Device and method for testing grabbing force of dexterous hand

Abstract

The invention discloses a dexterous hand grabbing force testing device, and relates to the technical field of manipulator testing. The profiling assembly comprises a profiling piece used for simulating different working conditions to grab workpieces, a storage ring fixedly connected to the inner side wall of the cabin body and a plurality of notches formed in the inner circle of the storage ring, and the profiling piece comprises a bolt, a profiling head in threaded connection to the top of the bolt and a baffle fixedly arranged at the bottom of the bolt in a sleeving mode; the short column is connected to the bottom end of the bolt through a chain, the magnetic attraction block is slidably connected to the outer portion of the short column, and the baffle is used for clamping and limiting the profiling piece to the top of the notch. A space included angle theta capable of being accurately regulated and controlled is formed between the cable chain and the vertical direction through the mechanical arm, by means of a binocular vision system and a vector dot product algorithm module, included angle data are accurately collected, grabbing working conditions of different angles are flexibly simulated, vertical and oblique grabbing force testing and other multi-dimensional grabbing force testing are achieved, and the mechanical arm better fits the actual application scene of the dexterous hand.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mechanical hand testing, in particular to a dexterous hand grasping force testing device and a testing method thereof. BACKGROUND

[0002] As the core end-of-arm of robots, dexterous hands can simulate human hands to complete precise grasping, handling and other operations, and are widely used in industrial automation production, dangerous environment operation, medical assistance, intelligent manufacturing equipment industry and other fields. The maximum grasping force is a key indicator of core performance, and directly determines the safety, reliability and load adaptation capability of the operation process.

[0003] There are related grasping force testing schemes in the prior art, for example, the Chinese invention patent with publication number CN117841050B "Test device for testing grasping capacity of mechanical hand", which realizes linear loading of the load by injecting water into the test ball, and obtains the limit grasping weight of the mechanical hand by using the weighing table. However, this scheme still has defects: first, the test scene is single, and only vertical grasping force testing can be realized, which cannot simulate the common oblique grasping working condition in actual application, and it is difficult to fully reflect the actual grasping capacity of the dexterous hand; second, the universality and testing efficiency are insufficient, the device relies on "test ball + water circulation system" to realize load loading, which can only adapt to spherical workpieces, and cannot meet the testing needs of irregularly shaped workpieces such as cubes and L-shaped workpieces, and needs to be equipped with complex components such as water tanks, water pumps and water outlet sealing mechanisms, and after testing, it needs to go through the tedious processes of draining, resetting and re-injection, which significantly reduces the testing efficiency; third, the anti-falling protection design is not perfect, and only relies on the buffer column foot of the test ball to reduce the impact of falling, and lacks a special anti-falling mechanism, which not only easily causes the test ball, weighing table and other components to be damaged due to impact, but also may interfere with the normal resetting of the water circulation system due to the rebound of the workpiece, thereby affecting the accuracy of the test data.

[0004] To solve the above problems, a dexterous hand grasping force testing device and a testing method thereof are proposed. SUMMARY

[0005] To solve the above technical problems, a dexterous hand grasping force testing device and a testing method thereof are provided, which solve the problems raised in the background art.

[0006] To achieve the above purposes, the following technical solutions can be used: The present application provides a dexterous hand grasping force testing device, comprising: a sealed test cabin; The profiling assembly comprises a profiling piece for simulating different working conditions to grab workpieces, a placing ring fixedly connected to the inner side wall of the cabin body of the sealed test cabin, and a plurality of notches opened in the inner circle of the placing ring. The profiling piece comprises a bolt, a profiling head threadedly connected to the top of the bolt, a baffle fixedly sleeved at the bottom of the bolt, a short column connected to the bottom end of the bolt through a chain, and a magnetic block slidingly connected to the outside of the short column. The baffle is used to limit the profiling piece to the top of the notch. The test assembly comprises a mechanical arm fixedly installed in the middle of the cabin cover, a vision recognition camera one fixedly installed on both sides of the execution end of the mechanical arm for assembling the dexterous hand to be tested, a vertical installation type tension and compression bidirectional force sensor fixedly installed in the middle of the bottom of the cabin body, and an electromagnet fixedly installed in the sensing area at the top end of the tension and compression bidirectional force sensor. The vision recognition camera one is used for feature recognition of the profiling head, and the positioning alignment between the magnetic block and the electromagnet. The test assembly further comprises two vision recognition camera twos fixedly installed on both sides of the top of the gear ring for collecting the spatial coordinates of both ends of the chain.

