Autonomous building robot intelligent monitoring system based on multi-sensor fusion

Abstract

The invention discloses an autonomous building robot intelligent monitoring system based on multi-sensor fusion, belongs to the field of intelligent monitoring, and aims to solve the problem that a traditional building robot scheme is difficult to adapt to a complex construction environment due to dependence on a single sensor and insufficient fusion algorithm. The system comprises a multi-source data acquisition system, an intelligent control system, an exception handling system and a mode switching system, the multi-source data collection system collects working environment data of the building robot through a main sensor, obtains a positioning result and transmits the positioning result to the intelligent control system, and the intelligent control system analyzes environment and task states in real time after receiving the data, autonomously decides and controls precise operation of the building robot, adaptively adjusts a decision instruction and cooperatively works, and controls the building robot to work. The exception handling system judges the exception level of the collected data and performs emergency processing, and the mode switching system monitors the state of the main sensor and switches to the standby sensor when the main sensor is exceptional.

Description

Technical Field

[0001] The present invention relates to the field of intelligent monitoring technology, and in particular to an autonomous construction robot intelligent monitoring system based on multi-sensor fusion. Background Art

[0002] With the rapid development of intelligent construction technology, construction robots have become a key enabler for construction automation. Current mainstream applications include: single-function robots, such as masonry robots (e.g., Hadrian X) and plastering robots (e.g., the HP series), which utilize pre-programmed robotic arms to perform a single process; navigation and positioning systems, which utilize lidar (e.g., the Trimble SPS986) or ultrawideband (UWB) solutions (e.g., the Decawave solution) for centimeter-level positioning; and quality inspection equipment, such as wall flatness inspection devices based on visual sensors (e.g., FLIR infrared cameras).

[0003] The industry is undergoing a transformation from "single-machine automation" to "group intelligence". According to Global Market Insights data, the market size of intelligent construction robots will reach US$5.4 billion in 2023, with an annual compound growth rate of 18.7%.

[0004] Currently, the existing technology has the following defects: (1) Traditional solutions often rely on a single sensor (such as pure vision or single-line laser radar), which is prone to the following problems in construction sites with dusty and changing lighting: point cloud loss (dust causes laser scattering), feature mismatching (visual SLAM fails on low-texture walls), and typical errors: MIT research shows that the median error of a single sensor in steel structure installation scenarios is as high as ±32mm; (2) Existing fusion algorithms (such as Kalman filtering) have difficulty in handling unstructured obstacles unique to construction scenes (scattered building materials causing point cloud mutations), dynamic interference (people moving around causing visual occlusion), and traditional EKF fusion produces a 17% misjudgment rate in the steel bar bundling scene. Summary of the Invention

[0005] In view of the shortcomings of the existing technology, the present invention proposes an autonomous construction robot intelligent monitoring system based on multi-sensor fusion, which includes a multi-source data acquisition system, an intelligent control system, an abnormality processing system and a mode switching system; The multi-source data acquisition system collects multi-source data from the working environment of the construction robot through sensors, obtains positioning results through the multi-source data, and transmits the multi-source data and positioning results to the intelligent control system; the sensors are each provided with two, namely a main sensor and a backup sensor; The intelligent control system receives multi-source data and positioning results transmitted by the multi-source data acquisition system; analyzes the environment and task status in real time by integrating the multi-source data and positioning results, and makes autonomous decisions to control the precise operation, adaptive adjustment, and collaborative operation of the construction robot in a complex construction environment; The exception handling system is used to determine the abnormality level of multi-source data received from the multi-source data acquisition system, obtain the abnormality level, and perform emergency processing based on the abnormality level; The mode switching system is used to monitor the output data and status of the main sensor in real time. If an abnormal output of the main sensor is detected, the fault judgment logic is triggered and the control signal is used to switch to the backup sensor.

[0006] As a further improvement of the present invention, the sensor includes: a laser radar LiDAR, a UWB sensor, an accelerometer, a gyroscope, a temperature sensor, a motor encoder, a force sensor, a dust sensor and a vibration sensor.

[0007] As a further improvement of the present invention, the multi-source data includes: point cloud data, visual sensor data, infrared ranging data, measurement unit data, joint torque data, positioning data, torque sensor data, temperature sensor data, vibration data, wireless communication data, time of flight ToF, angular velocity, linear acceleration, wheel rotation angular velocity and rotation speed, and dust concentration.

