A global self-adaptive application multi-sensing detection large model driven communication operation and maintenance robot

Abstract

The application provides a global self-adaptive application multi-sensing detection large model driving communication operation and maintenance robot, and relates to the technical field of communication operation and maintenance. The driving mechanism comprises an outer frame, an embedded frame, a connecting frame, a placing frame, a driver, a driving wheel, a driving rotating wheel, an auxiliary support, a connecting support, a mounting plate, a slope frame, a collision-proof roller, a binding subrope and a slope guiding frame. The driving mechanism realizes adaptive movement on complex terrain. The driving wheel adopts a Mecanum wheel structure with a diameter of 250 mm. The 24V DC brushless motor of the driver and the parallelogram linkage mechanism of the adjusting support can realize 360° turning in place, lateral movement and 15° slope climbing. The triangular truss structure of the slope frame, the spring shock absorber of the collision-proof roller and the inclination sensor of the slope guiding frame improve the stability and passability of the robot in complex environments such as indoor machine room tiles, outdoor cement ground and mountain gravel road.

Description

Technical Field

[0001] This invention relates to the field of communication operation and maintenance technology, and in particular to a large-scale model-driven communication operation and maintenance robot with global adaptive multi-sensor detection. Background Technology

[0002] In the field of telecommunications maintenance, with the continuous development of intelligent and automated technologies, robots have gradually become indispensable equipment in maintenance, inspection, and repair work. Existing telecommunications maintenance robots typically possess basic functions such as movement, detection, and performance of maintenance tasks, and collect data and are remotely controlled through sensors and communication modules. With technological advancements, robots need to possess stronger autonomy and adaptability to cope with complex and ever-changing environments and work scenarios. Integrating multi-sensor technology, large-model-driven algorithms, precise positioning technology, and efficient communication modules has become a key direction for improving robot maintenance capabilities. Existing telecommunications maintenance robots generally rely on a single sensor for environmental perception and typically lack intelligent scene recognition and adaptive adjustment functions.

[0003] Existing communication and maintenance robots suffer from several shortcomings. Their sensor configurations are mostly single or low-precision, resulting in limited perception capabilities in complex environments and an inability to acquire multi-dimensional environmental data in real-time and accurately. Existing robots often rely on single laser or infrared sensors for obstacle detection and localization, lacking sufficient redundancy and fusion mechanisms, making them prone to positioning errors and obstacle recognition failures in dynamic or complex environments. They typically lack strong adaptive capabilities, failing to automatically adjust their working modes according to different scenarios (such as indoor server rooms, outdoor courtyards, or mountainous areas), leading to performance degradation in complex terrains. Existing communication systems are mostly single-mode transmissions, unable to guarantee high stability and low latency in changing environments, easily causing remote control interruptions and information loss. The hardware architecture and algorithm models of existing technologies are mostly simplified, failing to fully utilize large-model-driven intelligent algorithms for scene recognition and task decision-making, making it difficult to achieve autonomous optimization and precise control of complex tasks. Existing robots lack optimization in power management and battery life, often failing to meet the requirements for stable operation in long-term, low-temperature, or extreme environments, limiting their application in certain special working environments. Therefore, we provide a large-scale model driven by multi-sensor detection with global adaptive capabilities to drive a communication maintenance robot. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: including a driving mechanism, a detection mechanism, and a maintenance mechanism. The driving mechanism includes an outer frame, an inner frame, a connecting frame, a placement frame, a driver, a driving wheel, a driving wheel, an auxiliary bracket, a connecting bracket, a mounting plate, a slope frame, anti-collision rollers, binding ropes, a slope guide frame, and an adjustment bracket. The outer frame is made of 6061-T6 aluminum alloy and forms a double-layer protective structure with the inner frame through laser welding. The connecting frame is an I-shaped connector made of carbon fiber composite material and is rigidly fixed to the outer frame and the inner frame through M8 high-strength bolts. The detection mechanism integrates a detection substrate, energy storage battery, main working box, limit card frame, detection base frame, infrared detection eye, laser detection eye, spare detection head, signal transceiver, external frame, side limit frame and limit pin. The detection substrate is made of aerospace-grade magnesium alloy plate and the surface is covered with a 0.2mm thick graphene thermal conductive film. The maintenance mechanism is equipped with an extension plate, a maintenance base plate, a support frame, a drive cavity, a drive motor, a rotator, a rotary motor, a main adjustment frame, an auxiliary adjustment frame, a derivative frame, a three-piece frame, a detection frame, a construction probe, a construction cylinder, a gripper frame, a gripper telescopic device, a gripper socket rotating frame, and a gripper. The expansion board is connected to the maintenance base plate through a tenon and mortise structure, and the joint is filled with nitrile rubber sealing gaskets. The robot is equipped with an adaptive hardware architecture consisting of a heterogeneous computing unit, a multimodal sensor interface and an adaptive power management system, as well as an artificial intelligence large model scene recognition algorithm deployed in the main working box. It achieves full-domain autonomous operation and maintenance, centimeter-level precise positioning and 99.99% communication stability through a 5G-A and LoRa dual-mode communication module.

[0006] In a preferred embodiment, the drive wheel adopts a 250mm diameter Mecanum wheel structure, with the wheel body forged from 7075 aluminum alloy and covered with a polyurethane tread with a Shore hardness of 75. The tread has a 3mm deep diamond-shaped anti-slip pattern. The drive rotor is a 2.5 module helical gear structure, connected to the driver output shaft via a keyway. The driver uses a 24V DC brushless motor with a rated power of 350W and a peak torque of 12N·m, and is equipped with a 1024-line encoder to achieve closed-loop control. The adjustment bracket includes two sets of parallelogram linkage mechanisms, connected via a DS3218 servo motor. The drive system has an adjustable stroke range of ±15° and a response time of ≤0.1s. Φ12mm positioning pin holes are provided at both ends of the connecting frame to form a clearance fit with the positioning boss of the outer frame, with a fit accuracy of H7 / g6. The placement frame has an internal heat dissipation duct and forced air cooling via an axial fan, with an operating temperature range of -40℃ to +70℃. This drive system can achieve 360° in-situ turning (turning angular velocity ≥60° / s), lateral movement (speed ≥0.8m / s), and 15° slope climbing. It adapts to complex terrains such as indoor computer room floor tiles, outdoor cement floors, and mountain gravel roads through a multi-round differential speed algorithm.

[0007] In a preferred embodiment, the transceiver integrates a 5G-A and LoRa dual-mode communication module. The 5G-A module adopts the 3GPP Release 18 standard, supports the n257 / n258 millimeter-wave band, achieves a peak rate of 10Gbps, has a transmission latency of ≤1ms, and is configured with a 4×4 MIMO antenna array. The LoRa module operates in the 433MHz ISM band, uses a spreading factor of SF12, has a maximum transmit power of 20dBm, a receive sensitivity of -148dBm, and supports Class A / C operating modes. The module has a built-in VSWR monitoring circuit that collects reflected power in real time through a directional coupler, with a measurement range of 1.0~3.0 and an accuracy of ±0.05. When the VSWR is >1.5, an antenna switching mechanism is automatically triggered. The communication module adopts a metal shield design, achieves an electromagnetic compatibility level of EN301489-1 / 3, supports IP67 protection, and can operate continuously in environments ranging from -30℃ to +65℃.