[0007] Further, the profiling assembly further comprises a containing cavity opened in the top of the magnetic block. When the baffle is buckled on the top of the magnetic block, the containing cavity is used to contain the naturally drooping chain, so as to realize the buckling stability between the baffle and the magnetic block.

[0008] Further, when the short column slides to the top end relative to the magnetic block, the connection part of the short column top end and the chain will extend out of the top end of the containing cavity, so as to ensure that the chain is not interfered by the edge of the top end of the containing cavity when it is inclined.

[0009] Further, the profiling assembly further comprises a positioning ring fixedly connected to the bottom of the magnetic block, and the test assembly further comprises a limiting groove opened in the top of the electromagnet. The positioning ring is coaxially matched with the limiting groove. When the positioning ring is embedded in the limiting groove, the lateral movement of the magnetic block relative to the electromagnet is limited through the lateral support of the limiting groove to the positioning ring.

[0010] Further, the anti-falling assembly comprises a gear ring rotationally connected to the bottom of the cabin body, a gear meshing with the gear ring, a motor for driving the gear to rotate, a cross frame fixedly connected to both sides of the top of the gear ring, a sliding table electric cylinder fixedly installed in the middle of each cross frame, and an anti-falling plate fixedly connected to the sliding table of each sliding table electric cylinder.

[0011] Further, the environmental simulation assembly comprises a humidifier fixedly installed on one side of the inside of the cabin body, and a humidity sensor fixedly installed on the side wall of the cabin body.

[0012] Further, the environmental simulation assembly further comprises a fan fixedly installed on one side of the inside of the cabin body and located on one side of the humidifier, an electric heating wire fixedly installed at the air inlet of the fan, a temperature sensor fixedly installed on the side wall of the cabin body and located on one side of the humidity sensor, and an exhaust fan fixedly installed on one side of the top of the cabin cover.

[0013] Furthermore, the robotic arm's control system incorporates a spatial coordinate recording module and a vector dot product algorithm module. The spatial coordinate recording module, using the center of the electromagnet as the origin, outputs real-time XYZ three-dimensional coordinate data from the execution end. This data assists the visual recognition camera in grasping and positioning the contoured part, aligning the magnetic block with the electromagnet. It also cross-verifies the coordinates of both ends of the chain collected by the visual recognition camera, calibrating the angle calculation results and eliminating anomalies. Additionally, it enables real-time monitoring and overtravel protection of the dexterous hand's movement trajectory during testing. The vector dot product algorithm module calculates the spatial angle θ between the chain and the vertical direction. The control system also includes a hand-eye calibration unit, a coordinate verification and filtering unit, and a vertical reference self-calibration unit. The specific calculation process includes: Step 1: Establish the conversion relationship between the pixel coordinates of the visual recognition camera 2 and the world coordinates of the robotic arm through the hand-eye calibration unit. The binocular vision system composed of two visual recognition cameras 2 identifies the connection between the bottom of the bolt and the chain, and the connection between the top of the short column and the chain. Calculate the true three-dimensional coordinates of the two by combining the parallax principle with the conversion relationship, and use them as the coordinates of the top and bottom of the chain, respectively. Step 2: The coordinate verification and filtering unit performs mean filtering on the collected coordinates of the top and bottom ends of the chain to remove outliers. Step 3: The vertical reference self-calibration unit calls the true vector when the chain is completely vertical as the vertical reference vector for angle calculation; Step 4: Substitute the calibrated coordinates of the top and bottom ends of the chain into the vector dot product algorithm module to calculate the spatial angle θ between the chain and the vertical direction.

