A control method for the five-finger dexterous hand of an intelligent bomb disposal robot

Abstract

The present invention discloses a control method for a five-finger dexterous hand of an intelligent bomb disposal robot, which relates to the technical field of program-controlled manipulators. The method comprises calibrating the five-finger dexterous hand, collecting and preprocessing surrounding environmental data using sensors, analyzing the data using image processing and pattern recognition algorithms to obtain the surrounding environmental data and the location, shape, and characteristics of the explosive; using a machine learning algorithm to plan a safe path from the five-finger dexterous hand to the explosive location based on the surrounding environmental data and the explosive's location, shape, and characteristics; and moving the five-finger dexterous hand near the explosive along the planned safe path. After the five-finger dexterous hand has been moved near the explosive location, a machine learning algorithm is used to generate an action sequence. Based on the generated action sequence, high-precision force feedback is used to grasp the explosive. After grasping, the bomb disposal action is executed. By performing precise calibration steps on the five-finger dexterous hand, the present invention achieves precise control of the movements of each finger.

Description

Technical Field

[0001] The present invention relates to the technical field of program-controlled manipulators, in particular to a control method for a five-finger dexterous hand of an intelligent bomb disposal robot. Background Art

[0002] In the development of intelligent bomb disposal robots, the control method of five-fingered dexterous hands is one of the key research areas. Recent advances in robotics and artificial intelligence, particularly the application of computer vision, machine learning algorithms, and high-precision force feedback systems, have enabled robots to perform more complex and delicate tasks.

[0003] In practice, traditional bomb disposal robots rely on fixed operating procedures, resulting in limited flexibility and adaptability, making them incapable of effectively handling complex and changing on-site environments. This is especially true when dealing with explosives of varying shapes, where traditional mechanical structures and control methods often lack sufficient precision and safety, resulting in a low mission success rate. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a control method for the five-finger dexterous hand of an intelligent bomb disposal robot to solve the problems of inaccurate path planning, lack of flexibility in grasping action and insufficient real-time monitoring of existing bomb disposal robots in complex environments.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for controlling a five-finger dexterous hand of an intelligent bomb disposal robot, comprising calibrating the five-finger dexterous hand, collecting and preprocessing surrounding environment data using sensors, and analyzing the data through image processing and pattern recognition algorithms to obtain the surrounding environment data and the location, shape, and characteristics of the explosive;

[0008] Based on the surrounding environment data and the location, shape, and characteristics of the explosive, a machine learning algorithm is used to plan a safe path for the five-fingered dexterous hand to reach the explosive. After the planning is complete, the five-fingered dexterous hand is moved near the explosive along the planned safe path.

[0009] After the five-fingered dexterous hand is moved near the location of the explosive, a machine learning algorithm is used to generate an action sequence. Based on the generated action sequence, high-precision force feedback is used to grasp the explosive. After the grasp is completed, the bomb disposal action is performed;

[0010] After performing the bomb disposal action, the status of the smart material is monitored in real time through sensors, providing a real-time video stream to the operator. After completing the task, the robot is returned to the safe area for a comprehensive inspection, any abnormalities are recorded, and maintenance is carried out.

[0011] As a preferred solution of the control method of the five-finger dexterous hand of the intelligent bomb disposal robot of the present invention, wherein: the five-finger dexterous hand is calibrated, the surrounding environment data is collected by sensors, and pre-processed, including the following steps:

[0012] Start the sensor array, communication device and controller, run the hardware and software self-test program, perform physical and electrical calibration of the five-finger dexterous hand, and calibrate the force sensor and other sensors;

[0013] Activate the sensor installed on the robot. After activation, the sensor starts to collect data about the surrounding environment and obtains the surrounding environment data.

[0014] The collected surrounding environment data is filtered, denoised and format converted to obtain pre-processed surrounding environment data.

[0015] As a preferred solution of the five-finger dexterous hand control method of the intelligent bomb disposal robot of the present invention, wherein: the data is analyzed by image processing and pattern recognition algorithm to obtain the surrounding environment data and the position, shape and characteristics of the explosive, including the following steps:

[0016] Use image processing algorithms to extract location information, shape information, size information, material information, background information, and posture information from the pre-processed surrounding environment data, and apply pattern recognition algorithms to analyze the pre-processed surrounding environment data to determine the location, shape, and characteristics of the surrounding data and the explosives;

[0017] Generate a 3D model and map of the surrounding environment and target objects based on the environmental data and the location, shape, and characteristics of the explosives, expressed as,

[0018]

[0019] Where M is the generated three-dimensional model, V is the integration area, p is any point in the environment, D(p) is the distance function, α is the attenuation coefficient, is the feature extraction function, is the shape matching function, N is the number of known feature points in the environment, i is the index of the feature point, and w i is the importance weight of the i-th feature point, β is the three-dimensional model enhancement coefficient, It is a complex information filtering function.