[0008] In a preferred embodiment, the outer frame of the detection mechanism is made of 304 stainless steel with a thickness of 2mm, and is formed into a honeycomb hollow structure by laser cutting, reducing the weight by 35% compared to traditional designs; the side limiting frame is an aluminum alloy bracket, fixed to the outer frame by two limiting pins with a diameter of 8mm, and the pin fitting clearance is 0.05mm~0.1mm; the infrared detection eye uses an uncooled focal plane array detector with a resolution of 640×512, thermal sensitivity <50mK, field of view of 45°×34°, and frame rate of 30Hz; the laser detection eye is a 1550nm wavelength fiber laser. The detector has an output power of 5mW, a measurement distance of 0.5m~100m, an accuracy of ±2mm, and a scanning frequency of 50Hz. The backup detector head includes a high-definition industrial camera (2 megapixels, 60fps) and a temperature and humidity sensor (measurement range -40℃~+85℃, humidity 0~100%RH, accuracy ±0.5℃ / ±2%RH). Data from multiple sensors is transmitted to the main control box via gigabit Ethernet. After spatiotemporal registration, a full-area communication heat map with a resolution of 0.5m×0.5m is generated. Color mapping adopts the Jet pseudo-color scheme, with a dynamic range of -120dBm~-30dBm.

[0009] In a preferred embodiment, the extension plate of the maintenance mechanism is made of 7075-T73 aluminum alloy sheet, 10mm thick, with a hard anodized surface; the support frame is a hollow cylindrical structure with an inner diameter of 80mm and a length of 150mm, and the side-mounted drive chamber is designed with an IP65 protection rating; the drive motor is a servo motor with a brake, a rated speed of 3000rpm, an encoder resolution of 17 bits, and drives the rotary motor through a planetary gear reducer (reduction ratio 10:1); the rotary motor uses a harmonic reducer, with no-load backlash <1 arc minute and repeatability accuracy ±0.01mm; the derivative frame is a telescopic arm structure made of carbon fiber, with a maximum extension length of 800m. The main adjustment frame and auxiliary adjustment frame form a parallelogram mechanism, and the attitude is controlled by a servo motor of model MG996R. The adjustment angle range is -45° to +90°. The construction cylinder is a double-acting pneumatic cylinder with a cylinder diameter of 20mm, a stroke of 50mm, and a working pressure of 0.5 to 0.8MPa. The gripper consists of a gripper frame, a gripper telescopic device, and a gripper socket rotating frame. The gripping range is 5mm to 80mm, and the gripping force is continuously adjustable from 0 to 50N. It achieves flexible gripping through a force sensor and can automatically tighten M2 to M8 screws (torque accuracy ±5%), plug and unplug RJ45 ports (insertion force 8 to 12N), and connect fiber optic connectors with a diameter ≤12mm.

[0010] As a preferred implementation, the AI ​​large-scale model scene recognition algorithm is deployed on the NVIDIA Jetson AGX Orin computing platform within the main workbench. The algorithm employs an improved YOLOv8-lite network architecture, with an input resolution of 640×640 and an inference speed ≥30FPS. The scene recognition module includes 12 scene classifiers: indoor computer room (98.7% accuracy), outdoor courtyard (96.5% accuracy), and mountainous site (94.2% accuracy). A ResNet50 feature extraction network optimized through transfer learning enables scene judgment within 50ms. The driving parameter adjustment strategy includes: indoor... In indoor scenes, a silent mode is enabled (noise ≤ 55dB); in outdoor scenes, full-power drive is switched; and in mountainous scenes, terrain-adaptive algorithms are activated. A dynamic communication strategy switching mechanism is implemented: when the 5G-A signal strength is > -85dBm, 5G-A transmission (bandwidth ≥ 100Mbps) is used; when the signal strength is between -85dBm and -105dBm, dual-mode redundant transmission of 5G-A and LoRa is enabled; and when the signal strength is < -105dBm, automatic switching to LoRa communication (transmission rate 1.2kbps~50kbps adaptive) is achieved. The algorithm supports OTA upgrades, with monthly feature library updates, and the model size optimized to 8MB to meet edge computing requirements.

[0011] In a preferred embodiment, the ramp frame of the drive mechanism is a triangular truss structure made of high-strength steel, connected to the inner frame via rotating wheels, with a maximum load capacity of 500 kg; the anti-collision rollers are 80 mm in diameter, made of natural rubber, and internally equipped with spring shock absorbers (stiffness coefficient 20 N / mm), with a buffer stroke of 20 mm, capable of absorbing 20 J of impact energy; the auxiliary support has a cross-shaped reinforcing rib structure with a thickness of 5 mm, connected to the connecting support, raising the overall structural modal frequency to 25 Hz; the mounting plate is an epoxy glass cloth laminate (… The FR-4 cable is 3mm thick and has 32 M3 mounting holes on its surface for fixing the controller, power module and related electrical components. The binding rope is made of aramid fiber, 3mm in diameter, and has a breaking strength ≥500N. It is used to bind and connect the cable to the anti-collision rollers and the slope guide frame. The slope guide frame integrates an inclination sensor (measurement range ±45°, accuracy ±0.1°), which adjusts the driving force distribution of the drive wheel through a PID algorithm. When the slope is detected to be >8°, the anti-slip mode is automatically activated to prevent slippage by increasing the adhesion of the drive wheel (the coefficient of friction is increased to 0.8).

[0012] In a preferred embodiment, the energy storage battery integrated on the detection substrate is a lithium iron phosphate battery pack with a nominal voltage of 24V, a capacity of 20Ah, an energy density of 150Wh / kg, and a cycle life of ≥2000 cycles; the battery management system (BMS) supports overcharge (protection voltage 29.2V), over-discharge (protection voltage 20V), overcurrent (protection current 30A), and short-circuit protection; the solar-assisted charging system includes two 100W monocrystalline silicon solar panels (conversion efficiency 23%), configured with an MPPT controller (tracking efficiency ≥99%), under standard illumination conditions (1000W / m²). 2 It can provide 150W charging power; the battery life meets the following requirements: 12 hours of continuous operation in an indoor constant temperature environment (25℃), 8 hours of continuous operation in an outdoor normal temperature environment (20℃~35℃), and 5 hours of continuous operation in a low temperature environment (-10℃); the battery insulation module adopts a polyimide heating film (power 50W), which automatically starts when the battery temperature is <5℃, maintains the battery operating temperature in the range of 10℃~45℃, and the heating response time is <3 minutes.