[0008] As a further improvement of the present invention, the sensors are each provided with two, namely a main sensor and a backup sensor, which are connected to the same data acquisition interface via a multiplexer MUX.

[0009] As a further improvement of the present invention, a laser SLAM module, a visual semantic module, a UWB positioning module, an IMU module, a displacement calculation module, a fusion module, a layering module and a unified navigation coordinate module; The laser SLAM module uses LiDAR to scan the environment, obtain high-precision point cloud data, and combines real-time positioning and map construction SLAM algorithm to generate an environmental map, which is transmitted to the visual semantic module and the fusion module; The visual semantic module uses a deep learning model to identify the semantic information of any object image collected by the camera set on the construction robot, builds a semantic map including doors, pedestrians, and obstacles, and combines it with the environmental map transmitted by the laser SLAM module to realize the semantic map construction and transmit it to the fusion module; The UWB positioning module calculates the position of the construction robot mobile device based on the time of flight ranging (ToF) of the ultra-wideband UWB radio signal through communication between multiple base stations and tags, and transmits it to the fusion module; The IMU module measures the angular velocity and linear acceleration of the construction robot through an accelerometer and a gyroscope, calculates the posture change of the construction robot over a period of time, and transmits the calculated posture change to the fusion module. The displacement calculation module measures the wheel rotation angle and speed through the wheel encoder, calculates the displacement of the construction robot, and transmits it to the fusion module; The fusion module receives the environment map transmitted by the laser SLAM module; receives the semantic map transmitted by the visual semantic module; receives the absolute position of the mobile device transmitted by the UWB module; receives the posture change transmitted by the IMU module; receives the displacement of the construction robot transmitted by the displacement calculation module; and obtains the positioning result of the construction robot by optimizing the environment map, the semantic map, the absolute position of the mobile device, the posture change, and the displacement of the construction robot, and transmits it to the hierarchical processing module; The hierarchical processing module is used to perform hierarchical processing on the positioning results received from the fusion module to balance the computing load, combine local high precision with global consistency, output navigation coordinates, and transmit them to the unified navigation coordinate module; The unified navigation coordinate module receives the navigation coordinates transmitted by the layered processing module and provides a globally consistent reference frame for each sensor to avoid coordinate confusion.

[0010] As a further improvement of the present invention, the hierarchical processing module includes a data preprocessing layer, a local positioning layer, a global positioning layer and a fusion output layer.

[0011] As a further improvement of the present invention, the intelligent control system further comprises: a storage and simulation module, a decision module, and an execution module; The storage and simulation module is used to store multi-source data, construction robot status data and historical task records in real time; for rapid retrieval and analysis; and to build a virtual construction environment based on the BIM model and multi-source data, and transmit it to the decision-making module; The decision module receives the virtual construction environment transmitted by the storage and simulation module; decomposes the task sequence according to the construction goal, combines the real-time environment map with the virtual construction environment, generates a global path and operation steps; replans the path in real time through reinforcement learning (RL); adjusts the trajectory of the robotic arm based on force feedback and visual deviation, generates decision instructions, and previews the motion path of the construction robot through the physics engine to perform collision detection and task feasibility verification, thereby optimizing the decision instructions; and transmits the decision instructions to the execution module; The execution module is used to convert the decision instructions received from the decision module into motor control signals, drive actuators such as robotic arms and mobile chassis, support high-precision position control and force control, and feed back the results of the execution instructions to the storage and simulation module for storage.

[0012] Beneficial effects of the present invention: Through the collaborative optimization of multi-sensor fusion, adaptive path planning and modular construction execution mechanisms, the accuracy, efficiency and safety of construction can be significantly improved, while reducing manpower dependence and construction costs.

[0013] By integrating LiDAR, dust sensors, IMU and other sensor data and adopting an adaptive weight distribution algorithm, a high-precision three-dimensional map of the construction environment is constructed in real time. Through the complementary nature of multiple sensors, positioning accuracy is improved by 80%. Even in harsh environments such as strong light and dust, it can still maintain stable mapping capabilities, with a measured point cloud matching success rate of >95%.

[0014] Combined with real-time point cloud + visual semantic segmentation, the path is dynamically adjusted to avoid collision accidents.

