system. S2. Visual capture and physical filtering: the drone enters the field of view of the ground camera; S3. Coordinate system air handshake; S4. Communication link switching and hierarchical guidance, including indoor approach and precise landing.

6. The seamless indoor / outdoor docking system and collaborative control method for unmanned aerial vehicles (UAVs) based on external active vision guidance as described in claim 1, characterized in that, The central processing unit is equipped with positioning calculation algorithms and process control logic. It is used to complete visual coordinate calculation, initial offset calculation, virtual coordinate system construction, communication link switching judgment and guidance command generation. It is the core control unit for achieving stable positioning, smooth coordinate switching and hierarchical guidance control in strong light environment.

7. The seamless indoor / outdoor docking system and collaborative control method for unmanned aerial vehicles (UAVs) based on external active vision guidance as described in claim 5, characterized in that, Step S3 includes the following sub-steps: S3.1 At the moment of acquisition t0, the system reads the GPS coordinates (converted local coordinates P_gps_t0) currently transmitted back by the UAV. S3.2 Simultaneously obtain the indoor coordinates P_vis_t0 calculated by the vision system; S3.3 Calculate the initial offset, which is expressed as follows: ； S3.4 Establishing a virtual coordinate system: At all subsequent times t, the ground system sends the target coordinates to the UAV. All are based on visual coordinates plus this bias: ； The drone is tricked into thinking it is still flying in its original GPS coordinate system, when in fact its relative displacement is completely controlled by the indoor vision system, thus achieving a "zero jump" switch.

8. The seamless indoor / outdoor docking system and collaborative control method for unmanned aerial vehicles (UAVs) based on external active vision guidance as described in claim 7, characterized in that, The principle of zero jump is to utilize the invariance of relative coordinates, without forcibly modifying the internal coordinate system of the UAV, but by superimposing a fixed initial deviation value, allowing external guidance commands to adapt to the current state of the UAV, thus eliminating control abrupt changes caused by inconsistencies in the coordinate system.

9. A seamless indoor / outdoor docking system and collaborative control method for unmanned aerial vehicles (UAVs) based on external active vision guidance, as described in claim 5, is characterized in that... In step S4, the indoor approach uses the first communication link, i.e., the data link. The ground system sends the target position coordinates with an offset through ROS Topic. The UAV flight controller treats this as an external GPS signal and uses its own algorithm to plan a path to fly to the charging dock.

10. A seamless indoor / outdoor docking system and collaborative control method for unmanned aerial vehicles (UAVs) based on external active vision guidance, as described in claim 5, is characterized in that... The precise landing refers to the system switching to the second communication link, i.e. the remote control link, when the visual judgment indicates that the drone is above the charging dock and the error is less than a preset value. The ground system directly sends the underlying attitude and throttle simulation signals through the NRF module, bypassing the navigation logic of the airborne flight control, and directly controls the motors to perform a precise vertical landing.

Images