[0013] In a preferred embodiment, the positioning system employs a multi-source fusion scheme of "LiDAR + Visual SLAM + Inertial Navigation". The LiDAR is a 16-line mechanical rotating radar (360° horizontal field of view, -15° to +15° vertical field of view, ranging range 0.1m to 100m, angular resolution 0.1°); the Visual SLAM module includes two fisheye cameras (190° field of view, 1280×800 resolution) and one depth camera (ranging range 0.5m to 10m, accuracy ±1%); the inertial navigation unit is a MEMSMU (three-axis accelerometer range ±16g, three-axis gyroscope). The instrument has a range of ±2000° / s and a zero-bias stability of ≤0.1° / h. The fusion positioning algorithm uses a tightly coupled Kalman filter. In areas with base station signals (GPS / BeiDou positioning accuracy 1m), the positioning error is ≤0.5m. In areas without base station signals, centimeter-level positioning (planar accuracy ±3cm, elevation accuracy ±5cm) is achieved through environmental feature matching (keyframe matching quantity ≥50 / frame). The positioning data is updated at a frequency of 10Hz and is fused with the global communication heat map through the ROS system. The superimposed 3D scene is displayed on the control terminal, and the coordinate system adopts the UTM projection coordinate system.

[0014] In one preferred embodiment, the maintenance mechanism has a three-piece frame made of aluminum alloy profiles, with three standard interfaces (compliant with ISO9409-1 robot quick-change interface standard), an interface repeatability of ±0.02mm, and a construction probe for detecting and monitoring the operation of the gripper.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: 1. This invention achieves adaptive movement in complex terrain through a drive mechanism. The drive wheels adopt a 250mm diameter Mecanum wheel structure, which, together with the 24V DC brushless motor of the driver and the parallelogram linkage mechanism of the adjustment bracket, can achieve 360° in-situ turning, lateral movement, and 15° slope climbing. The triangular truss structure of the slope frame, the spring shock absorbers of the anti-collision rollers, and the tilt sensor of the slope guide frame improve the stability and passability of the robot in complex environments such as indoor machine room floor tiles, outdoor cement ground, and mountain gravel roads.

[0016] 2. This invention achieves precise detection and stable communication across the entire domain through a detection mechanism and communication system. The transceiver integrates a 5G-A and LoRa dual-mode communication module, with a 5G-A peak rate of 10Gbps and a transmission latency of ≤1ms, and a LoRa receiving sensitivity of -148dBm. Combined with a VSWR monitoring circuit, it ensures 99.99% communication stability. The infrared detection eye, laser detection eye, and backup detection head of the detection mechanism generate a 0.5m×0.5m resolution full-domain communication heat map through spatiotemporal registration. Combined with an artificial intelligence large-scale model scene recognition algorithm, it achieves centimeter-level environmental perception and scene adaptation.

[0017] 3. This invention achieves high-precision operation and maintenance through the maintenance mechanism. The derivative frame, together with the parallelogram mechanism of the main adjustment frame and the auxiliary adjustment frame, can flexibly adjust the working position. The gripper achieves flexible gripping through the force sensor and can automatically tighten M2~M8 screws, plug and unplug RJ45 ports and connect fiber optic connectors with a diameter ≤12mm. The ISO9409-1 standard interface of the three-piece frame further improves the accuracy and compatibility of maintenance operations. Attached Figure Description

[0018] Figure 1 This invention proposes a three-dimensional diagram of a communication operation and maintenance robot driven by a large model of multi-sensor detection with global adaptive capability; Figure 2 This invention proposes a partial disassembly 3D diagram of a large-scale model driven by multiple sensors for global adaptive communication operation and maintenance robot. Figure 3 This invention presents a three-dimensional disassembly of the detection and maintenance mechanisms of a communication operation and maintenance robot driven by a large model of multi-sensor detection with global adaptive capability. Figure 4 This invention presents a three-dimensional diagram of a drive mechanism for a communication maintenance robot driven by a large model using multi-sensor detection for global adaptive operation. Figure 5 This invention presents a three-dimensional disassembly of the drive mechanism of a communication maintenance robot driven by a large model using multi-sensor detection for global adaptive operation. Figure 6This invention presents a three-dimensional disassembly of the detection mechanism of a communication maintenance robot driven by a large model of multi-sensor detection with global adaptive capability. Figure 7 This invention proposes a partial three-dimensional view of the maintenance mechanism of a communication operation and maintenance robot driven by a large model of multi-sensor detection with global adaptive capability; Figure 8 This invention proposes a partial three-dimensional view of a derivative frame of a communication operation and maintenance robot driven by a large model of multi-sensor detection with global adaptive capability.

[0019] Legend: 1. Drive mechanism; 11. Outer frame; 12. Inner frame; 13. Connecting frame; 14. Placement frame; 15. Driver; 16. Drive wheel; 17. Drive wheel; 18. Auxiliary bracket; 19. Connecting bracket; 110. Mounting plate; 111. Slope frame; 112. Anti-collision roller; 113. Binding rope; 114. Slope guide frame; 115. Adjustment bracket; 2. Detection mechanism; 21. Detection base plate; 22. Energy storage battery; 23. Main working box; 24. Limiting bracket; 25. Detection base frame; 26. Infrared detection eye; 27. Laser detection eye; 28. Spare detection head; 29. ​​Signal transceiver; 210. External frame; 211. Side limiting bracket; 212. Limiting pin; 3. Maintenance mechanism; 31. Expansion plate; 32. Maintenance base plate; 33. Support frame; 34. Drive cavity; 35. Drive motor; 36. Rotator; 37. Rotary motor; 38. Main adjustment frame; 39. Auxiliary adjustment frame; 310. Derivative frame; 311. Three-piece frame; 312. Detection frame; 313. Construction probe; 314. Construction cylinder; 315. Hand clamp frame; 316. Hand clamp telescopic device; 317. Hand clamp rotating frame; 319. Hand clamp. Detailed Implementation

[0020] To make the technical problems, technical solutions, and beneficial effects of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the described embodiments are only a part of the embodiments of this application, not all of them. The specific embodiments described herein are only used to explain the invention and are not intended to limit the invention. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] It should be further noted that the accompanying drawings and embodiments of the present invention mainly describe the concept of the present invention. Based on this concept, some specific forms and arrangements of connection relationships, positional relationships, power mechanisms, power supply systems, hydraulic systems and control systems may not be fully described. However, under the premise that those skilled in the art understand the concept of the present invention, they can implement the above-mentioned specific forms and arrangements in a well-known manner.

[0022] When a component is referred to as being "fixed to" or "set on" another component, it can be directly on or indirectly on that other component. When a component is referred to as being "connected to" another component, it can be directly connected to or indirectly connected to that other component.

[0023] The directional terms "inner" and "outer" refer to the inner and outer contours of each component itself. The terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the present invention.