[0015] Compared with traditional workers and ordinary construction robots, construction accuracy is improved by 5-10 times, meeting the stringent requirements of super-high-rise / special-shaped buildings; overall efficiency is improved by 50% to 100%, shortening construction period and reducing labor costs; fully autonomous construction solves the labor shortage problem in the construction industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a structural diagram of the intelligent monitoring system of the autonomous construction robot based on multi-sensor fusion in this embodiment. DETAILED DESCRIPTION

[0017] To facilitate understanding of the present application, the present application will be described more fully below with reference to the accompanying drawings. The accompanying drawings provide embodiments of the description. Conversely, these embodiments are provided for the purpose of providing a more thorough and comprehensive understanding of the disclosure of the present application.

[0018] In this embodiment, the camera is an RGB-D camera; The present invention proposes an autonomous construction robot system based on multi-sensor fusion, which includes a multi-source data acquisition system, an intelligent control system, an abnormality processing system and a mode switching system; The multi-source data acquisition system collects multi-source data in the working environment of the construction robot through sensors, obtains positioning results through the multi-source data, and transmits the multi-source data and positioning results to the intelligent control system. The multi-source data acquisition system further includes: a laser SLAM module, a visual semantic module, an ultra-wideband UWB positioning module, an inertial measurement IMU module, a displacement calculation module, a fusion module, a layering module and a unified navigation coordinate module; The multi-source data acquisition system includes a laser radar LiDAR, a UWB sensor, an accelerometer, a gyroscope, a temperature sensor, a motor encoder, a force sensor, a dust sensor, and a vibration sensor to collect multi-source data in real time. The multi-source data includes: point cloud data, visual sensor data, infrared ranging data, measurement unit data, joint torque data, positioning data, torque sensor data, temperature sensor data, vibration data, wireless communication data, time of flight ranging (ToF), angular velocity, linear acceleration, wheel rotation angular velocity and rotation speed, and dust concentration; Among them, each of the above sensors is set to two, divided into a main sensor and a backup sensor, which are connected to the same data acquisition interface through a multiplexer MUX.

[0019] The laser SLAM module uses LiDAR to scan the environment, obtain high-precision point cloud data, and combines real-time positioning and map construction SLAM algorithm to generate an environmental map, which is transmitted to the visual semantic module and the fusion module; The visual semantic module uses a deep learning model to identify the semantic information of any object image collected by the camera set on the construction robot, builds a semantic map including doors, pedestrians, and obstacles, and combines it with the environmental map transmitted by the laser SLAM module to realize the semantic map construction and transmit it to the fusion module; The UWB positioning module calculates the position of the construction robot mobile device based on the time of flight ranging (ToF) of the ultra-wideband UWB radio signal through communication between multiple base stations and tags, and transmits it to the fusion module; The IMU module measures the angular velocity and linear acceleration of the construction robot through an accelerometer and a gyroscope, calculates the posture change of the construction robot over a period of time, and transmits the calculated posture change to the fusion module. The displacement calculation module measures the wheel rotation angle and speed through the wheel encoder, calculates the displacement of the construction robot, and transmits it to the fusion module; The fusion module receives the environment map transmitted by the laser SLAM module; receives the semantic map transmitted by the visual semantic module; receives the absolute position of the mobile device transmitted by the UWB module; receives the posture change transmitted by the IMU module; receives the displacement of the construction robot transmitted by the displacement calculation module; and obtains the positioning result of the construction robot by optimizing the environment map, the semantic map, the absolute position of the mobile device, the posture change, and the displacement of the construction robot, and transmits it to the hierarchical processing module; The hierarchical processing module is used to perform hierarchical processing on the positioning results received from the fusion module to balance the computing load, combine local high precision with global consistency, output navigation coordinates, and transmit them to the unified navigation coordinate module; the hierarchical processing system further includes: a data preprocessing layer, a local positioning layer, a global positioning layer, and a fusion output layer; The data preprocessing layer uses the PTP protocol to ensure the alignment of sensor data timestamps, unifies laser / visual / UWB data into the carrier coordinate system by calibrating external parameters, and dynamically updates the coordinate origin; filters abnormal data from each sensor based on the 3σ criterion; and uses the IQR algorithm to eliminate transient interference; The local positioning layer maintains positioning even when any sensor fails through laser-vision tightly coupled SLAM: the front end uses ORB / SIFT to match visual feature points with laser edge features; the back end uses Ceres Solver for joint optimization (reprojection error + point cloud matching error). It utilizes multimodal sensor assistance, IMU pre-integration to provide motion priors, and encoder odometry as a short-term motion constraint. In the local positioning layer, for masonry structures: add mortar joint edge detection (based on the normal vector mutation threshold θ>35°); for steel structures: use curvature clustering to eliminate bolt point cloud interference; The global positioning layer optimizes base station layout based on GDOP geometric precision factor; dynamic power adjustment resists multipath interference; uses PointNet+ to process laser point cloud semantic segmentation and visual-semantic feature graph optimization matching; The fusion output layer uses adaptive Kalman filtering and dynamic noise covariance matrix adjustment (online estimation of Q / R matrix); multi-model switching for different motion states; sensor failure detection based on chi-square test; and missing sensor prediction based on historical data. Through the above layers, high-precision fusion positioning of multi-source sensor data is achieved, solving the problem of inconsistent timestamps and coordinate systems of any sensor data, the contradiction between the local high precision of laser SLAM and the global calibration of UWB, and maintaining stable output even when some sensors fail.

