Unmanned aerial vehicle for underground detection and bionic autonomous positioning and navigation method thereof

Abstract

The invention provides an unmanned aerial vehicle for underground detection and a bionic autonomous positioning and navigation method thereof, and belongs to the technical field of mine robot inspection and rescue. The technical problems that in the prior art, an unmanned aerial vehicle is poor in maneuverability, difficult in electric energy supplement and inaccurate in positioning and navigation are solved. According to the technical scheme, the explosion-proof unmanned aerial vehicle comprises an explosion-proof type rack, an explosion-proof type speed-increasing power device, an intrinsic safety type sensing device, an intrinsic safety type navigation device and an explosion-proof type rapid docking and charging device, the navigation device is used for navigation of the unmanned aerial vehicle, and the sensing device is used for sensing of the unmanned aerial vehicle; the explosion-proof type rapid docking and charging device is used for enabling the unmanned aerial vehicle to rapidly land on a charging carrier and to be safely charged when multiple robots work cooperatively in a well, and the explosion-proof type speed increasing power device is used for increasing the rotating speed output by a motor of the unmanned aerial vehicle. The robot can rapidly land on other robots and supplement electric energy, and meanwhile long-distance accurate positioning and navigation can be executed.

Description

Technical Field

[0001] The present invention relates to the technical field of mine robot inspection and rescue, and particularly to an underground detection unmanned aerial vehicle and a bionic autonomous positioning and navigation method thereof. Background Art

[0002] When an unmanned aerial vehicle (UAV) operates underground, it faces the following difficulties. First, the UAV needs to be explosion-proofed before underground operation. The heavy flameproof enclosure will greatly reduce the load-carrying capacity and maneuverability of the UAV, and may even prevent the UAV from taking off. Second, in the task of multi-robot collaborative operation underground, the UAV needs to land on other robots in time for charging after a long time of operation. However, current UAVs are difficult to quickly land, dock on other robots and charge. Finally, before the UAV performs long-distance precise navigation in extreme environments such as mine roadways, it often needs to use the Simultaneous Localization and Mapping (SLAM) technology to build a high-precision map. However, this method will consume a large amount of computing and power resources of the UAV, and there will be large positioning errors, trajectory drifts, global map consistency failures, and even map building failures after a long time of operation. In addition, due to a large amount of computing and power resources being consumed in SLAM, the UAV is often difficult to undertake additional high-level tasks such as inspection, monitoring, and human-machine interaction at the body end.

[0003] Therefore, in view of the problems that the UAV has a decreased maneuverability, is difficult to quickly replenish electrical energy, and is difficult to autonomously perform high-precision positioning and long-distance precise navigation in the underground extreme environment, the present invention proposes an underground detection UAV and a bionic autonomous positioning and navigation method thereof to solve the above technical problems. Summary of the Invention

[0004] The purpose of the present invention is to provide an underground detection UAV and a bionic autonomous positioning and navigation method thereof. The UAV has high maneuverability, can quickly land on other robots for electrical energy replenishment in the task of multi-robot collaborative operation underground, and can perform long-distance precise positioning and navigation through the method in a narrow and confined space. In addition, the present invention also provides a high-precision positioning method. The UAV senses fixed equipment in the mine through multiple sensors and measures the distance, and then inversely calculates the relative position of the UAV with respect to the equipment. Coarse positioning of the absolute position between nodes is performed based on the semantic-topological map, and then mileage estimation is performed based on the data recorded by the inertial measurement unit and the previously obtained relative position information, so as to complete the precise positioning of the absolute position.

[0005] To achieve the above-mentioned invention object, the technical solution adopted by the present invention is specifically as follows: An underground exploration unmanned aerial vehicle, comprising an explosion-proof frame, an explosion-proof speed increasing power device, an intrinsically safe sensing device, an intrinsically safe navigation device, and an explosion-proof quick docking and charging device;

[0006] The navigation device is arranged inside the explosion-proof frame and is used for the navigation of the unmanned aerial vehicle. The sensing device is installed below the explosion-proof frame and is used for the sensing of the unmanned aerial vehicle. The explosion-proof quick docking and charging device is arranged below the explosion-proof frame and is used for the unmanned aerial vehicle to quickly land on the charging carrier and safely charge during the collaborative operation of multiple robots underground. A plurality of explosion-proof speed increasing power devices are provided and are evenly distributed around the explosion-proof frame, and are used for increasing the rotational speed output by the motor of the unmanned aerial vehicle to increase the lift of the unmanned aerial vehicle.

[0007] Furthermore, the explosion-proof frame includes an explosion-proof fuselage, explosion-proof arms, propeller collision prevention frames, an on-board computer, and a power supply battery. The explosion-proof arms are installed on the explosion-proof fuselage. The propeller collision prevention frames are installed on the explosion-proof arms. The on-board computer and the power supply battery are installed inside the explosion-proof fuselage;

[0008] The navigation device includes an inertial measurement unit and a millimeter wave altimeter, and the navigation device is installed inside the explosion-proof fuselage.

[0009] Furthermore, the explosion-proof speed increasing power device includes a brushless motor, planet gears, a planet carrier, a sun gear, a ring gear, a sealing cover, an end cover, bearings, a propeller, and a bottom plate. The bottom plate is installed below the explosion-proof machine. The brushless motor is installed above the bottom plate and is located inside the explosion-proof arm. The planet carrier is installed above the output shaft of the brushless motor. A plurality of planet gears are provided. The planet gears and the sun gear are both installed above the planet carrier. The bearing is installed on the output shaft of the sun gear. The ring gear is installed above the explosion-proof arm and its interior is engaged with the planet gears. The sealing cover is installed above the ring gear. The end cover is installed above the sealing cover and is used for sealing the brushless motor and the plurality of planet gears inside the explosion-proof arm. The propeller is installed on the output shaft of the sun gear and is located above the end cover.