[0020] As a preferred embodiment of the control method of the five-finger dexterous hand of the intelligent bomb disposal robot of the present invention, the method includes: planning a safe path from the five-finger dexterous hand to the explosive location using a machine learning algorithm based on the surrounding environment data and the location, shape and characteristics of the explosive; and moving the five-finger dexterous hand to the vicinity of the explosive along the planned safe path after the planning is completed, including the following steps:

[0021] Based on the generated 3D model and map of the surrounding environment and target object, the trained machine learning model is loaded, the surrounding environment data and explosive feature information are input into the machine learning model, and the current state of the robot is input to obtain the constructed machine learning model;

[0022] A machine learning model is used to simulate different approach paths. Combined with energy-aware adaptive motion planning, the system takes into account the current battery state and optimizes the path to minimize energy consumption. Furthermore, it considers safety factors and plans a safe path for the five-fingered dexterous hand to reach the explosive through iterative optimization.

[0023] The five-finger dexterous hand starts its motion execution program, moves step by step along the planned safe path, monitors the execution of each action in real time, and continuously monitors the applied force through high-precision force feedback.

[0024] As a preferred solution of the control method of the five-finger dexterous hand of the intelligent bomb disposal robot of the present invention, after the five-finger dexterous hand is moved to the vicinity of the explosive location, a machine learning algorithm is used to generate an action sequence, including the following steps:

[0025] After moving the five-fingered dexterous hand to the designated location according to the planned path, the machine learning model is loaded and the model parameters are verified;

[0026] The collected environmental data and explosive feature information are input into the machine learning model, and the machine learning model is used to simulate different grasping and manipulation actions to generate action sequences, which are expressed as,

[0027]

[0028] Where A is the generated action sequence, T is the total time period, and λ is the time decay coefficient. is the machine learning model output function, is a security assessment function, K is the number of known key points in the environment, k is the index of the key point, γ is the action sequence enhancement coefficient, is the information filtering function, a k is the action parameter of the kth key point;

[0029] The finalized motion sequence is loaded into the robot's embedded control panel.

[0030] As a preferred solution of the control method of the five-finger dexterous hand of the intelligent bomb disposal robot of the present invention, wherein: according to the generated action sequence, high-precision force feedback is used to grab the explosive, and after the grabbing is completed, the bomb disposal action is performed, including the following steps:

[0031] Load the generated motion sequence from the control panel, activate the high-precision force sensor, and use the vision sensor to continuously monitor the position deviation based on the first motion of the motion sequence, fine-tune the path to ensure accurate positioning, and monitor the contact force in real time through high-precision force feedback during the approach process;

[0032] Use force feedback data to fine-tune the dexterous hand's posture and strength, dynamically adjusting the finger's opening and closing angles and pressure based on the object's shape and surface features. Based on the grasping instructions in the action sequence, high-precision force feedback is used to precisely control the strength of each finger for grasping.

[0033] After completing the grasping, high-precision force feedback is used to fine-tune the posture of the five-fingered dexterous hand to better perform bomb disposal actions, transfer the explosives to a safe area, implement harmless treatment and dismantle the device.

[0034] As a preferred solution of the five-finger dexterous hand control method of the intelligent bomb disposal robot of the present invention, wherein: after the bomb disposal action is performed, the state of the intelligent material is monitored in real time by the sensor, and a real-time video stream is provided to the operator, including the following steps:

[0035] After executing the bomb disposal action, the sensors installed on the five-fingered dexterous hand and the robot continuously collect data on the state of the smart material to obtain the smart material state data, and the smart material state data is preprocessed to obtain the preprocessed smart material state data;

[0036] Use machine learning models to analyze pre-processed smart material status data to identify abnormal patterns and potential problems;

[0037] When an abnormal situation is detected, the alarm mechanism is immediately triggered to notify the operator and take protective measures. All abnormal events and their handling process are recorded;

[0038] Activate cameras and visual sensors to capture real-time images of the operation site, use compression technology to reduce bandwidth usage, and transmit video streams to the operator's monitoring terminal, allowing the operator to monitor progress in real time and intervene when necessary.