[0020] The unified navigation coordinate module receives the navigation coordinates transmitted by the layered processing module and provides a globally consistent reference frame for each sensor to avoid coordinate confusion; The intelligent control system receives multi-source data and positioning results transmitted by the multi-source data acquisition system; analyzes the environment and task status in real time by integrating the multi-source data and positioning results, and makes autonomous decisions to control the construction robot's precise operation, adaptive adjustment and efficient collaborative operation in a complex construction environment; the intelligent control system further includes: a storage and simulation module, a decision module, and an execution module.

[0021] The storage and simulation module is used to store multi-source data, construction robot status data and historical task records in real time; for rapid retrieval and analysis; and to build a virtual construction environment based on the BIM (Building Information Modeling) model and multi-source data, and transmit it to the decision-making module; The decision module receives the virtual construction environment transmitted by the storage and simulation module; decomposes the task sequence according to the construction goal, combines the real-time environment map with the virtual construction environment, generates a global path and operation steps; replans the path in real time through reinforcement learning (RL); adjusts the trajectory of the robotic arm based on force feedback and visual deviations (such as uneven wall surfaces), generates decision instructions, and previews the construction robot's motion path through a physics engine to perform collision detection and task feasibility verification, thereby optimizing the decision instructions; and transmits the decision instructions to the execution module; The execution module is used to convert the decision instructions received from the decision module into motor control signals, drive actuators such as robotic arms and mobile chassis, support high-precision position control and force control, and feed back the results of the execution instructions to the storage and simulation module for storage.

[0022] The exception handling system is used to determine the abnormality level of multi-source data received from the multi-source data acquisition system, obtain the abnormality level, and perform emergency processing based on the abnormality level; According to the severity of the abnormality, it is divided into three levels: level one abnormality, level two abnormality and level three abnormality. Corresponding measures are taken according to the level of abnormality; The first-level abnormality includes: temporary sensor data exceeding the limit, that is, occasional noise in the temperature sensor, or detection of temperature sensor abnormality but determined to be self-recoverable. In this case, an alarm prompt is issued to notify maintenance personnel to handle it; The alarm prompt displays a yellow warning on the monitoring platform, records a log, saves the abnormal time, value and context data; and continuously monitors the abnormality of the temperature sensor. If it does not recover in a short period of time, the judgment level is raised to level 2 abnormality.