[0010] Furthermore, the sensing device includes an explosion-proof housing, a visible light camera, an infrared camera, a lidar, and a pan-tilt head. The sensing device is installed below the explosion-proof fuselage. The visible light camera is installed on a first-layer platform inside the explosion-proof housing. The infrared camera is installed on a second-layer platform inside the explosion-proof housing. The lidar is installed on a third-layer platform inside the explosion-proof housing. The pan-tilt head is connected to the interface outside the explosion-proof housing by screws;

[0011] The explosion-proof quick docking and charging device includes an explosion-proof housing, clamping plates, clamping plate springs, a protective cover, a charging female head, a female head spring, and a charging male head disposed on a robot carrier. A charging jack is vertically provided at the bottom of the explosion-proof housing. Mounting holes are respectively provided on opposite side walls of the charging jack. Ramps are provided at the bottoms of the sides of the two clamping plates close to each other. The clamping plate springs are sleeved on the clamping plates and are installed together with the clamping plates on the horizontal mounting holes of the explosion-proof housing. The protective cover is installed outside the horizontal mounting holes of the explosion-proof housing. The female head spring is sleeved on the charging female head and is disposed together with the charging female head at the top of the charging jack;

[0012] The lower end of the charging male head is connected to an electric lead screw on the robot carrier through a vertical rod and a base. An upper stopper is fixedly sleeved on the upper end of the vertical rod. A lower stopper is slidably sleeved on the outer surface of the vertical rod below the upper stopper. The upper stopper is of a frustum structure. The lower stopper is of an inverted frustum structure. The lower stopper can be fitted to the bottom wall of the upper stopper after moving upward;

[0013] The present invention also provides a bionic autonomous positioning and navigation method, including the following steps:

[0014] S1. The unmanned aerial vehicle continuously flies and collects visible light source images, infrared source images, and point cloud data through a sensing device. The visible light source images are used to judge the scene illumination intensity, and the flight altitude is maintained through a millimeter-wave altimeter;

[0015] S2. The unmanned aerial vehicle selects the sensing mode of the sensing device according to the scene illumination intensity, performs target detection on the underground fixed equipment, marks semantic tags and marks them as road signs;

[0016] S3. The unmanned aerial vehicle constructs a semantic-topological map, uses the trajectory data recorded by the inertial measurement unit as the side lines, and records all the semantic information of the scene containing road signs through the sensing device and uses it as the nodes;

[0017] S4. The unmanned aerial vehicle uses the semantic-topological map for navigation, detects the node road signs through the sensing device, generates a navigation vector pointing to the node, and extracts the side line data between the current position and the node in the semantic-topological map to move;

[0018] S5. After the unmanned aerial vehicle arrives at the node, the cumulative error of the inertial measurement unit is cleared, and step S4 is repeatedly executed and moved to the next node until the navigation is completed;

[0019] S6. The unmanned aerial vehicle measures the distance to the road signs through the sensing device and inversely calculates its relative position to the road signs, and combines the data recorded by the inertial measurement unit and the semantic-topological map to clarify its absolute position.

[0020] Further, in step S1, the judgment of the scene illumination intensity specifically includes the following steps:

[0021] S11. The drone captures the front scene through a visible light camera, determines the light intensity in the environment, and selects the sensing mode of the sensing device according to the light intensity, which is divided into three types: single infrared camera and lidar sensing, dual-camera fusion and lidar sensing, and single visible light camera and lidar sensing.

[0022] S12. Select the sensing mode of the sensing device through the high-light ratio P. The sensing mode determination method is as follows:

[0023]

[0024] In step S12, the calculation method of the high-light ratio P is as follows:

[0025] S121. After using the Retinex model to separate the illumination component L(x, y) that reflects the light source intensity and distribution and the reflection component R(x, y) that reflects the surface characteristics of the scene, extract the illumination component L(x, y). The light source position and intensity can be intuitively presented through the high-light area of the illumination component.

[0026] S122. Crop the image so that its width and height are both N. Divide the image into an n×n grid, with the size of each area being (N / n)×(N / n) pixels. Calculate the mean value T of the gray values for each area to obtain the local brightness distribution, and judge the pixel T in the image through the mean value T i whether it is a low-brightness pixel T L and a high-brightness pixel T H， The determination method is as follows:

[0027]

[0028] S123. Statistically count the total amount I of high-brightness pixels Htotal , calculate the total number I of high-brightness pixels Htotal and the ratio P of the total number I of high-brightness pixels to the total number of pixels I

[0029]

[0030] Furthermore, the specific steps of step S2 are as follows:

[0031] S21. Convert the visible light camera coordinate system, infrared camera coordinate system, and lidar coordinate system to the same world coordinate system through a transformation matrix, and select the drone sensing method according to the environmental light intensity.

[0032] S22. The drone obtains the visual information in the scene through visible light images, infrared images, or visible light-infrared fusion images, and uses an object detection algorithm to judge whether there are fixed mine equipment in the image.

[0033] S23. When the algorithm detects fixed mine equipment, it marks it as a road sign and assigns semantic tags to the road sign.

[0034] Further, in step S3, the construction of the semantic-topological map includes the following steps:

[0035] S31. When the drone completes step S2 at a certain location, the drone uses the sensing device to record the complete scene information of this location and creates a node. This node includes the number and category of road signs in the scene and the spatial relationship of each road sign in this scene.