[0039] As a preferred solution of the five-finger dexterous hand control method of the intelligent bomb disposal robot of the present invention, wherein: after completing the task, the robot is returned to the safe area, a comprehensive inspection of the robot is carried out, any abnormalities are recorded, and maintenance is carried out, including the following steps:

[0040] When the mission is completed, the robot's return program is started, the navigation panel is activated, and the collected environmental data and the latest sensor information are used to recalculate a safe path back to the safe zone, generate path planning instructions, and send them to the motion control panel;

[0041] Control the robot to move to the designated safety zone according to the planned safety path, start the comprehensive inspection program, activate all self-test functions, generate a detailed inspection report, and submit it to the relevant technicians and management department. Based on the inspection results, a specific maintenance plan will be formulated;

[0042] After completing all maintenance work, all inspection records, abnormality reports and maintenance logs will be archived to establish a complete task file, which will be stored in the storage medium and cloud server.

[0043] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the method for controlling the five-finger dexterous hand of an intelligent bomb disposal robot as described in the first aspect of the present invention is implemented.

[0044] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the method for controlling the five-finger dexterous hand of an intelligent bomb disposal robot as described in the first aspect of the present invention is implemented.

[0045] The beneficial effects of the present invention are as follows: by performing precise calibration steps on the five-fingered dexterous hand, precise control of the movements of each finger is achieved; by utilizing a highly sensitive sensor network to collect data on the surrounding environment and preprocessing it, the position, shape and characteristics of the explosive can be quickly and accurately determined; through the previously obtained environmental and target information, a machine learning algorithm is used to plan a safe path from the current position to the explosive, thereby achieving efficient approach to the target; by calling the machine learning algorithm to generate a series of precise action instructions, the dexterous hand is guided to complete the capture of the explosive, and fine-tuning of the grasping force is achieved to avoid danger caused by excessive force. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0047] Figure 1 This is a flow chart of the control method of the five-finger dexterous hand of the intelligent bomb disposal robot in Example 1.

[0048] Figure 2 This is a flowchart of the five-finger dexterous hand grasping in Example 1. DETAILED DESCRIPTION

[0049] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0050] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0051] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0052] Example 1, with reference to Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a method for controlling the five-finger dexterous hand of an intelligent bomb disposal robot, comprising the following steps:

[0053] S1. Calibrate the five-finger dexterous hand, use sensors to collect surrounding environment data, and perform preprocessing. The specific steps are as follows:

[0054] Turn on the robot's main power switch, ensure that all components have a stable power supply, activate all types of sensors, including but not limited to cameras, lidar, infrared sensors, and temperature sensors, check their connection status, start wireless and wired communications, ensure that the communication link with the operator console is normal, activate the embedded control panel, and load the operator panel and control software;

[0055] Automatically run hardware diagnostic programs to check the status of various hardware components, including but not limited to the CPU, memory, storage, and sensor interfaces, confirm that all hardware is functioning properly, record any abnormalities and generate reports, load and run software self-test scripts to verify that all applications and services are properly launched, and check system logs to confirm that there are no error prompts or warning messages;

[0056] Place the five-fingered dexterous hand in a standard position, use precision measuring tools to measure the angles and displacements of each joint, adjust the zero point setting of the robotic arm to ensure that the initial position of each joint is accurate, calibrate the motor driver and encoder to ensure that they can accurately reflect the position changes of the joints, and calibrate the motor driver and encoder to ensure that they can accurately reflect the position changes of the joints;

[0057] Perform multi-point calibration of force sensors using standard weights and force gauges to ensure accurate readings throughout their entire range. Record calibration results and update sensor calibration parameters. Perform zero-point and span calibration of temperature, humidity, and distance sensors to ensure reliable data under various environmental conditions. Verify sensor accuracy using calibration certificates and standard reference values.

[0058] Activate all sensors installed on the robot in sequence, ensuring that each sensor is in working condition, verify the sensor's communication interface to ensure normal data transmission, set the sensor's working mode (including but not limited to continuous collection and timed collection), and prepare to start collecting data about the surrounding environment;

[0059] Start acquiring real-time data from various sensors, including but not limited to visual information, distance measurement, temperature, and humidity, and temporarily store the raw data in a local cache;

[0060] Apply digital filters to remove high-frequency noise and smooth the signal curve. Use advanced noise reduction algorithms to further reduce random noise and improve data quality. Convert data from different sensors into a format suitable for subsequent processing and ensure that the timestamps of all data are synchronized.