[0024] For ease of description, spatial relative terms such as "above," "on top of," "on the upper surface of," "above," etc., are used herein to describe the spatial positional relationship of a device or feature as shown in the figures to other devices or features. It should be understood that spatial relative terms are intended to encompass different orientations in use or operation beyond the orientation of the device as described in the figures. For example, if the device in the figures were inverted, a device described as "above" or "on top of" other devices or structures would subsequently be positioned as "below" or "under" other devices or structures. Thus, the exemplary term "above" can include both "above" and "below." The device may also be positioned in other different ways, and the spatial relative descriptions used herein will be interpreted accordingly.

[0025] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, and "several" means one or more, unless otherwise explicitly specified.

[0026] The present invention will now describe a large-model driven communication maintenance robot with global adaptive multi-sensor detection. Example 1

[0027] like Figure 1-5 As shown, the present invention provides a technical solution: a large-model driven communication operation and maintenance robot with full-domain adaptive multi-sensor detection, including: a drive mechanism 1, a detection mechanism 2 and a maintenance mechanism 3. The drive mechanism 1 includes an outer frame 11, an inner frame 12, a connecting frame 13, a placement frame 14, a driver 15, a drive wheel 16, a drive wheel 17, an auxiliary support 18, a connecting support 19, a mounting plate 110, a slope frame 111, anti-collision rollers 112, binding ropes 113, a slope guide frame 114 and an adjustment support 115; The outer frame 11 is made of 6061-T6 aluminum alloy and forms a double-layer protective structure with the inner frame 12 through laser welding. The connecting frame 13 is an I-shaped connector made of carbon fiber composite material and is rigidly fixed to the outer frame 11 and the inner frame 12 through M8 high-strength bolts. The detection mechanism 2 integrates a detection substrate 21, an energy storage battery 22, a main working box 23, a limit card holder 24, a detection base frame 25, an infrared detection eye 26, a laser detection eye 27, a spare detection head 28, a signal transceiver 29, an external frame 210, a side limit bracket 211, and a limit pin 212. The detection substrate 21 is made of aerospace-grade magnesium alloy plate, and the surface is covered with a 0.2mm thick graphene thermal conductive film. The maintenance mechanism 3 is equipped with an extension plate 31, a maintenance base plate 32, a support sleeve 33, a drive cavity 34, a drive motor 35, a rotator 36, a rotary motor 37, a main adjustment frame 38, an auxiliary adjustment frame 39, a derivative frame 310, a three-piece frame 311, a detection frame 312, a construction probe 313, a construction cylinder 314, a gripper frame 315, a gripper telescopic device 316, a gripper socket rotating frame 317, and a gripper 319. The expansion board 31 is connected to the maintenance base plate 32 through a tenon and mortise structure, and the joint is filled with nitrile rubber sealing gaskets; the robot is equipped with an adaptive hardware architecture consisting of a heterogeneous computing unit, a multimodal sensor interface and an adaptive power management system, as well as an artificial intelligence large model scene recognition algorithm deployed in the main working box 23. It achieves full-domain autonomous operation and maintenance, centimeter-level precise positioning and 99.99% communication stability through 5G-A and LoRa dual-mode communication modules; The drive wheel 16 adopts a 250mm diameter Mecanum wheel structure, with the wheel body forged from 7075 aluminum alloy and covered with a polyurethane tread with a Shore hardness of 75. The tread has a 3mm deep diamond-shaped anti-slip pattern. The drive wheel 17 is a 2.5 module helical gear structure, connected to the output shaft of the driver 15 via a keyway. The driver 15 uses a 24V DC brushless motor with a rated power of 350W and a peak torque of 12N·m, and is equipped with a 1024-line encoder to achieve closed-loop control. The adjusting bracket 115 includes two sets of parallelogram linkage mechanisms, driven by a DS3218 servo motor. The adjustable stroke range is ±15°, and the response time is ≤0.1s. The connecting frame 13 is equipped with Φ12mm positioning pin holes at both ends, which form a clearance fit with the positioning boss of the outer frame 11, with a fit accuracy of H7 / g6. The placement frame 14 is equipped with a heat dissipation air duct, which achieves forced air cooling through an axial flow fan, with an operating temperature range of -40℃ to +70℃. This drive system can achieve 360° in-situ turning with a turning angular velocity ≥60° / s, a lateral movement speed ≥0.8m / s, and 15° slope climbing. It can adapt to complex terrains such as indoor computer room floor tiles, outdoor cement ground, and mountain gravel roads through a multi-wheel differential speed algorithm. The ramp frame 111 of the drive mechanism 1 is a triangular truss structure made of high-strength steel. It is connected to the inner frame 12 via rotating wheels and has a maximum load capacity of 500 kg. The anti-collision rollers 112 are 80 mm in diameter, made of natural rubber, and have internal spring shock absorbers with a stiffness coefficient of 20 N / mm and a buffer stroke of 20 mm, capable of absorbing 20 J of impact energy. The auxiliary support 18 has a cross-shaped reinforcing rib structure with a thickness of 5 mm and is connected to the connecting support 19, raising the overall structural modal frequency to 25 Hz. The mounting plate 110 is an epoxy fiberglass laminate FR-4 with a thickness of [missing information]. The surface has 32 M3 mounting holes, 3mm in diameter, for fixing the controller, power module and related electrical components; the binding rope 113 is made of aramid fiber, 3mm in diameter, with a breaking strength ≥500N, and is used to bind and connect with the anti-collision roller 112 and the slope guide frame 114; the slope guide frame 114 integrates an inclination sensor with a measurement range of ±45° and an accuracy of ±0.1°. It adjusts the driving force distribution of the drive wheel 16 through a PID algorithm. When the slope is detected to be >8°, the anti-slip mode is automatically activated, and the adhesion friction coefficient of the drive wheel 16 is increased to 0.8 to prevent slippage.

[0028] In this embodiment, adaptive movement over complex terrain is achieved through the drive mechanism 1. The drive wheel 16 adopts a Mecanum wheel structure with a diameter of 250mm. Combined with the 24V DC brushless motor (rated power 350W, peak torque 12N·m) of the driver 15 and the parallelogram linkage mechanism of the adjustment bracket 115, it can achieve 360° in-situ turning (turning angular velocity ≥60° / s), lateral movement (speed ≥0.8m / s), and 15° slope climbing. The triangular truss structure of the slope frame 111 (bearing 500kg), the spring shock absorber of the anti-collision roller 112 (absorbing 20J of impact energy), and the tilt sensor of the slope guide frame 114 (activating anti-slip mode when the slope is >8°) improve the stability and passability of the robot in complex environments such as indoor machine room floor tiles, outdoor cement ground, and mountain gravel roads. Example 2