[0036] S32. After the drone establishes a node of the topological map, the drone moves forward and continuously searches for fixed mine equipment with significant features in the new scene through the sensing device. When a new road sign is found, the drone repeats step S31 and creates a new node.

[0037] S33. The drone records the trajectory information between two nodes through the inertial measurement unit and stores the trajectory information in the edge between the two nodes in the semantic-topological map.

[0038] S34. The drone will keep repeating steps S31, S32, and S33 until the complete mine semantic-topological navigation map is constructed.

[0039] Further, the positioning and navigation model in steps S4 and S5 is as follows:

[0040] Location C, target point S, and landmark L are in three-dimensional space. The distance from location C to landmark L is Rc, the angle between the horizontal direction at C and Rc is φc, the distance from target point S to landmark L is Rs, and the angle between the horizontal direction at S and Rs is φs.

[0041] The projections of location C, target point S, and landmark L on the two-dimensional plane form a triangle LCS. The angle between landmark L and the moving direction Qs of the drone at S is θ, and the angle between landmark L and the moving direction Qc of the drone at C is θ + δ.

[0042] The angles between the side CS of triangle LCS and the moving directions Qs and Qc are both a. The angle of ∠LCS is π - (θ + δ - a), the angle of ∠LSC is θ - a, and the angle of ∠CLS is δ.

[0043] In step S4, the calculation process of the navigation vector is as follows:

[0044] S41. The drone obtains the distances R c 、R s from itself to landmark L at locations C and S through lidar and calculates its length on the projection plane The formula is as follows:

[0045]

[0046] S42. By the sine theorem The ratio with ∠LCS and ∠LSC is as follows:

[0047]

[0048] S4.3. Further derivation gives and The ratio of is as follows:

[0049]

[0050] S44. Further derivation gives the included angle a between CS and the moving direction Q s as follows:

[0051]

[0052] S4.5. Finally, the navigation angle β and the moving vector u are obtained as follows:

[0053] β = π - a

[0054]

[0055] Furthermore, in step S6, the specific steps of measuring the distance to the road sign and inverting its relative position to the road sign are as follows:

[0056] S61. Take the center pixel point I(x, y) of the confidence box of the road sign in the image as the position information of the road sign, and take I(x, y) as the center to extract a circle with a radius of 0.5 times the width of the confidence box;

[0057] S62. Convert the circle into a cone in the lidar coordinate system, filter the point cloud in the cone, and then calculate its average depth R and average angle

[0058] S63. The drone inversely calculates its relative position information to the road sign through the position I(x, y), depth R and angle information of the road sign to complete high-precision relative positioning;

[0059] S64. The drone first performs rough positioning of the absolute position between nodes according to the semantic-topological map, and then performs mileage estimation based on the data recorded by the inertial measurement unit (4-1) and the relative position information obtained in step S61, and further completes fine positioning of the absolute position.

[0060] Compared with the prior art, the beneficial effects of the present invention are:

[0061] 1. The present invention provides an explosion-proof speed-increasing power device, which improves the maneuverability of the unmanned aerial vehicle through a planetary gear train, and solves the problems that the maneuverability and load-carrying capacity of the unmanned aerial vehicle are damaged due to the excessive weight of the explosion-proof shell, and it is even difficult to take off.

[0062] 2. The present invention provides an explosion-proof quick docking and charging device, which overcomes the difficulty that the unmanned aerial vehicle is difficult to quickly land on other robots and replenish electric energy in the task of multi-robot cooperative operation underground.

[0063] 3. The present invention imitates the visual homing mechanism of the genus Cataglyphis in ants, and proposes a high-precision and lightweight positioning and navigation method based on semantic information, which overcomes the problem that it is difficult for the unmanned aerial vehicle to perform long-distance precise positioning and navigation through the SLAM method in a long and narrow restricted space.

[0064] 4. The present invention provides a high-precision positioning method. The unmanned aerial vehicle senses fixed equipment in the mine through multiple sensors and measures the distance, and then inversely calculates the relative position of the unmanned aerial vehicle with respect to the equipment. Coarse positioning of the absolute position between nodes is carried out according to the semantic-topological map, and then mileage estimation is carried out according to the data recorded by the inertial measurement unit and the previously obtained relative position information, so as to complete the precise positioning of the absolute position. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention, and do not constitute a limitation to the present invention.

[0066] Figure 1 It is a detection schematic diagram of the unmanned aerial vehicle of the present invention in a roadway scene.

[0067] Figure 2 It is a mechanism diagram of the bionic insect navigation of the present invention.

[0068] Figure 3 It is a three-dimensional view of the unmanned aerial vehicle of the present invention Figure 1 .

[0069] Figure 4 It is a three-dimensional view of the unmanned aerial vehicle of the present invention Figure 2 .

[0070] Figure 5 It is an external view and an internal view of the explosion-proof speed-increasing power device of the present invention.

[0071] Figure 6 It is a front view of the unmanned aerial vehicle of the present invention.

[0072] Figure 7 It is a structural schematic diagram of the explosion-proof quick docking and charging device of the present invention in the charging state.

[0073] Figure 8 This is a schematic diagram of the structure of the explosion-proof type rapid docking and charging device of the present invention in the uncharged state.

[0074] Figure 9 This is a working view of the explosion-proof type rapid docking and charging device of the present invention.

[0075] Figure 10 This is a flowchart of the positioning and navigation method of the present invention.

[0076] Figure 11 This is a schematic diagram of the calculation method of the high light ratio P of the present invention.

[0077] Figure 12 This is a mathematical model of the bionic positioning and navigation device of the present invention.