[0061] S2. Analyzing data through image processing and pattern recognition algorithms to obtain surrounding environment data and the location, shape, and characteristics of the explosive, including the following steps:

[0062] Load pre-processed surrounding data, initialize image processing libraries and tools, apply Canny edge detection and Sobel operators to highlight edges in the image, extract object outlines using contour finding algorithms, and preliminarily separate foreground and background. Based on grayscale or color thresholds, the image is divided into foreground and background. Starting from a seed point, similar pixels are gradually merged to form connected regions. Water flow simulation is used to separate different regions. This approach is particularly suitable for object segmentation in complex backgrounds.

[0063] Apply pattern recognition algorithms to calculate the center position of objects in images using a coordinate system, detect lines and circles using the Hough transform, describe object shapes, measure length, width, and height, and calculate their area and perimeter. Obtain surface characteristics through color histogram and texture analysis, identify and mark non-target areas, provide contextual information, and estimate the orientation and angle of objects using PCA and geometric methods. All extracted information is then integrated into a structured data format.

[0064] Use depth sensors and stereo vision technology to generate point cloud data to represent the three-dimensional point set in the environment. Apply Delaunay triangulation and Poisson reconstruction algorithms to convert the point cloud into a continuous three-dimensional grid. Combine RGB images, depth maps and perception data to generate a complete three-dimensional model. Use SLAM technology to construct a three-dimensional map containing the environment and target objects, which is expressed as:

[0065]

[0066] Where M is the generated three-dimensional model, V is the integration area, which is used to represent the volume of the entire environment space, p is any point in the environment, which is used to describe every possible position in the environment, D(p) is the distance function, which is used to represent the distance from point p to the nearest known environmental feature point, and α is the attenuation coefficient, which is used to control the degree of influence of distance on density. It is a feature extraction function used to extract the features of each point in the environment. (p) is a shape matching function used to evaluate the similarity between point p and the shapes in the known shape library. N is the number of known feature points in the environment, which is used to represent the number of key points detected by the sensor. i is the index of the feature point, which is used to process each feature point in the summation operation. w i is the importance weight of the i-th feature point, which is used to dynamically adjust according to the type of feature point. β is the three-dimensional model enhancement coefficient, which is used to adjust the influence of the feature point on the overall model. It is a complex information filtering function used to process data from multi-source sensors.

[0067] S3. Based on the surrounding environment data and the location, shape, and characteristics of the explosive, a machine learning algorithm is used to plan a safe path from the five-fingered dexterous hand to the location of the explosive. After the planning is completed, the five-fingered dexterous hand is moved to the vicinity of the explosive along the planned safe path, including the following steps:

[0068] Collect surrounding environment data, including but not limited to three-dimensional maps, obstacle locations, terrain characteristics, collect explosive samples of different types, shapes, sizes, and materials, and record their location, shape, size, and surface characteristics; and record the robot's operational data in different scenarios, including but not limited to path planning, motion sequences, and force feedback information;

[0069] Select a machine learning algorithm based on the task requirements and design the model architecture, including the design of the input layer, hidden layer, and output layer, as well as the connection method between each layer. For reinforcement learning models, define the state space, action space, and reward function, initialize model parameters, set initial weights and biases, and divide the dataset into training, validation, and test sets, usually with a ratio of 70%, 15%, and 15%. Select a loss function and optimizer, use batch gradient descent and mini-batch gradient descent for training, gradually adjust model parameters to minimize the loss function, and continuously optimize model performance through multiple iterations until the predetermined convergence condition and maximum number of iterations are reached;

[0070] Evaluate model performance on the validation set, use grid search, random search, and Bayesian optimization methods to tune hyperparameters, and use k-fold cross validation to further verify the stability and generalization ability of the model;

[0071] Load a pre-trained machine learning model, check the model parameters to ensure they are appropriate for the current task environment and target object characteristics, and integrate the machine learning model into the robot control panel to ensure that the model can receive sensor data and output path instructions;

[0072] Loading the generated three-dimensional model and map of the surrounding environment and target object into the machine learning model, inputting data collected from the environment, including but not limited to the location of obstacles, terrain characteristics, the specific location, shape, size and surface characteristics of the explosives, and the current state of the robot so that the model can take into account physical limitations and energy optimization;

[0073] The path planning module starts and prepares to simulate different approach paths. It uses a machine learning model to simulate multiple possible approach paths, evaluates the safety and feasibility of each path, and optimizes the path to minimize energy consumption based on the current battery level, taking safety factors into account and avoiding obstacles and other potential risk points. Through iterative optimization, it selects the optimal path, ensuring that the path is both safe and efficient in reaching the target location.