[0029] like Figure 1-6 As shown, the transceiver 29 integrates a 5G-A and LoRa dual-mode communication module. The 5G-A module adopts the 3GPP Release 18 standard, supports the n257 / n258 millimeter wave band, has a peak rate of 10Gbps, a transmission latency of ≤1ms, and is configured with a 4×4 MIMO antenna array. The LoRa module operates in the 433MHz ISM band, uses a spreading factor of SF12, has a maximum transmit power of 20dBm, a receive sensitivity of -148dBm, and supports Class A / C operating modes. The module has a built-in VSWR monitoring circuit, which collects reflected power in real time through a directional coupler, with a measurement range of 1.0~3.0 and an accuracy of ±0.05. When the VSWR is >1.5, it automatically triggers the antenna switching mechanism. The communication module adopts a metal shield design, and its electromagnetic compatibility level reaches EN301489-1 / 3 standard, supports IP67 protection level, and can work continuously in an environment of -30℃~+65℃. The outer frame 210 of the detection mechanism 2 is made of 304 stainless steel with a thickness of 2mm. It features a honeycomb-shaped hollow structure formed by laser cutting, reducing weight by 35% compared to traditional designs. The side limiting bracket 211 is an aluminum alloy support, fixed to the outer frame 210 by two 8mm diameter limiting pins 212 with a pin clearance of 0.05mm~0.1mm. The infrared detection eye 26 uses an uncooled focal plane array detector with a resolution of 640×512, thermal sensitivity <50mK, field of view of 45°×34°, and frame rate of 30Hz. The laser detection eye 27 uses a 1550nm wavelength fiber laser. The instrument has an output power of 5mW, a measurement distance of 0.5m~100m, an accuracy of ±2mm, and a scanning frequency of 50Hz; the spare probe head 28 includes a high-definition industrial camera with 2 megapixels and a frame rate of 60fps, and a temperature and humidity sensor with a measurement range of -40℃~+85℃ and humidity of 0~100%RH, with an accuracy of ±0.5℃ / ±2%RH; multi-sensor data is transmitted to the main working box 23 via gigabit Ethernet, and after spatiotemporal registration, a full-area communication heat map with a resolution of 0.5m×0.5m is generated. The color mapping adopts the Jet pseudo-color scheme, with a dynamic range of -120dBm~-30dBm; The AI ​​large-scale scene recognition algorithm is deployed on the NVIDIA Jetson AGX Orin computing platform within the main workbench 23. The algorithm employs an improved YOLOv8-lite network architecture, with an input resolution of 640×640 and an inference speed ≥30FPS. The scene recognition module includes 12 scene classifiers with accuracy rates of 98.7% for indoor computer rooms, 96.5% for outdoor courtyards, and 94.2% for mountainous sites. A ResNet50 feature extraction network optimized through transfer learning enables scene judgment within 50ms. Driving parameter adjustment strategies include: enabling static... The audio mode noise is ≤55dB, outdoor scenes switch to full power drive, and mountain scenes activate terrain adaptive algorithm; the communication strategy dynamic switching mechanism: when the 5G-A signal strength is >-85dBm, the 5G-A transmission bandwidth is ≥100Mbps; when the signal strength is in the range of -85dBm to -105dBm, 5G-A and LoRa dual-mode redundant transmission is enabled; when the signal strength is <-105dBm, it automatically switches to LoRa communication transmission rate of 1.2kbps~50kbps adaptive; the algorithm supports OTA upgrades, pushes feature library updates monthly, and the model size is optimized to 8MB to adapt to edge computing needs; The energy storage battery 22 integrated on the detection substrate 21 is a lithium iron phosphate battery pack with a nominal voltage of 24V, a capacity of 20Ah, an energy density of 150Wh / kg, and a cycle life of ≥2000 cycles. The battery management system (BMS) supports overcharge protection voltage of 29.2V, over-discharge protection voltage of 20V, overcurrent protection current of 30A, and short-circuit protection. The solar-assisted charging system includes two 100W monocrystalline silicon solar panels with a conversion efficiency of 23%, and is equipped with an MPPT controller to track an efficiency of ≥99%, achieving 1000W / m² under standard illumination conditions. 2 It can provide 150W charging power; the battery life meets the following requirements: 12 hours of continuous operation in an indoor constant temperature environment of 25℃, 8 hours of continuous operation in an outdoor normal temperature environment of 20℃~35℃, and 5 hours of continuous operation in a low temperature environment of -10℃; the battery insulation module adopts a polyimide heating film with a power of 50W, which automatically starts when the battery temperature is <5℃, maintains the battery operating temperature in the range of 10℃~45℃, and the heating response time is <3 minutes. The positioning system employs a multi-source fusion scheme of "LiDAR + Visual SLAM + Inertial Navigation". The LiDAR is a 16-line mechanical rotating radar with a horizontal field of view of 360°, a vertical field of view of -15° to +15°, a ranging range of 0.1m to 100m, and an angular resolution of 0.1°. The Visual SLAM module includes two fisheye cameras with a 190° field of view and a resolution of 1280×800, and one depth camera with a ranging range of 0.5m to 10m and an accuracy of ±1%. The inertial navigation unit is a MEMSIMU three-axis accelerometer with a range of ±16g and a three-axis gyroscope with a range of ±200. 0° / s, zero-bias stability ≤0.1° / h; the fusion positioning algorithm adopts tightly coupled Kalman filtering, and the positioning error is ≤0.5m when the GPS / BeiDou positioning accuracy is 1m in the base station signal area; in the area without base station signal, the centimeter-level positioning plane accuracy is ±3cm and the elevation accuracy is ±5cm by matching ≥50 key frames / frame through environmental feature matching; the positioning data is updated at a frequency of 10Hz, and the data is fused with the global communication heat map through the ROS system. The superimposed 3D scene is displayed on the control terminal, and the coordinate system adopts the UTM projection coordinate system.

[0030] In this embodiment, accurate full-area detection and stable communication are achieved through the detection mechanism 2 and the communication system. The transceiver 29 integrates a 5G-A and LoRa dual-mode communication module, with a 5G-A peak rate of 10Gbps and a transmission latency of ≤1ms, and a LoRa receiving sensitivity of -148dBm. Combined with the VSWR monitoring circuit (automatic antenna switching when VSWR > 1.5), 99.99% communication stability is guaranteed. The infrared detection eye 26 (resolution 640×512), the laser detection eye 27 (range 0.5m~100m, accuracy ±2mm), and the spare detection head 28 (including a high-definition camera and temperature and humidity sensor) of the detection mechanism 2 generate a full-area communication heat map with a resolution of 0.5m×0.5m through spatiotemporal registration. Combined with the artificial intelligence large model scene recognition algorithm (inference speed ≥30FPS, 12 scene classifications), centimeter-level environmental perception and scene adaptation are achieved. Example 3