[0078] Among them, the reference numerals are:

[0079] 1. Explosion-proof frame; 1-1. Explosion-proof fuselage; 1-2. Explosion-proof arm; 1-3. Propeller anti-collision frame; 1-4. On-board computer; 1-5. Power supply battery;

[0080] 2. Explosion-proof speed increasing power device; 2-1. Brushless motor; 2-2. Planet gear; 2-3. Planet carrier; 2-4. Sun gear; 2-5. Ring gear; 2-6. End cover; 2-7. Bearing; 2-8. Speed increasing output shaft; 2-9. Propeller;

[0081] 3. Sensing device; 3-1. Explosion-proof housing; 3-2. Visible light camera; 3-3. Infrared camera; 3-4. Lidar; 3-5. Pan-tilt head;

[0082] 4. Navigation device; 4-1. Inertial measurement unit; 4-2. Millimeter wave altimeter;

[0083] 5. Explosion-proof type rapid docking and charging device; 5-1. Explosion-proof housing; 5-2. Splint spring; 5-3. Male charging head; 5-4. Female charging head; 5-5. Upper ejector; 5-6. Protective cover; 5-7. Splint; 5-8. Female head spring; 5-9. Charging jack; 5-10. Vertical rod; 5-11. Base; 5-12. Lower ejector;

[0084] 6. Belt conveyor;

[0085] 7. Pipeline;

[0086] 8. Miner's lamp;

[0087] 9. Robot carrier; 9-1. Electric lead screw. Detailed implementation manners

[0088] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Of course, the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0089] Embodiment

[0090] As Figure 1 shown, when the drone detects in the roadway scenario, it detects fixed equipment such as belt conveyors 6, pipelines 7, miner's lamps 8, etc. existing in the roadway through the sensing device 3, and records its own trajectory through the navigation device 4.

[0091] As Figure 2 shown, the visual homing mechanism of the genus Cataglyphis of the present invention uses a semantic-topological map to complete real-time high-precision navigation in the roadway.

[0092] As Figure 3 and Figure 4 shown, an underground inspection drone includes: an explosion-proof frame 1, an explosion-proof speed-increasing power device 2, an intrinsically safe sensing device 3, an intrinsically safe navigation device 4, and an explosion-proof quick docking and charging device 5; the navigation device 4 is arranged inside the explosion-proof frame 1 for drone navigation, the sensing device 3 is installed below the explosion-proof frame 1 for drone sensing, the explosion-proof quick docking and charging device 5 is arranged below the explosion-proof frame 1 for the drone to quickly land on the charging carrier and safely charge during multi-robot collaborative operations underground, and 4 explosion-proof speed-increasing power devices 2 are arranged and evenly distributed around the explosion-proof frame 1 for increasing the rotational speed of the drone motor output to increase the lift of the drone.

[0093] As Figure 3 shown, the explosion-proof frame 1 includes an explosion-proof fuselage 1-1, explosion-proof arms 1-2, propeller collision protection frames 1-3, an on-board computer 1-4, and a power supply battery 1-5. The explosion-proof arms 1-2 are installed on the explosion-proof fuselage 1-1, the propeller collision protection frames 1-3 are installed on the explosion-proof arms 1-2, and the on-board computer 1-4 and the power supply battery 1-5 are installed inside the explosion-proof fuselage 1-1; the navigation device 4 includes an inertial measurement unit 4-1 and a millimeter-wave altimeter 4-2, and the navigation device 4 is installed inside the explosion-proof fuselage 1-1.

[0094] As Figure 5As shown in the figure, the explosion-proof speed-increasing power device 2 includes a brushless motor 2-1, a planetary gear 2-2, a planetary carrier 2-3, a sun gear 2-4, a ring gear 2-5, a sealing cover 2-6, an end cover 2-7, a bearing 2-8, a propeller 2-9 and a base plate 2-10. The base plate 2-10 is installed below the explosion-proof machine arm 1-2, and the brushless motor 2-1 is installed above the base plate 2-10 and inside the explosion-proof machine arm 1-2. The planetary carrier 2-3 is installed above the output shaft of the brushless motor 2-1. The planetary gear 2-2 is installed above the planetary carrier 2-3. The sun gear 2-4 is also installed above the planetary carrier 2-3. The bearing 2-8 is installed on the output shaft of the sun gear 2-4. The ring gear 2-5 is installed above the explosion-proof machine arm 1-2 and is internally engaged with the planetary gear 2-2. The sealing cover 2-6 is installed above the ring gear, and the end cover 2-7 is installed above the sealing cover 2-6 to seal the brushless motor and the planetary gear train inside the explosion-proof machine arm 1-2. The propeller 2-9 is installed on the output shaft of the sun gear 2-4 and above the end cover 2-7.

[0095] As Figure 6 shown, the sensing device 3 includes an explosion-proof housing 3-1 for the sensing device, a visible light camera 3-2, an infrared camera 3-3, a lidar 3-4 and a pan-tilt head 3-5. The sensing device 3 is installed below the explosion-proof fuselage 1-1. The visible light camera 3-2 is installed on a platform on the first floor inside the explosion-proof housing 3-1. The infrared camera 3-3 is installed on a platform on the second floor inside the explosion-proof housing 3-1. The lidar 3-4 is installed on a platform on the third floor inside the explosion-proof housing 3-1. The pan-tilt head 3-5 is connected to the interface outside the explosion-proof housing 3-1 of the sensing device by screws.