[0074] Output the finalized safe path, including the starting position, intermediate nodes, target position, obstacle avoidance strategy, and emergency response. Simulate the generated safe path in a virtual environment to observe whether the entire process is smooth and risk-free. If any potential problems or unreasonable paths are found, return to the previous step for adjustment and optimization.

[0075] Load the finalized safe path into the robot's embedded control panel, confirm that all path instructions are correct, prepare for actual execution, and set up monitoring mechanisms to ensure that the path can be adjusted in real time during execution;

[0076] The five-finger dexterous hand starts its motion execution program and moves step by step along the planned safe path. The execution of each movement is monitored in real time, and fine-tuned according to the actual situation to ensure the safety and accuracy of the operation. The applied force is continuously monitored through high-precision force feedback to ensure that no unnecessary pressure is placed on the environment or the user. During the movement, synchronization with other sensors is maintained to ensure the accuracy of the path.

[0077] When the five-fingered dexterous hand successfully moves to the predetermined position near the explosive, it stops moving and confirms that the position is accurate, and performs a final environmental perception and verification to ensure that the dexterous hand is at the optimal operating distance and posture.

[0078] S4. After moving the five-fingered dexterous hand to the location of the explosive, a machine learning algorithm is used to generate an action sequence, including the following steps:

[0079] The five-fingered dexterous hand starts its motion execution program and moves step by step along a pre-planned safe path. The execution of each action is monitored in real time, and fine-tuned based on actual conditions to ensure safety and accuracy. The applied force is continuously monitored through a high-precision force feedback system to ensure no unnecessary stress is placed on the environment or the hand itself. Once the five-fingered dexterous hand successfully moves to the predetermined position near the explosive, it stops and confirms that the position is correct.

[0080] Start model loading, prepare to load a pre-trained machine learning model, load the model file from the storage medium, including the model architecture and weight parameters, verify the integrity of the model file to ensure that no key parts are damaged or missing, check the model parameters to ensure that they are suitable for the current task environment and target object characteristics, and adjust the model parameters if necessary to adapt to the new environment or task requirements;

[0081] Activate all relevant sensors to recollect data on the current environment, including but not limited to the latest three-dimensional map, obstacle locations, terrain characteristics, and the specific location, shape, size, and surface characteristics of the explosives; pre-process the collected data; and input the pre-processed environmental data and explosive characteristics information into the machine learning model, along with the robot's current state, so that the model can take into account physical limitations and energy optimization;

[0082] Start motion planning and prepare to simulate different grasping and manipulation actions. Use machine learning models to simulate multiple possible grasping actions, evaluate the safety and feasibility of each action, and simulate subsequent manipulation actions to ensure that each step meets safety standards and energy optimization requirements.

[0083] According to the current power state, the action sequence is optimized to minimize energy consumption, taking safety factors into consideration to avoid causing additional pressure on explosives and other risks. Through iterative optimization, the optimal action sequence is selected to ensure that the action is both safe and can complete the task efficiently. The final action sequence is output, including the approach path, grasping action, manipulation action, safety checkpoints and emergency response strategy, which is expressed as,

[0084]

[0085] Among them, A is the generated action sequence, T is the total time period, which is used to represent the time range of the entire action sequence, and λ is the time decay coefficient, which is used to control the degree of influence of different time periods on the final action sequence. It is the output function of the machine learning model, which is used to represent the action suggestions or feature extraction results output by the machine learning model based on the current environment and target information. is a safety assessment function, which is used to express the safety score based on the current environment data and operation risk assessment. K is the number of known key points in the environment, which is used to enhance the safety and accuracy of the action sequence. k is the index of the key point, which is used to process each key point in the summation operation. γ is the action sequence enhancement coefficient, which is used to adjust the degree of influence of the key point on the overall action sequence. is an information filtering function used to process data from multiple source sensors, a k is the action parameter of the kth key point, which is used to represent the role and influencing factors of the key point in the action sequence;

[0086] S5. According to the generated action sequence, high-precision force feedback is used to grab the explosives. After the grab is completed, the bomb disposal action is performed, including the following steps:

[0087] Start the robot's embedded control panel and prepare to load the generated action sequence. Load the pre-generated action sequence file from the control panel, including but not limited to timestamps and action parameters.