[0014] A testing method for a dexterous hand grasping force testing device includes the following steps: Step 1: The temperature and humidity inside the sealed test chamber are adjusted to the preset test conditions through the environmental simulation component, and the temperature and humidity sensors provide real-time feedback to maintain stable operating conditions. Step 2: The robotic arm moves the dexterous hand to the placement ring. The visual recognition camera identifies the appearance features of the profiling head to determine the target profiling part. Then, the dexterous hand grabs the profiling head and removes the profiling part from the positioning notch. The robotic arm moves the profiling part directly above the electromagnet. The visual recognition camera assists in positioning, so that the positioning ring at the bottom of the magnetic block is embedded in the limiting groove of the electromagnet. The electromagnet is activated and the magnetic block is attracted and fixed on the top of the electromagnet. The gravity value G of the profiling part is measured. Step 3: The dexterous hand grabs the contouring head of the contouring part again, and the robotic arm drives the dexterous hand to lift it vertically upward until the chain is taut. The electric cylinder of the slide table drives the two anti-fall plates to move closer to each other above the magnetic block, with a safe gap between the anti-fall plates to prevent them from touching the chain. Step 4: The robotic arm drives the dexterous hand to continue pulling vertically upwards. A bidirectional force sensor collects vertical force data in real time. When the profiling head detaches from the dexterous hand, the force detected by the sensor instantly drops to zero. The maximum force value Fmax1 before detachment is recorded. After the profiling head falls, it is caught by a fall arrestor to prevent damage to the equipment. This is achieved using vector mechanics formulas. F_max = G + |Fmax1|; The maximum vertical gripping force of the dexterous hand, F_straight_grip_max, was calculated. Step 5: The dexterous hand grabs the contouring head of the contouring part again, and the robotic arm drives the dexterous hand to continue to pull vertically upward until the chain is taut. The motor drives the gear ring to rotate at a preset angle. At this time, the direction of the gap between the fall arrestor plates is the direction of the dexterous hand's lateral movement. Then, the robotic arm drives the dexterous hand to move along the gap direction, keeping the chain taut during the movement. At this time, the chain will form a certain angle θ with the vertical direction. Then, the dexterous hand is kept still, and the visual recognition camera 2 obtains the three-dimensional coordinate data of the top and bottom of the chain. The vector dot product algorithm module built into the robotic arm control system calculates the angle and obtains the spatial angle θ value between the chain and the vertical direction. Step 6: The robotic arm drives the dexterous hand to continue pulling along the extension direction of the chain until the contour head falls off. The tension and compression bidirectional force sensor records the maximum vertical tension value Fmax2. After the contour head falls, it is caught again by the fall arrestor to prevent damage to the equipment. Step 7: Using vector mechanics formulas; Ftension = |Fmax2| / cosθ; To reconstruct the true diagonal tension transmitted by the chain, we then use vector mechanics formulas: Foblique grip max = G*cosθ + Fpull oblique; The maximum gripping force of the dexterous hand in the direction of the included angle θ is calculated, namely F_oblique_grip_max.

[0015] As described above, the features and advantages of the dexterous hand grasping force testing device and its testing method of the present invention are as follows: This solution utilizes a three-dimensional robotic arm to drive a dexterous hand in lateral translation, creating a precisely adjustable spatial angle θ between the chain and the vertical direction. Simultaneously, a binocular vision system and a vector dot product algorithm module accurately collect and calculate the angle data. Combined with optimized mechanical formulas, this allows for precise measurement of the oblique grasping force. This design can flexibly simulate grasping conditions at different angles, enabling multi-dimensional grasping force testing, including vertical and oblique directions. This makes the test results more closely reflect the actual application scenarios of the dexterous hand, avoiding the limitations of existing technologies that rely on water loading onto a test ball to measure only the vertical grasping force. In practical applications, dexterous hands often need to perform oblique grasping operations at different angles, resulting in a significant disconnect between the test scenario and real-world conditions. This effectively overcomes the core deficiency of existing technologies that can only test grasping force in a single vertical direction.

[0016] This solution utilizes interchangeable contoured parts, whose shape, material, and surface friction coefficient can be flexibly adjusted according to testing requirements, adapting to diverse workpiece gripping testing scenarios. Simultaneously, the gravity of each contoured part is individually calibrated, and combined with vertically mounted tension / compression sensors, tensile force data is directly collected, eliminating the need for complex load loading and reset mechanisms. This significantly simplifies the testing process and avoids the limitations of existing technologies that rely on a "test ball + water circulation system" for load loading. On one hand, the test objects are limited to spherical workpieces; although an arc-shaped replacement plate can be used, it still cannot adapt to irregularly shaped workpieces such as cubes and L-shaped parts, resulting in poor versatility. On the other hand, it requires complex water circulation components such as water tanks, water pumps, water outlet plugs, and compression springs, and after testing, it involves cumbersome processes such as draining, resetting, and refilling, leading to low testing efficiency. This solution improves the versatility of the testing device and significantly shortens the testing cycle, effectively addressing the shortcomings of existing technologies in terms of insufficient versatility and cumbersome processes. Furthermore, it simulates gripping conditions through environmental simulation components, making the testing more accurate.