[0023] The secondary abnormality includes: the key sensor data is continuously abnormal, that is, the motor encoder signal is unstable, but it can still partially operate. At this time, an alarm prompt is issued, a blue warning is displayed on the monitoring platform, and maintenance personnel are notified to handle it to avoid deterioration; The three-level abnormality includes: complete sensor failure, that is, pressure sensor disconnection and personal danger. At this time, an alarm prompt is issued, a red warning is displayed on the monitoring platform, and maintenance personnel are notified to check to avoid deterioration; In addition, the hardware protection of the construction robot adopts hardware safety protection design, and the mechanical body protection adopts anti-collision structure: omnidirectional buffer layer, using EPP material, thickness ≥ 50mm, collapsible joint design, trigger force threshold 500N; For monitoring of building scenes: In normal operation, environmental parameters are within the safety threshold, that is, dust concentration ≤ 50mg / m³, vibration amplitude < 4mm / s; all sensor data are in a stable fluctuation state; the system outputs a green operation indicator light; Real-time monitoring indicators: real-time dust concentration curve 20 ~ 40mg / m³ fluctuation, vibration spectrum analysis main frequency band 1 ~ 100Hz, amplitude 3.2mm / s, equipment temperature 65±5℃; If the main dust sensor continuously detects >50mg / m³ for 3 seconds and the backup dust sensor verifies the difference <15%; Start the graded response. When the real-time dust concentration curve fluctuates between 50 and 80 mg / m³, the first-level response is triggered. The system outputs a blue indicator light, prompting maintenance personnel to start the second-level filtration system and increase the ventilation volume to 120%. When the real-time dust concentration curve is greater than 80mg / m³, a secondary response is triggered. At this time, the system outputs a red indicator light to prompt maintenance personnel to close the raw material input valve, trigger the sound and light alarm, and activate the emergency brake mode; This hierarchical response strategy ensures both operational safety and production continuity, in compliance with the GB / T5226.1-2019 standard for mechanical and electrical safety. All emergency response delays are controlled within 500ms, and critical braking actions meet the SIL2 safety level.

[0024] The mode switching system monitors the output data and status of the main sensor in real time. Once an abnormal output of the main sensor is detected, the fault judgment logic is triggered and the system switches to the backup sensor through a control signal.

[0025] The specific switching methods for different sensors are as follows: LiDAR: Monitor the lidar's scanning frequency, point cloud density, and communication status.

[0026] If the primary lidar fails, switch to the backup lidar and recalibrate its position and angle.

[0027] UWB Sensor: Detect the strength, positioning accuracy and synchronization status of the UWB signal, and switch to the backup sensor when the main UWB sensor fails.

[0028] The accelerometer and gyroscope are integrated into the IMU module to monitor whether the acceleration and angular velocity data they output are reasonable. If the main IMU fails, it switches to the backup accelerometer and gyroscope and recalibrates the zero bias.

[0029] Temperature sensor: Monitor whether the temperature reading is within a reasonable range and whether there is a jump or communication failure. If the main temperature sensor fails, switch to the backup sensor and perform data calibration.

[0030] Motor encoder: Monitor the encoder's pulse output, resolution, and communication status. If the primary encoder fails, switch to the backup encoder via hardware switching relays or software instructions.

[0031] Force sensor: Check whether the output of the force sensor is stable, whether there is abnormal fluctuation or communication failure. When the main sensor fails, switch to the backup sensor and recalibrate the zero point.

[0032] Dust sensor: Monitor dust concentration readings for plausibility, jumps, or communication failures, and switch to backup sensors when the primary sensor fails.

[0033] Vibration sensor: Detect the frequency, amplitude and communication status of the vibration signal, and switch to the backup sensor when the main sensor fails.

[0034] After switching to the backup sensor, perform a calibration to ensure its output data is consistent with the primary sensor.

[0035] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the patent for this invention shall be determined by the appended claims.

Claims

1. An intelligent monitoring system for autonomous construction robots based on multi-sensor fusion, characterized in that: Including multi-source data acquisition system, intelligent control system, exception handling system and mode switching system; The multi-source data acquisition system collects multi-source data from the working environment of the construction robot through sensors, obtains positioning results through the multi-source data, and transmits the multi-source data and positioning results to the intelligent control system; the sensors are each provided with two, namely a main sensor and a backup sensor; The intelligent control system receives multi-source data and positioning results transmitted by the multi-source data acquisition system; analyzes the environment and task status in real time by integrating the multi-source data and positioning results, and makes autonomous decisions to control the precise operation, adaptive adjustment, and collaborative operation of the construction robot in a complex construction environment; The exception handling system is used to determine the abnormality level of multi-source data received from the multi-source data acquisition system, obtain the abnormality level, and perform emergency processing based on the abnormality level; The mode switching system is used to monitor the output data and status of the main sensor in real time. If an abnormal output of the main sensor is detected, the fault judgment logic is triggered and the control signal is used to switch to the backup sensor.

2. The intelligent monitoring system for autonomous construction robots based on multi-sensor fusion according to claim 1 is characterized in that: The sensors include: LiDAR, UWB sensor, accelerometer, gyroscope, temperature sensor, motor encoder, force sensor, dust sensor and vibration sensor.