[0088] Start the high-precision force sensor and other related sensors to ensure they are working properly, calibrate the force sensor to ensure its readings are accurate;

[0089] According to the first action instruction in the action sequence, the five-finger dexterous hand starts moving. The visual sensor continuously monitors the position deviation of the dexterous hand. The position error is calculated in real time through image processing algorithms. Based on the position deviation, the path of the dexterous hand is fine-tuned to ensure that it is accurately positioned at the target position.

[0090] Enable high-precision force feedback to monitor the contact force of the dexterous hand in real time during the approach process. Based on the force feedback data, the dexterous hand's posture and strength are dynamically adjusted to avoid unnecessary pressure on the environment or the hand itself.

[0091] Adjust the finger opening and closing angle and pressure according to the object's shape and surface features to ensure the best grasping effect. Before the actual grasping, make a final posture adjustment to ensure that the dexterous hand is in the best grasping position.

[0092] According to the grasping instructions in the action sequence, high-precision force feedback is used to accurately control the strength of each finger to grasp the object. The force feedback data is used to confirm whether the grasping is successful, ensuring that the object is firmly grasped.

[0093] After the grasping is completed, high-precision force feedback is used to fine-tune the posture of the five-fingered dexterous hand to ensure that it is suitable for the subsequent bomb disposal action. The angle and strength of the dexterous hand are adjusted to prepare for the next bomb disposal action.

[0094] According to the predetermined route, the captured explosives are smoothly transferred to a safe area. In the safe area, harmless treatment measures are implemented, such as dismantling the device and other necessary operations. After the bomb disposal operation is completed, a comprehensive inspection is carried out to ensure that all operations have been completed safely.

[0095] S6. After executing the bomb disposal action, the status of the smart material is monitored in real time by the sensor, and a real-time video stream is provided to the operator, including the following steps:

[0096] After completing the bomb disposal action, the smart material status sensors installed on the five-fingered dexterous hand and the robot are activated to continuously collect the status data of the smart material, including but not limited to stress, strain, and temperature changes. Each collected data is timestamped to ensure the time series integrity and traceability of the data;

[0097] Preprocess the collected data to extract key features, including but not limited to maximum stress and average temperature change rate, simplify the data analysis process, and obtain smart material status data;

[0098] Loading a pre-trained machine learning model Input the pre-processed smart material state data into the machine learning model;

[0099] Use models to analyze data, identify abnormal patterns and potential problems, including but not limited to excessive stress and abnormal temperature changes, and output analysis results, including normal state assessments and anomaly detection reports;

[0100] When an abnormal situation is detected, the alarm mechanism is immediately triggered, notifying the operator through the sound and light alarm, and sending the alarm information to the operator's monitoring terminal to ensure that the operator receives the notification in time. The system automatically executes the preset protective measures, including but not limited to stopping the current operation and returning to a safe position, recording all abnormal events and their handling process, and generating detailed log files;

[0101] Activate the cameras and visual sensors installed on the robot to capture real-time images of the operation site, continuously capture high-definition images of the operation site to ensure coverage of the entire operation area, use efficient video compression technology to reduce bandwidth usage, ensure smooth transmission of video streams, and transmit the compressed video streams to the operator's monitoring terminal in real time, allowing the operator to monitor progress in real time and intervene when necessary.

[0102] S7. After completing the task, return the robot to the safe zone, conduct a comprehensive inspection of the robot, record any abnormalities, and perform maintenance, including the following steps:

[0103] After completing the mission, start the return process in the embedded control panel, prepare to return to the safe zone, start the navigation panel, prepare to plan the route, and load the latest 3D map and environmental data, including but not limited to obstacle locations and terrain characteristics;

[0104] Activate all relevant sensors, obtain the latest sensor information, use the latest environmental data and sensor information to recalculate a safe path back to the safe zone, combine energy-aware adaptive motion planning, optimize the path to minimize energy consumption and ensure safety, and convert the recalculated safe path into specific path planning instructions, including but not limited to speed and steering angle;

[0105] Load the path planning instructions into the robot's motion control panel, confirm that all path instructions are correct, and prepare to perform actual operations. Start the robot and move it step by step along the planned safe path to the designated safe zone. Monitor the execution of each action in real time and make fine adjustments based on actual conditions to ensure the safety and accuracy of the operation. When the robot successfully moves to the designated safe zone, stop moving and confirm that the position is correct.