[0096] As Figure 7 and Figure 8 shown, the explosion-proof quick docking and charging device 5 includes an explosion-proof housing 5-1, a clamping plate 5-7, a clamping plate spring 5-2, a protective cover 5-6, a charging female head 5-4, a female head spring 5-8 and a charging male head 5-3 installed on the robot carrier 9. A charging jack 5-9 is vertically provided at the bottom of the explosion-proof housing 5-1. Mounting holes are respectively provided on the opposite side walls of the charging jack 5-9. Ramps are provided at the bottom of the sides of the two clamping plates 5-7 close to each other. The clamping plate spring 5-2 is sleeved on the clamping plate 5-7 and is installed together with the clamping plate 5-7 in the horizontal mounting holes of the explosion-proof housing 5-1. The protective cover 5-6 is installed outside the horizontal mounting holes of the explosion-proof housing 5-1. The female head spring 5-8 is sleeved on the charging female head 5-4 and is arranged together with the charging female head 5-4 at the top of the charging jack 5-9. The charging male head 5-3 is installed inside the robot carrier 9. When the drone lands on the robot carrier 9 for charging, the explosion-proof quick docking and charging device 5, the drone and the robot carrier 9 together form an explosion-proof structure.

[0097] As Figure 9As shown in the figure, the working process of the explosion-proof type quick docking and charging device 5 in the present invention is as follows:

[0098] In the first step, as Figure 9 (A) shows, when the unmanned aerial vehicle lands on the robot carrier 9 through the explosion-proof type quick docking and charging device 5, the male charging head 5-3 enters the charging jack 5-9. After the motor inside the robot carrier 9 starts, the electric lead screw 9-1 will push the male charging head 5-3 into the explosion-proof type quick docking and charging device 5. When the male charging head 5-3 moves upward, it will push the two clamping plates 5-7 to both sides along the bottom slope of the clamping plate 5-7 through the upper ejector 5-5. After being pushed open, the male charging head 5-3 enters the charging bin and continues to move upward until the male charging head 5-3 is pushed into the female charging head 5-4. The female head spring 5-8 installed on the female charging head 5-4 can provide buffering during docking to protect the female charging head 5-4 from damage, thereby ensuring the normal connection of the male charging head 5-3; the side of the two clamping plates 5-7 close to each other is an arc structure and is adapted to the side wall of the vertical rod 5-10. During this charging insertion process, after the upper ejector 5-5 passes through the two clamping plates 5-7, due to the restoring force of the clamping plate spring 5-2, the two clamping plates 5-7 can move towards each other and approach for reset. Since the upper ejector 5-5 is a frustum structure, the bottom of the upper ejector 5-5 will be stuck on the upper side of the clamping plate 5-7 after passing through the clamping plate 5-7, thereby ensuring the stability and safety of charging.

[0099] In the second step, as Figure 9 (B) shows, when the unmanned aerial vehicle finishes charging, start the motor to make the lead screw 9-1 continue to push the male charging head 5-3 until the lower ejector 5-12 pushes open the clamping plate 5-7 and enters the charging bin. And due to the action of the clamping plate 5-7, the two ejectors can be attached to each other up and down. During this process, the female head spring 5-8 on the female charging head 5-4 will be compressed to ensure that the female charging head 5-4 will not be damaged due to extrusion.

[0100] In the third step, as Figure 9 (C) shows, the lead screw 9-1 moves downward and pulls out the male charging head 5-3. During this process, since the lower ejector 5-12 is a reverse frustum structure, it can easily push open the two side clamping plates 5-7. Since the upper and lower ejectors are attached to each other, the upper ejector 5-5 will also follow the lower ejector 5-12 through the pushed-open clamping plate 5-7 until the male charging head 5-3 is completely pulled out of the charging bin and recovered into the robot carrier 9. During this process, the female charging head 5-4 will be blocked by the reset clamping plate 5-7 in the charging bin and will not be taken out of the charging bin.

[0101] As Figure 10 shown, the present invention also provides a method for autonomous positioning and navigation of an underground unmanned aerial vehicle with a bionic insect homing mechanism, including the following steps:

[0102] S1. The drone continuously flies and collects visible light source images, infrared source images, and point cloud data through the sensing device 3. It uses the visible light source images to judge the scene illumination intensity and maintains the flight altitude through the millimeter-wave altimeter 4-2.

[0103] S2. The drone selects the sensing mode of the sensing device 3 according to the scene illumination intensity, conducts target detection on the underground fixed equipment, marks semantic tags and labels them as road signs.

[0104] S3. The drone constructs a semantic-topological map, uses the trajectory data recorded by the inertial measurement unit 4-1 as the side lines, and records all the semantic information of the scene containing road signs through the sensing device 3 and uses it as the nodes.

[0105] S4. The drone uses the semantic-topological map for navigation, detects the node road signs through the sensing device 3, generates a navigation vector pointing to the node, and extracts the side line data between the current position and the node in the semantic-topological map to move.

[0106] S5. After the drone arrives at the node, it clears the cumulative error of the inertial measurement unit 4-1, and repeats step S6 to move to the next node until the navigation is completed.

[0107] S6. The drone measures the distance to the road sign through the sensing device 3 and inversely calculates its relative position to the road sign, and combines the data recorded by the inertial measurement unit 4-1 and the semantic-topological map to determine its absolute position.

[0108] As Figure 11 shown, the scene illumination intensity judgment steps in step S1 include:

[0109] S11. When the drone takes pictures of the front scene through the visible light camera and judges the illumination intensity in the environment, it selects the sensing mode of the sensing device 3 according to the illumination intensity, which is divided into three types: single infrared camera and lidar sensing, dual camera fusion and lidar sensing, and single visible light camera and lidar sensing.

[0110] S12. Select the sensing mode of the sensing device 3 through the highlight ratio P. The sensing mode determination method is as follows:

[0111]

[0112] As Figure 11 shown, the calculation method of the highlight ratio P in step S12 is as follows:

[0113] S121. After using the Retinex model to separate the illumination component L(x, y) that reflects the light source intensity and distribution and the reflection component R(x, y) that reflects the surface characteristics of the scene, extract the illumination component L(x, y). The light source position and intensity can be intuitively presented through the highlight area of the illumination component.