3. The intelligent monitoring system for autonomous construction robots based on multi-sensor fusion according to claim 2 is characterized in that: The multi-source data includes: point cloud data, visual sensor data, infrared ranging data, measurement unit data, joint torque data, positioning data, torque sensor data, temperature sensor data, vibration data, wireless communication data, time of flight ToF, angular velocity, linear acceleration, wheel rotation angular velocity and rotation speed, and dust concentration.

4. The intelligent monitoring system for autonomous construction robots based on multi-sensor fusion according to claim 3 is characterized in that: The sensors are each configured as two, namely a main sensor and a backup sensor, which are connected to the same data acquisition interface via a multiplexer MUX.

5. The intelligent monitoring system for autonomous construction robots based on multi-sensor fusion according to claim 1 is characterized in that: The multi-source data acquisition system further includes: a laser SLAM module, a visual semantic module, a UWB positioning module, an IMU module, a displacement calculation module, a fusion module, a layering module and a unified navigation coordinate module; The laser SLAM module uses LiDAR to scan the environment, obtain high-precision point cloud data, and combines real-time positioning and map construction SLAM algorithm to generate an environmental map, which is transmitted to the visual semantic module and the fusion module; The visual semantic module uses a deep learning model to identify the semantic information of any object image collected by the camera set on the construction robot, builds a semantic map including doors, pedestrians, and obstacles, and combines it with the environmental map transmitted by the laser SLAM module to realize the semantic map construction and transmit it to the fusion module; The UWB positioning module calculates the position of the construction robot mobile device based on the time of flight ranging (ToF) of the ultra-wideband UWB radio signal through communication between multiple base stations and tags, and transmits it to the fusion module; The IMU module measures the angular velocity and linear acceleration of the construction robot through an accelerometer and a gyroscope, calculates the posture change of the construction robot over a period of time, and transmits the calculated posture change to the fusion module. The displacement calculation module measures the wheel rotation angle and speed through the wheel encoder, calculates the displacement of the construction robot, and transmits it to the fusion module; The fusion module receives the environment map transmitted by the laser SLAM module; receives the semantic map transmitted by the visual semantic module; receives the absolute position of the mobile device transmitted by the UWB module; receives the posture change transmitted by the IMU module; receives the displacement of the construction robot transmitted by the displacement calculation module; and obtains the positioning result of the construction robot by optimizing the environment map, the semantic map, the absolute position of the mobile device, the posture change, and the displacement of the construction robot, and transmits it to the hierarchical processing module; The hierarchical processing module is used to perform hierarchical processing on the positioning results received from the fusion module to balance the computing load, combine local high precision with global consistency, output navigation coordinates, and transmit them to the unified navigation coordinate module; The unified navigation coordinate module receives the navigation coordinates transmitted by the layered processing module and provides a globally consistent reference frame for each sensor to avoid coordinate confusion.

6. The intelligent monitoring system for autonomous construction robots based on multi-sensor fusion according to claim 5 is characterized in that: The hierarchical processing module includes a data preprocessing layer, a local positioning layer, a global positioning layer and a fusion output layer.

7. The intelligent monitoring system for autonomous construction robots based on multi-sensor fusion according to claim 1 is characterized in that: The intelligent control system further includes: a storage and simulation module, a decision module, and an execution module; The storage and simulation module is used to store multi-source data, construction robot status data and historical task records in real time; for rapid retrieval and analysis; and to build a virtual construction environment based on the BIM model and multi-source data, and transmit it to the decision-making module; The decision module receives the virtual construction environment transmitted by the storage and simulation module; decomposes the task sequence according to the construction goal, combines the real-time environment map with the virtual construction environment, generates a global path and operation steps; replans the path in real time through reinforcement learning (RL); adjusts the trajectory of the robotic arm based on force feedback and visual deviation, generates decision instructions, and previews the motion path of the construction robot through the physics engine to perform collision detection and task feasibility verification, thereby optimizing the decision instructions; and transmits the decision instructions to the execution module; The execution module is used to convert the decision instructions received from the decision module into motor control signals, drive actuators such as robotic arms and mobile chassis, support high-precision position control and force control, and feed back the results of the execution instructions to the storage and simulation module for storage.

Images