[0106] Start all self-check functions, including but not limited to hardware diagnosis, software testing, and sensor calibration, run comprehensive inspection procedures, evaluate the status of the Robot, including but not limited to mechanical components, electronic systems, and sensors, and generate a detailed inspection report documenting all inspection results, including normal status and potential problems;

[0107] Export the inspection report to a standard format to ensure it is easy to read and analyze. Submit the inspection report to relevant technicians and management departments to ensure they can understand the status of the robot in a timely manner. Technicians and management departments will jointly analyze the inspection report to identify parts that require maintenance. Based on the inspection results, a specific maintenance plan will be developed, including but not limited to parts replacement, software updates, and sensor calibration.

[0108] Perform all necessary maintenance work according to the maintenance plan to ensure that the robot is in optimal condition. After completing the maintenance, run the self-test program again to confirm that all problems have been resolved;

[0109] Organize all inspection records, exception reports, and maintenance logs into complete mission files, store them in the robot's local storage media to ensure data security, and upload them synchronously to the storage media and cloud servers for remote access and backup.

[0110] This embodiment also provides a computer device, which is suitable for the control method of the five-finger dexterous hand of an intelligent bomb disposal robot, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the control method of the five-finger dexterous hand of an intelligent bomb disposal robot proposed in the above embodiment.

[0111] The computer device may be a terminal, comprising a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0112] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for controlling the five-finger dexterous hand of an intelligent bomb disposal robot as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0113] In summary, the present invention achieves precise control of the movements of each finger by performing precise calibration steps on the five-fingered dexterous hand, quickly and accurately determines the position, shape and characteristics of the explosive by utilizing a highly sensitive sensor network to collect data on the surrounding environment and preprocess the data, uses a machine learning algorithm to plan a safe path from the current position to the explosive based on the environmental and target information obtained previously, and achieves efficient approach to the target, generates a series of precise action instructions by calling the machine learning algorithm to guide the dexterous hand to complete the capture of the explosive, and achieves fine-tuning of the grasping force to avoid danger caused by excessive force.

[0114] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for controlling the five-finger dexterous hand of an intelligent bomb disposal robot, characterized by: include, The five-fingered dexterous hand is calibrated, and sensors are used to collect and pre-process surrounding environmental data. The data is analyzed through image processing and pattern recognition algorithms to obtain surrounding environmental data and the location, shape, and characteristics of the explosives. Based on the surrounding environment data and the location, shape, and characteristics of the explosive, a machine learning algorithm is used to plan a safe path for the five-fingered dexterous hand to reach the explosive. After the planning is complete, the five-fingered dexterous hand is moved near the explosive along the planned safe path. After the five-fingered dexterous hand is moved near the location of the explosive, a machine learning algorithm is used to generate an action sequence. Based on the generated action sequence, high-precision force feedback is used to grasp the explosive. After the grasp is completed, the bomb disposal action is performed; After performing the bomb disposal action, the status of the smart material is monitored in real time through sensors, providing a real-time video stream to the operator. After completing the task, the robot is returned to the safe zone for a comprehensive inspection, recording any abnormalities, and performing maintenance. The data is analyzed by image processing and pattern recognition algorithms to obtain the surrounding environment data and the location, shape and characteristics of the explosives. The following steps are included: Use image processing algorithms to extract location information, shape information, size information, material information, background information, and posture information from the pre-processed surrounding environment data, and apply pattern recognition algorithms to analyze the pre-processed surrounding environment data to determine the location, shape, and characteristics of the surrounding data and the explosives; Generate a 3D model and map of the surrounding environment and target objects based on the environmental data and the location, shape, and characteristics of the explosives, expressed as, Where M is the generated three-dimensional model, V is the integration area, p is any point in the environment, D(p) is the distance function, α is the attenuation coefficient, is the feature extraction function, is the shape matching function, N is the number of known feature points in the environment, i is the index of the feature point, and w i is the importance weight of the i-th feature point, β is the three-dimensional model enhancement coefficient, It is a complex information filtering function; The method includes the following steps: planning a safe path from the five-fingered dexterous hand to the explosive location using a machine learning algorithm based on the surrounding environment data and the location, shape and characteristics of the explosive; and moving the five-fingered dexterous hand to the vicinity of the explosive along the planned safe path after the planning is completed. Based on the generated 3D model and map of the surrounding environment and target object, the trained machine learning model is loaded, the surrounding environment data and explosive feature information are input into the machine learning model, and the current state of the robot is input to obtain the constructed machine learning model; A machine learning model is used to simulate different approach paths. Combined with energy-aware adaptive motion planning, the system takes into account the current battery state and optimizes the path to minimize energy consumption. Furthermore, it considers safety factors and plans a safe path for the five-fingered dexterous hand to reach the explosive through iterative optimization. The five-finger dexterous hand starts its motion execution program, moves step by step along the planned safe path, monitors the execution of each movement in real time, and continuously monitors the applied force through high-precision force feedback; After the five-fingered dexterous hand is moved near the location of the explosive, a machine learning algorithm is used to generate an action sequence, including the following steps: After moving the five-fingered dexterous hand to the designated location according to the planned path, the machine learning model is loaded and the model parameters are verified; The collected environmental data and explosive feature information are input into the machine learning model, and the machine learning model is used to simulate different grasping and manipulation actions to generate action sequences, which are expressed as, Where A is the generated action sequence, T is the total time period, and λ is the time decay coefficient. is the machine learning model output function, is the security assessment function, K is the number of known key points in the environment, k is the index of the key point, γ is the action sequence enhancement coefficient, is the information filtering function, a k is the action parameter of the kth key point; The finalized motion sequence is loaded into the robot's embedded control panel.