[0114] S122. Crop the image so that both its width and height are N. Divide the image into an n×n grid, where each region has a size of (N / n)×(N / n) pixels. Calculate the mean value T of the grayscale values for each region to obtain the local brightness distribution. Determine the pixels T in the image based on the mean value T i whether they are low-brightness pixels T L and high-brightness pixels T H , and the determination method is as follows:

[0115]

[0116] S123. Statistically count the total amount I of high-brightness pixels Htotal , and calculate the total number I of high-brightness pixels Htotal and the ratio P of the total number I of high-brightness pixels to the total number of pixels I

[0117]

[0118] Step S2 in the present invention specifically includes the following steps:

[0119] S21. Convert the visible light camera coordinate system, the infrared camera coordinate system, and the lidar coordinate system to the same world coordinate system through a transformation matrix, and select the UAV sensing method according to the ambient light intensity;

[0120] S22. The UAV obtains the visual information in the scene through the visible light image, the infrared image, or the visible light-infrared fusion image, and uses the object detection algorithm to determine whether there are significant feature mine fixed equipment such as belt conveyors, miner's lamps, pipelines, and rails in the image;

[0121] S23. When the algorithm detects significant feature mine fixed equipment, mark it as a road sign and assign a semantic label to the road sign.

[0122] In step S3, the detailed steps for constructing the topological map include:

[0123] S31. When the UAV completes step S62 at a certain location, the UAV will use the sensing device 3 to record the complete scene information at this location and create a node, which includes the number of road signs, categories, and the spatial relationships of each road sign in the scene, etc.;

[0124] S32. After the UAV establishes a node of the topological map, the UAV will move forward and continuously search for significant feature mine fixed equipment in the new scene through the sensing device 3. When a new road sign is found, the UAV will repeat step S101 and create a new node;

[0125] S33. The drone records the trajectory information between two nodes through the inertial measurement unit 4-1 and stores the trajectory information in the edge between the two nodes in the semantic-topological map.

[0126] S34. The drone will keep repeating steps S31, S32, and S33 until a complete mine semantic-topological navigation map is constructed.

[0127] As Figure 12 shown, the positioning and navigation model in steps S4 and S5 can be expressed as:

[0128] Location C, target point S, and landmark L are in three-dimensional space. The distance from location C to landmark L is R c , and the angle between the horizontal direction at C and R c is The distance from target point S to landmark L is R s , and the angle between the horizontal direction at S and R s is

[0129] The projections of location C, target point S, and landmark L on the two-dimensional plane form a triangle LCS. The angle between landmark L and the moving direction Q s at S of the drone is θ, and the angle between landmark L and the moving direction Q c at C of the drone is θ + δ.

[0130] The angle between side CS of triangle LCS and the moving directions Q s , Q c is both a. The angle of ∠LCS is π - (θ + δ - a), the angle of ∠LSC is θ - a, and the angle of ∠CLS is δ.

[0131] As Figure 12 shown, the calculation process of the navigation vector in step S4 is represented as follows:

[0132] S41. The drone obtains the distances R c , R s from itself to landmark L at locations C and S through lidar, and the angles to calculate its length on the projection plane The formula is as follows:

[0133]

[0134] S42. Through the sine theorem and the ratios of ∠LCS and ∠LSC, the formula is as follows:

[0135]

[0136] S43. Further derivation gives The ratio with is as follows:

[0137]

[0138] S44, further derive CS and the movement direction Q s The included angle a is as follows:

[0139]

[0140] S45, finally obtain the navigation angle β and the movement vector u, as follows:

[0141] β = π - a

[0142]

[0143] As Figure 10 and Figure 12 shown, the specific steps for distance measurement of the road sign and inversion of its relative position to the road sign in step S6 are as follows:

[0144] S61, use the central pixel point I(x, y) of the confidence box of the road sign in the image as the position information of the road sign, and take I(x, y) as the center to extract a circle with a radius of 0.5 times the width of the confidence box;

[0145] S62, convert the circle into a cone in the lidar coordinate system, filter the point cloud within the cone, and calculate its average depth R and average angle

[0146] S63, the drone inversely calculates its relative position information to the road sign through information such as the position I(x, y), depth R, and angle of the road sign to complete high-precision relative positioning;

[0147] S64, the drone first performs rough positioning of the absolute position between nodes according to the semantic-topological map, and then performs mileage estimation based on the data recorded by the inertial measurement unit (4-1) and the relative position information obtained in step S61, and further completes fine positioning of the absolute position.

[0148] In summary, the present invention provides an explosion-proof speed-increasing power device. This device improves the maneuverability of the drone through a planetary gear train, solving the problem that its maneuverability is impaired or even difficult to take off due to the excessive weight of the explosion-proof housing. Similarly, the present invention also provides an explosion-proof rapid docking and charging device 5, which solves the difficulty of rapid docking and charging of drones in the task of multi-robot collaborative operation underground. In addition, the present invention also imitates the homing mechanism of the genus Cataglyphis in the ant family, realizing lightweight and high-precision autonomous navigation of underground drones under the condition of limited computing resources, and overcoming the problem that it is difficult for underground drones to accurately perform long-distance navigation through non-SLAM methods. Finally, the present invention also uses a large number of fixed devices with known positions in the mine to provide accurate relative orientation information for the drone, and constructs a low-cost and high-precision positioning method for underground drones through the constructed semantic-topological map and the mileage information recorded by the inertial measurement unit.