[0031] like Figure 1-8 As shown, the expansion plate 31 of the maintenance mechanism 3 is made of 7075-T73 aluminum alloy sheet, 10mm thick, with a hard anodized surface; the support frame 33 is a hollow cylindrical structure with an inner diameter of 80mm and a length of 150mm; the side-mounted drive cavity 34 is designed with an IP65 protection rating; the drive motor 35 is a servo motor with a brake, a rated speed of 3000rpm, an encoder resolution of 17 bits, and drives the rotary motor 36 through a planetary gear reducer with a reduction ratio of 10:1; the rotary motor 37 uses a harmonic reducer, with no-load backlash <1 arc minute and repeatability accuracy ±0.01mm; the derivative frame 310 is a telescopic arm structure made of carbon fiber, with a maximum extension length of 800mm and a load-bearing capacity of 5k. g; The main adjustment frame 38 and the auxiliary adjustment frame 39 form a parallelogram mechanism, and the attitude is controlled by a servo motor of model MG996R. The adjustment angle range is -45° to +90°; The construction cylinder 314 is a double-acting pneumatic cylinder with a cylinder diameter of 20mm, a stroke of 50mm, and a working pressure of 0.5 to 0.8MPa; The gripper 319 consists of a gripper frame 315, a gripper telescopic device 316, and a gripper socket rotating frame 317. The gripping range is 5mm to 80mm, and the gripping force is continuously adjustable from 0 to 50N. Flexible gripping is achieved through a force sensor. It can automatically tighten M2 to M8 screws with a torque accuracy of ±5%, insert RJ45 port with an insertion force of 8 to 12N, and perform docking operations on fiber optic connectors with a diameter ≤12mm. The maintenance mechanism 3 has a three-piece frame 311 made of aluminum alloy profiles. It has three standard interfaces that conform to the ISO9409-1 robot quick-change interface standard. The interface repeatability is ±0.02mm. It is equipped with a construction probe 313 to detect and monitor the operation of the gripper 319.

[0032] In this embodiment, high-precision maintenance operations are achieved through maintenance mechanism 3. The derivative frame 310 (carbon fiber telescopic arm, maximum extension 800mm, load capacity 5kg) works in conjunction with the parallelogram mechanism of the main adjustment frame 38 and the auxiliary adjustment frame 39 (adjustment angle -45°~+90°) to flexibly adjust the working position. The gripper 319 (gripping range 5mm~80mm, gripping force 0~50N adjustable) achieves flexible gripping through force sensor, and can automatically tighten M2~M8 screws (torque accuracy ±5%), plug and unplug RJ45 ports (insertion force 8~12N), and connect fiber optic connectors with a diameter ≤12mm. The ISO9409-1 standard interface of the three-panel frame 311 (repeat positioning accuracy ±0.02mm) further improves the accuracy and compatibility of maintenance operations.

[0033] Working principle: like Figure 1-8 As shown, this device achieves autonomous movement across complex terrain through drive mechanism 1. Drive wheel 16 uses a 250mm diameter Mecanum wheel structure, forged from 7075 aluminum alloy, with a polyurethane tread of Shore hardness 75 and 3mm deep diamond-shaped anti-slip tread. Combined with the 24V DC brushless motor (rated power 350W, peak torque 12N·m) and 1024-line encoder closed-loop control of the driver 15, it can achieve 360° in-situ turning (steering angular velocity ≥60° / s) and lateral movement (speed ≥0.8m / s). Adjustment bracket 115 is driven by two sets of parallelogram linkage mechanisms via a DS3218 servo motor, with an adjustment stroke of ±15° (response time ≤0.8m / s). The slope guide frame 114, with its triangular truss structure (bearing 500kg) and tilt sensor (measuring range ±45°, accuracy ±0.1°), automatically activates the anti-slip mode when a slope > 8° is detected. It adjusts the adhesion of the drive wheel through a PID algorithm (increasing the friction coefficient to 0.8) to achieve 15° slope climbing. The anti-collision roller 112 (80mm diameter natural rubber material, built-in spring shock absorber, buffer stroke 20mm) can absorb 20J of impact energy. The cross-shaped reinforcing rib structure of the auxiliary support 18 increases the overall modal frequency to 25Hz, ensuring stability on complex terrains such as indoor floor tiles, outdoor cement floors, and mountain gravel roads.

[0034] The detection mechanism 2 achieves full-area environmental perception through multi-sensor fusion. The infrared detection eye 26 employs a 640×512 resolution uncooled focal plane array detector (thermal sensitivity <50mK, field of view 45°×34°, frame rate 30Hz). The laser detection eye 27 uses a 1550nm fiber laser (range 0.5m~100m, accuracy ±2mm, scanning frequency 50Hz). The backup detection head 28 integrates a 2-megapixel high-definition industrial camera (frame rate 60fps) and a temperature and humidity sensor (-40℃~+85℃, humidity 0~100%RH, accuracy ±0.5℃ / ±2%RH). Multi-sensor data is transmitted to the main control box 23 via gigabit Ethernet, and a 0.5m×0.5m resolution full-area communication heatmap (Jet pseudo-color) is generated through spatiotemporal registration. The dynamic range is -120dBm to -30dBm. The transceiver 29 integrates a 5G-A and LoRa dual-mode communication module: the 5G-A module supports the 3GPP Release 18 standard (n257 / n258 band, peak rate 10Gbps, latency ≤1ms, 4×4 MIMO), and the LoRa module operates in the 433MHz ISM band (SF12 spreading factor, receiver sensitivity -148dBm). It has a built-in VSWR monitoring circuit (measurement range 1.0~3.0, accuracy ±0.05), which automatically switches the antenna when the VSWR is >1.5. Combined with a metal shield (EN301489-1 / 3 electromagnetic compatibility standard) and IP67 protection level, it ensures 99.99% communication stability in an environment of -30℃ to +65℃.

[0035] The robot's adaptive hardware architecture includes heterogeneous computing units, multimodal sensor interfaces, and an adaptive power management system. The large-scale AI scene recognition algorithm is deployed on the NVIDIA Jetson AGX Orin platform in the main workbench 23. The algorithm uses an improved YOLOv8-lite network (640×640 input, inference speed ≥30FPS) and a ResNet50 feature extraction network optimized through transfer learning, achieving 12 scene classifications within 50ms (98.7% accuracy in indoor computer rooms, 96.5% in outdoor courtyards, and 94.2% in mountainous areas). Based on the scene recognition results, the operating strategy is dynamically adjusted: a silent mode is activated indoors (noise ≤5). The signal strength is 5dB, with full-power drive outdoors and terrain-adaptive algorithm activated in mountainous areas. The communication strategy switches according to the 5G-A signal strength: single-mode transmission (bandwidth ≥ 100Mbps) when > -85dBm, dual-mode redundant transmission when -85dBm to -105dBm, and automatic switching to LoRa (adaptive rate 1.2kbps to 50kbps) when < -105dBm. The algorithm supports OTA upgrades (monthly feature library updates), and the model size is optimized to 8MB to meet edge computing needs. Combined with multi-source fusion positioning of "LiDAR + visual SLAM + inertial navigation" (planar accuracy ±3cm, elevation accuracy ±5cm), it can achieve autonomous operation and maintenance and centimeter-level precision operation across the entire domain.

[0036] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0037] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0038] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of this application. It should also be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale. Techniques, methods, and devices known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and devices should be considered part of the specification. In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values. It should be noted that similar reference numerals and letters in the following drawings denote similar items; therefore, once an item is defined in one drawing, it need not be further discussed in subsequent drawings.