2. The method for controlling the five-finger dexterous hand of an intelligent bomb disposal robot according to claim 1, wherein: The five-finger dexterous hand is calibrated, and the surrounding environment data is collected by sensors and pre-processed. The following steps are included: Start the sensor array, communication device and controller, run the hardware and software self-test program, perform physical and electrical calibration of the five-finger dexterous hand, and calibrate the force sensor and other sensors; Activate the sensor installed on the robot. After activation, the sensor starts to collect data about the surrounding environment and obtains the surrounding environment data. The collected surrounding environment data is filtered, denoised and format converted to obtain pre-processed surrounding environment data.

3. The method for controlling the five-finger dexterous hand of an intelligent bomb disposal robot according to claim 1, wherein: The method comprises the following steps: Load the generated motion sequence from the control panel, activate the high-precision force sensor, and use the vision sensor to continuously monitor the position deviation based on the first motion of the motion sequence, fine-tune the path to ensure accurate positioning, and monitor the contact force in real time through high-precision force feedback during the approach process; Use force feedback data to fine-tune the dexterous hand's posture and strength, dynamically adjusting the finger's opening and closing angles and pressure based on the object's shape and surface features. Based on the grasping instructions in the action sequence, high-precision force feedback is used to precisely control the strength of each finger for grasping. After completing the grasping, high-precision force feedback is used to fine-tune the posture of the five-fingered dexterous hand to better perform bomb disposal actions, transfer the explosives to a safe area, implement harmless treatment and dismantle the device.

4. The method for controlling the five-finger dexterous hand of an intelligent bomb disposal robot according to claim 1, wherein: After the bomb disposal action is executed, the status of the smart material is monitored in real time by the sensor, and a real-time video stream is provided to the operator, including the following steps: After executing the bomb disposal action, the sensors installed on the five-fingered dexterous hand and the robot continuously collect data on the state of the smart material to obtain the smart material state data, and the smart material state data is preprocessed to obtain the preprocessed smart material state data; Use machine learning models to analyze pre-processed smart material status data to identify abnormal patterns and potential problems; When an abnormal situation is detected, the alarm mechanism is immediately triggered to notify the operator and take protective measures. All abnormal events and their handling process are recorded; Activate cameras and visual sensors to capture real-time images of the operation site, use compression technology to reduce bandwidth usage, and transmit video streams to the operator's monitoring terminal, allowing the operator to monitor progress in real time and intervene when necessary.

5. The method for controlling the five-finger dexterous hand of an intelligent bomb disposal robot according to claim 1, wherein: After completing the task, the robot is returned to the safe area, a comprehensive inspection is performed on the robot, any abnormalities are recorded, and maintenance is performed, including the following steps: When the mission is completed, the robot's return program is started, the navigation panel is activated, and the collected environmental data and the latest sensor information are used to recalculate a safe path back to the safe zone, generate path planning instructions, and send them to the motion control panel; Control the robot to move to the designated safety zone according to the planned safety path, start the comprehensive inspection program, activate all self-test functions, generate a detailed inspection report, and submit it to the relevant technicians and management department. Based on the inspection results, a specific maintenance plan will be formulated; After completing all maintenance work, all inspection records, abnormality reports and maintenance logs will be archived to establish a complete task file, which will be stored in the storage medium and cloud server.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for controlling the five-finger dexterous hand of an intelligent bomb disposal robot according to any one of claims 1 to 5 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for controlling the five-finger dexterous hand of an intelligent bomb disposal robot according to any one of claims 1 to 5 are implemented.

Images