[0149] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An unmanned aerial vehicle for underground detection, characterized in that It includes an explosion-proof frame (1), an explosion-proof speed-increasing power device (2), an intrinsically safe sensing device (3), an intrinsically safe navigation device (4), and an explosion-proof quick docking and charging device (5); The navigation device (4) is arranged inside the explosion-proof frame (1) for UAV navigation. The sensing device (3) is installed below the explosion-proof frame (1) for UAV sensing. The explosion-proof quick docking and charging device (5) is arranged below the explosion-proof frame (1) for the UAV to quickly land on the charging carrier and safely charge during underground multi-robot collaborative operations. A plurality of explosion-proof speed-increasing power devices (2) are provided and evenly distributed around the explosion-proof frame (1) to increase the rotational speed output by the UAV motor and thus increase the lift of the UAV.

2. The downhole detection UAV according to claim 1, wherein, The explosion-proof frame (1) includes an explosion-proof fuselage (1-1), explosion-proof arms (1-2), propeller anti-collision frames (1-3), an on-board computer (1-4), and a power supply battery (1-5). The explosion-proof arms (1-2) are installed on the explosion-proof fuselage (1-1). The propeller anti-collision frames (1-3) are installed on the explosion-proof arms (1-2). The on-board computer (1-4) and the power supply battery (1-5) are installed inside the explosion-proof fuselage (1-1); The navigation device (4) includes an inertial measurement unit (4-1) and a millimeter-wave altimeter (4-2), and the navigation device (4) is installed inside the explosion-proof fuselage (1-1).

3. The downhole detection drone according to claim 2, characterized in that, The explosion-proof speed-increasing power device (2) includes a brushless motor (2-1), planet gears (2-2), a planet carrier (2-3), a sun gear (2-4), a ring gear (2-5), a sealing cover (2-6), an end cover (2-7), a bearing (2-8), a propeller (2-9), and a base plate (2-10). The base plate (2-10) is installed below the explosion-proof arm (1-2). The brushless motor (2-1) is installed above the base plate (2-10) and inside the explosion-proof arm (1-2). The planet carrier (2-3) is installed above the output shaft of the brushless motor (2-1). A plurality of planet gears (2-2) are provided. The planet gears (2-2) and the sun gear (2-4) are both installed above the planet carrier (2-3). The bearing (2-8) is installed on the output shaft of the sun gear (2-4). The ring gear (2-5) is installed above the explosion-proof arm (1-2) and its interior is engaged with the planet gears (2-2). The sealing cover (2-6) is installed above the ring gear (2-5). The end cover (2-7) is installed above the sealing cover (2-6) to seal the brushless motor (2-1) and the plurality of planet gears (2-2) inside the explosion-proof arm (1-2). The propeller (2-9) is installed on the output shaft of the sun gear (2-4) and above the end cover (2-7).

4. The downhole detection UAV according to claim 2, wherein, The perception device (3) includes an explosion-proof housing (3-1), a visible light camera (3-2), an infrared camera (3-3), a lidar (3-4) and a pan-tilt head (3-5). The perception device (3) is installed below the flameproof fuselage (1-1). The visible light camera (3-2) is installed on a first-layer platform inside the explosion-proof housing (3-1). The infrared camera (3-3) is installed on a second-layer platform inside the explosion-proof housing (3-1). The lidar (3-4) is installed on a third-layer platform inside the explosion-proof housing (3-1). The pan-tilt head (3-5) is connected to an interface outside the explosion-proof housing (3-1) by screws. The flameproof quick docking and charging device (5) includes an explosion-proof housing (5-1), clamping plates (5-7), clamping plate springs (5-2), a protective cover (5-6), a charging female head (5-4), a female head spring (5-8) and a charging male head (5-3) arranged on a robot carrier (9). A charging jack (5-9) is vertically arranged at the bottom of the explosion-proof housing (5-1). Mounting holes are respectively arranged on opposite side walls of the charging jack (5-9). Ramps are arranged at the bottoms of the closer sides of the two clamping plates (5-7). The clamping plate springs (5-2) are sleeved on the clamping plates (5-7) and are installed together with the clamping plates (5-7) in the horizontal mounting holes of the explosion-proof housing (5-1). The protective cover (5-6) is installed outside the horizontal mounting holes of the explosion-proof housing (5-1). The female head spring (5-8) is sleeved on the charging female head (5-4) and is arranged together with the charging female head (5-4) at the top of the charging jack (5-9). The lower end of the charging male head (5-3) is connected to an electric lead screw (9-1) on the robot carrier (9) through a vertical rod (5-10) and a base (5-11). An upper stopper (5-5) is fixedly sleeved on the upper end of the vertical rod (5-10). A lower stopper (5-12) is slidably sleeved on the outer surface of the vertical rod (5-10) below the upper stopper (5-5). The upper stopper (5-5) is of a frustum structure. The lower stopper (5-12) is of an inverted frustum structure. The lower stopper (5-12) can be attached to the bottom wall of the upper stopper (5-5) after moving upward.

5. A bionic autonomous positioning and navigation method, which is based on the downhole detection UAV according to any one of claims 1-4, and is characterized in that, It includes the following steps: S1. The drone continuously flies and collects visible light source images, infrared source images and point cloud data through the perception device (3), judges the scene illumination intensity by using the visible light source images, and maintains the flight altitude through a millimeter wave altimeter (3-3). S2. The drone selects the perception mode of the perception device (3) according to the scene illumination intensity, performs target detection on the underground fixed equipment, marks semantic tags and labels them as road signs. S3. The drone constructs a semantic-topological map, uses the trajectory data recorded by the inertial measurement unit (4-1) as the side lines, and records all the semantic information of the scene containing road signs through the perception device (3) and uses it as the nodes. S4. The drone uses the semantic-topological map for navigation, detects node road signs through the sensing device (3), generates a navigation vector pointing to the node, and extracts the edge line data between the current position and the node in the semantic-topological map to move. S5. After the drone arrives at the node, clear the cumulative error of the inertial measurement unit (4-1), repeat step S4, and move to the next node until the navigation is completed. S6. The drone measures the distance to the road sign through the sensing device (3) and inverses its relative position to the road sign, and determines its absolute position in combination with the data recorded by the inertial measurement unit (4-1) and the semantic-topological map.