Claims

1. A large-scale model-driven communication maintenance robot with global adaptive multi-sensor detection, characterized in that, include: The drive mechanism (1), the detection mechanism (2), and the maintenance mechanism (3) include an outer frame (11), an inner frame (12), a connecting frame (13), a placement frame (14), a driver (15), a drive wheel (16), a drive wheel (17), an auxiliary bracket (18), a connecting bracket (19), a mounting plate (110), a slope frame (111), a crash roller (112), a binding rope (113), a slope guide frame (114), and an adjustment bracket (115). The outer frame (11) is made of 6061-T6 aluminum alloy and forms a double-layer protective structure with the inner frame (12) through laser welding process. The connecting frame (13) is an I-shaped connector made of carbon fiber composite material. The outer frame (11) and the inner frame (12) are rigidly fixed by M8 high-strength bolts. The detection mechanism (2) integrates a detection substrate (21), an energy storage battery (22), a main working box (23), a limit card holder (24), a detection base frame (25), an infrared detection eye (26), a laser detection eye (27), a spare detection head (28), a signal transceiver (29), an external frame (210), a side limit frame (211), and a limit pin (212). The detection substrate (21) is made of aerospace-grade magnesium alloy plate, and the surface is covered with a 0.2mm thick graphene thermal conductive film. The maintenance mechanism (3) is equipped with an extension plate (31), a maintenance base plate (32), a support frame (33), a drive cavity (34), a drive motor (35), a rotator (36), a rotary motor (37), a main adjustment frame (38), an auxiliary adjustment frame (39), a derivative frame (310), a three-piece frame (311), a detection frame (312), a construction probe (313), a construction cylinder (314), a gripper frame (315), a gripper telescopic device (316), a gripper socket rotating frame (317), and a gripper (319). The expansion board (31) is connected to the maintenance base plate (32) through a tenon and mortise structure, and the joint is filled with nitrile rubber sealing gaskets; the robot is equipped with an adaptive hardware architecture consisting of a heterogeneous computing unit, a multimodal sensor interface and an adaptive power management system, as well as an artificial intelligence large model scene recognition algorithm deployed in the main working box (23), and achieves full-domain autonomous operation and maintenance, centimeter-level precise positioning and 99.99% communication stability through the 5G-A and LoRa dual-mode communication module.

2. The large-scale model-driven communication maintenance robot with multi-sensor detection for global adaptive operation according to claim 1, characterized in that: The drive wheel (16) adopts a Mecanum wheel structure with a diameter of 250mm. The wheel body is forged from 7075 aluminum alloy and the surface is covered with a polyurethane tread with a Shore hardness of 75. The tread is provided with a 3mm deep diamond anti-slip pattern. The drive wheel (17) is a helical gear structure with a module of 2.

5. It is connected to the output shaft of the driver (15) through a keyway. The driver (15) adopts a 24V DC brushless motor with a rated power of 350W and a peak torque of 12N·m. It is equipped with a 1024-line encoder to realize closed-loop control. The adjustment bracket (115) contains two sets of parallelogram linkage mechanisms, which are driven by a DS3218 servo motor. The drive system can adjust the stroke range to ±15° and the response time to ≤0.1s. The connecting frame (13) is provided with Φ12mm positioning pin holes at both ends, which form a clearance fit with the positioning boss of the outer frame (11), with a fit accuracy of H7 / g6. The placement frame (14) is provided with a heat dissipation air duct inside, and forced air cooling is achieved through an axial flow fan. The working temperature range is -40℃ to +70℃. The drive system can achieve 360° in-situ turning (turning angle speed ≥60° / s), lateral movement (speed ≥0.8m / s), and 15° slope climbing. It can adapt to complex terrains such as indoor machine room floor tiles, outdoor cement ground, and mountain gravel roads through a multi-round differential speed algorithm.

3. The large-scale model-driven communication maintenance robot with multi-sensor detection for global adaptive operation according to claim 1, characterized in that: The transceiver (29) integrates a 5G-A and LoRa dual-mode communication module. The 5G-A module adopts the 3GPP Release 18 standard, supports the n257 / n258 millimeter wave band, has a peak rate of 10Gbps, a transmission delay of ≤1ms, and is configured with a 4×4 MIMO antenna array. The LoRa module operates in the 433MHz ISM band, employs a spreading factor of SF12, has a maximum transmit power of 20dBm, a receive sensitivity of -148dBm, and supports Class A / C operating modes. The module incorporates a built-in VSWR monitoring circuit, which acquires reflected power in real time via a directional coupler, with a measurement range of 1.0~3.0 and an accuracy of ±0.

05. When the VSWR > 1.5, an automatic antenna switching mechanism is triggered. The communication module features a metal shielding design, achieving an electromagnetic compatibility level of EN301489-1 / 3, supporting IP67 protection, and can operate continuously in environments ranging from -30℃ to +65℃.

4. The large-model driven communication maintenance robot with global adaptive multi-sensor detection according to claim 1, characterized in that: The outer frame (210) of the detection mechanism (2) is made of 304 stainless steel with a thickness of 2mm. It is formed by laser cutting to create a honeycomb hollow structure, which reduces the weight by 35% compared to the traditional design. The side limiting bracket (211) is an aluminum alloy bracket, which is fixed to the outer frame (210) by two limiting pins (212) with a diameter of 8mm. The pin fitting gap is 0.05mm~0.1mm. The infrared detection eye (26) adopts an uncooled focal plane array detector with a resolution of 640×512, thermal sensitivity <50mK, field of view of 45°×34°, and frame rate of 30Hz. The laser detection eye (27) has a 1550nm wavelength. The long fiber laser has an output power of 5mW, a measurement distance of 0.5m~100m, an accuracy of ±2mm, and a scanning frequency of 50Hz. The backup probe (28) includes a high-definition industrial camera (2 million pixels, 60fps) and a temperature and humidity sensor (measurement range -40℃~+85℃, humidity 0~100%RH, accuracy ±0.5℃ / ±2%RH). The multi-sensor data is transmitted to the main working box (23) via gigabit Ethernet. After spatiotemporal registration, a full-domain communication heat map with a resolution of 0.5m×0.5m is generated. The color mapping adopts the Jet pseudo-color scheme with a dynamic range of -120dBm~-30dBm.