6. The bionic autonomous positioning and navigation method according to claim 5, wherein In step S1, the judgment of the scene illumination intensity specifically includes the following steps: S11. The drone takes a picture of the front scene through the visible light camera (3-2) and judges the illumination intensity in the environment. According to the illumination intensity, select the sensing mode of the sensing device (3), which is divided into three types: single infrared camera and lidar sensing, dual camera fusion and lidar sensing, and single visible light camera and lidar sensing. S12. Select the sensing mode of the sensing device (3) through the high-light ratio P. The sensing mode determination method is: In step S12, the calculation method of the high-light ratio P is as follows: S121. After using the Retinex model to separate the illumination component L(x, y) that reflects the light source intensity and distribution and the reflection component R(x, y) that reflects the surface characteristics of the scene, extract the illumination component L(x, y). The light source position and intensity can be intuitively presented through the highlighted area of the illumination component. S122. Crop the image so that its width and height are both N. Divide the image into an n×n grid, where each region has a size of (N / n)×(N / n) pixels. Calculate the mean value T of the grayscale values for each region to obtain the local brightness distribution. Determine the pixels T in the image based on the mean value T i whether they are low-brightness pixels T L and high-brightness pixels T H， The determination method is as follows: S123, count the total number of highlighted pixels I Htotal , calculate the total number of highlighted pixels I Htotal , the ratio P of the total number of highlighted pixels I to the total number of pixels I 7. A bionic autonomous positioning and navigation method according to claim 5, characterized in that The specific steps of step S2 include the following steps: S21. Convert the visible light camera coordinate system, infrared camera coordinate system, and lidar coordinate system to the same world coordinate system through the transformation matrix, and select the drone sensing method according to the environmental illumination intensity. S22. The drone obtains the visual information in the scene through the visible light image, infrared image, or visible light-infrared fusion image, and uses the target detection algorithm to judge whether there are mine fixed devices in the image. S23. When the algorithm detects the mine fixed device, record it as a road sign and assign a semantic label to the road sign.

8. The bionic autonomous positioning and navigation method according to claim 5, characterized in that In step S3, the construction of the semantic-topological map includes the following steps: S31. When the drone completes step S2 at a certain location, the drone will use the sensing device (3) to record the complete scene information of this location and create a node. This node includes the number and category of road signs in the scene and the spatial relationship of each road sign in this scene. S32. After the drone establishes a node of the topological map, the drone will move forward and continuously search for mine fixed devices with significant features in the new scene through the sensing device (3). When a new road sign is found, the drone will repeat step S31 and create a new node. S33. The drone records the trajectory information between two nodes through the inertial measurement unit (4-1) and stores the trajectory information in the edge line between the two nodes in the semantic-topological map. S34. The drone will keep repeating steps S31, S32, and S33 until the complete mine semantic-topological navigation map is constructed.

9. The bionic autonomous positioning and navigation method according to claim 5, wherein The positioning and navigation model in steps S4 and S5 is: Location C, target point S, and landmark L are in three-dimensional space. The distance from location C to landmark L is Rc, the horizontal angle at C with respect to Rc is φc, the distance from target point S to landmark L is Rs, and the horizontal angle at S with respect to Rs is φs; The projections of location C, target point S, and landmark L on the two-dimensional plane form a triangle LCS. The angle between landmark L and the motion direction Qs of the drone at S is θ, and the angle between landmark L and the motion direction Qc of the drone at C is θ + δ; The angles between side CS of triangle LCS and the motion directions Qs and Qc are both a, the angle of ∠LCS is π - (θ + δ - a), the angle of ∠LSC is θ - a, and the angle of ∠CLS is δ. In step S4, the calculation process of the navigation vector is as follows: S41, the drone obtains the distances R from it to the landmark L at locations C and S through lidar c 、R s and the angle to calculate its length on the projection plane The formula is as follows: S42, by the sine theorem The ratio with ∠LCS and ∠LSC is as follows: S4.3, further derive to obtain and The ratio is as follows: S44, further derive the included angle a between CS and the moving direction Q, and the formula is as follows: s ​ S45. Finally, the navigation angle β and the movement vector u are obtained, and the formula is as follows: β＝π-a 10. A bionic autonomous positioning and navigation method according to claim 9, characterized in that, In step S6, the specific steps for measuring the distance to the road sign and inverting its relative position with respect to the road sign are as follows: S61. Use the center pixel point I(x, y) of the confidence box of the road sign in the image as the position information of the road sign, and take I(x, y) as the center to extract a circle with a radius of 0.5 times the width of the confidence box; S62. Convert the circle into a cone in the lidar coordinate system, filter the point cloud within the cone, and then calculate its average depth R and average angle S63. The drone uses the position I(x, y), depth R, and angle of the road sign to invert the relative position information of itself with respect to the road sign, completing high-precision relative positioning; S64. The drone first performs a rough positioning of the absolute positions between nodes according to the semantic-topological map, and then estimates the mileage based on the data recorded by the inertial measurement unit (4-1) and the relative position information obtained in step S61, and further completes the fine positioning of the absolute position.

Images