5. The large-model driven communication maintenance robot with global adaptive multi-sensor detection according to claim 1, characterized in that: The expansion plate (31) of the maintenance mechanism (3) is made of 7075-T73 aluminum alloy plate with a thickness of 10mm and a hard anodized surface. The support frame (33) is a hollow cylindrical structure with an inner diameter of 80mm and a length of 150mm. The side-mounted drive chamber (34) is designed with an IP65 protection rating. The drive motor (35) is a servo motor with a brake, a rated speed of 3000rpm, an encoder resolution of 17 bits, and drives the rotary motor (36) through a planetary gear reducer (reduction ratio 10:1). The rotary motor (37) uses a harmonic reducer with a no-load backlash of <1 arc minute and a repeatability of ±0.01mm. The derivative frame (310) is a telescopic arm structure made of carbon fiber with a maximum extension length of 800mm and a load-bearing capacity of 5. kg; the main adjustment frame (38) and the auxiliary adjustment frame (39) form a parallelogram mechanism, and the attitude is controlled by a servo motor of model MG996R, with an adjustment angle range of -45° to +90°; the construction cylinder (314) is a double-acting pneumatic cylinder with a cylinder diameter of 20mm, a stroke of 50mm, and a working pressure of 0.5 to 0.8MPa; the gripper (319) consists of a gripper frame (315), a gripper telescopic device (316), and a gripper socket rotating frame (317), with a gripping range of 5mm to 80mm and a gripping force of 0 to 50N that is continuously adjustable. It achieves flexible gripping through a force sensor and can complete the automatic tightening of M2 to M8 screws (torque accuracy ±5%), RJ45 port insertion and removal (insertion force 8 to 12N), and docking operation of fiber optic connectors with a diameter ≤12mm.

6. The large-model driven communication maintenance robot with multi-sensor detection for global adaptive operation according to claim 1, characterized in that: The AI ​​large-scale model scene recognition algorithm is deployed on the NVIDIA Jetson AGX Orin computing platform in the main work box (23). The algorithm adopts an improved YOLOv8-lite network architecture, with an input resolution of 640×640 and an inference speed of ≥30FPS. The scene recognition module includes 12 scene classifiers, including indoor computer room (recognition accuracy 98.7%), outdoor courtyard (recognition accuracy 96.5%), and mountain site (recognition accuracy 94.2%). The scene judgment within 50ms is achieved through the ResNet50 feature extraction network optimized by transfer learning. The drive parameter adjustment strategies include: enabling silent mode (noise ≤ 55dB) in indoor scenes, switching to full power drive in outdoor scenes, and activating terrain adaptive algorithm in mountainous scenes; Dynamic switching mechanism for communication strategy: When the 5G-A signal strength is > -85dBm, 5G-A transmission is used (bandwidth ≥ 100Mbps); when the signal strength is in the range of -85dBm to -105dBm, dual-mode redundant transmission of 5G-A and LoRa is enabled; when the signal strength is < -105dBm, it automatically switches to LoRa communication (transmission rate 1.2kbps~50kbps adaptive); the algorithm supports OTA upgrades, pushes feature library updates monthly, and optimizes the model size to 8MB to adapt to edge computing needs.

7. The large-scale model-driven communication maintenance robot with multi-sensor detection for global adaptive operation according to claim 1, characterized in that: The slope frame (111) of the drive mechanism (1) is a triangular truss structure made of high-strength steel. It is connected to the inner frame (12) via rotating wheels and has a maximum load capacity of 500kg. The anti-collision roller (112) has a diameter of 80mm and is made of natural rubber. It is equipped with a spring shock absorber (stiffness coefficient 20N / mm) with a buffer stroke of 20mm, which can absorb 20J of impact energy. The auxiliary support (18) has a cross-shaped reinforcing rib structure with a thickness of 5mm and is connected to the connecting support (19) to increase the overall structural modal frequency to 25Hz. The mounting plate (110) is an epoxy glass cloth laminate (FR-4). The thickness is 3mm, and the surface is provided with 32 M3 mounting holes for fixing the controller, power module and related electrical components; the binding rope (113) is made of aramid fiber, with a diameter of 3mm and a breaking strength ≥500N, and is bound to the anti-collision roller (112) and the slope guide frame (114); the slope guide frame (114) integrates an inclination sensor (measurement range ±45°, accuracy ±0.1°), and adjusts the driving force distribution of the drive wheel (16) through the PID algorithm. When the slope is detected to be >8°, the anti-slip mode is automatically activated, and the adhesion of the drive wheel (16) is increased (the friction coefficient is increased to 0.8) to prevent slippage.

8. The large-model driven communication maintenance robot with multi-sensor detection for global adaptive operation according to claim 1, characterized in that: The energy storage battery (22) integrated on the detection substrate (21) is a lithium iron phosphate battery pack with a nominal voltage of 24V, a capacity of 20Ah, an energy density of 150Wh / kg, and a cycle life of ≥2000 cycles; the battery management system (BMS) supports overcharge (protection voltage 29.2V), over-discharge (protection voltage 20V), overcurrent (protection current 30A), and short-circuit protection; the solar-assisted charging system includes two 100W monocrystalline silicon solar panels (conversion efficiency 23%), and is equipped with an MPPT controller (tracking efficiency ≥99%), under standard illumination conditions (1000W / m 2 It can provide 150W charging power; the battery life meets the following requirements: 12 hours of continuous operation in an indoor constant temperature environment (25℃), 8 hours of continuous operation in an outdoor normal temperature environment (20℃~35℃), and 5 hours of continuous operation in a low temperature environment (-10℃); the battery insulation module adopts a polyimide heating film (power 50W), which automatically starts when the battery temperature is <5℃, maintains the battery operating temperature in the range of 10℃~45℃, and the heating response time is <3 minutes.

9. A large-model driven communication maintenance robot with global adaptive multi-sensor detection according to claim 1, characterized in that: The positioning system employs a multi-source fusion scheme of "LiDAR + Visual SLAM + Inertial Navigation". The LiDAR is a 16-line mechanical rotating radar (360° horizontal field of view, -15° to +15° vertical field of view, ranging range 0.1m to 100m, angular resolution 0.1°); the Visual SLAM module includes two fisheye cameras (190° field of view, 1280×800 resolution) and one depth camera (ranging range 0.5m to 10m, accuracy ±1%); the inertial navigation unit is a MEMSMU (three-axis accelerometer range ±16g, three-axis gyroscope range ±20g). The positioning accuracy is 00° / s, with zero bias stability ≤0.1° / h. The fusion positioning algorithm uses tightly coupled Kalman filtering. In areas with base station signals (GPS / BeiDou positioning accuracy 1m), the positioning error is ≤0.5m. In areas without base station signals, centimeter-level positioning (planar accuracy ±3cm, elevation accuracy ±5cm) is achieved through environmental feature matching (keyframe matching quantity ≥50 / frame). The positioning data is updated at a frequency of 10Hz and fused with the global communication heat map through the ROS system. The superimposed 3D scene is displayed on the control terminal, and the coordinate system adopts the UTM projection coordinate system.

10. A large-scale model-driven communication maintenance robot with global adaptive multi-sensor detection as described in claim 1, characterized in that: The maintenance mechanism (3) has a three-piece frame (311) made of aluminum alloy profile splicing structure, with 3 standard interfaces (compliant with ISO9409-1 robot quick-change interface standard), interface repeatability accuracy ±0.02mm, and equipped with a construction probe (313) to detect and monitor the operation of the gripper (319).

Images