[0017] The fall arrestor in this solution utilizes a combination design of a drive motor, gear transmission, and electric cylinder for the sliding table. This design allows for precise control of the fall arrestor's opening, closing, and rotation. During vertical testing, the fall arrestor quickly converges to form a support surface, stably supporting the falling contoured part and preventing collision damage between the equipment and the workpiece. During oblique testing, the gear rotation drives the fall arrestor to avoid lateral movement, without interfering with the chain tilting or the dexterous lifting action. Simultaneously, a safety gap is reserved between the fall arrestor and the magnetic block to ensure no contact with the chain, thus not affecting the transmission of data or the accuracy of test results. This design achieves reliable fall protection while avoiding interference with the testing process. It effectively overcomes the shortcomings of existing fall protection technologies, such as insufficient protection, susceptibility to equipment damage, and testing errors. It also avoids the damage to test balls, buckets, or weighing platforms caused by falling impacts, which can affect equipment lifespan. Furthermore, it is organically integrated with a binocular vision system, enabling synchronous rotation and preventing interference from the chain itself with the camera's line of sight when acquiring coordinates at both ends of the chain. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the overall structure of the dexterous hand grasping force testing device and its testing method according to the present invention. Figure 2 This is a schematic diagram of the internal structure of the cabin of the dexterous hand grasping force testing device and testing method of the present invention. Figure 3 This is a schematic diagram of the contour component structure of the dexterous hand grasping force testing device and testing method of the present invention. Figure 4 This is an exploded view of the contoured component structure of the dexterous hand grasping force testing device and testing method of the present invention. Figure 5 This is a schematic diagram of the mechanical arm and dexterous hand structure assembly of the dexterous hand grasping force testing device and testing method of the present invention. Figure 6 This is a schematic diagram of the anti-fall component assembly of the dexterous hand grasping force testing device and testing method shown in this invention. Figure 7 for Figure 6 Schematic diagram of the middle section; Figure 8 This is a simplified schematic diagram of the state of the chain when it is tilted in an embodiment of the present invention.

[0019] The reference numerals in the accompanying drawings of this invention are as follows: 11. Sealed test chamber; Contouring components: 21. Storage ring; 22. Notch; 23. Bolt; 24. Contouring head; 25. Baffle; 26. Short post; 27. Magnetic block; 28. Receiving cavity; 29. ​​Positioning ring; Test components: 31. Robotic arm; 32. Visual recognition camera one; 33. Bidirectional force sensor (tension and compression); 34. Electromagnet; 35. Limiting groove; 36. Visual recognition camera two; Fall protection components: 41. Gear ring; 42. Gear; 43. Motor; 44. Crossbeam; 45. Slide table electric cylinder; 46. Fall protection plate; Environmental simulation components: 51. Humidifier; 52. Humidity sensor; 53. Fan; 54. Temperature sensor; 55. Exhaust fan. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0021] See Figures 1-8 As shown in the figure, an embodiment of the present invention is provided, and a dexterous hand grasping force testing device and its testing method will be described in detail below: A dexterous hand grasping force testing device, comprising: See Figure 1 As shown, the sealed test chamber 11, in this embodiment, includes a chamber body and a cover rotatably connected to the top of the chamber body. The two are connected by a hydraulic cylinder or a telescopic electric cylinder to achieve a labor-saving opening and closing effect. The hydraulic cylinder / telescopic electric cylinder driving the opening and closing of the cover is a conventional labor-saving driving technology in the field. Its core principle is to convert fluid pressure or electrical energy into mechanical thrust to achieve smooth opening and closing of the cover. No additional innovative design is required, and industrial-grade standard components can be directly selected for adaptation, ensuring the sealing reliability and operation convenience of the test chamber.

[0022] See Figures 2-4 As shown, the contouring assembly includes a contouring component for simulating different working conditions in grasping workpieces, a storage ring 21 fixedly connected to the inner wall of the sealed test chamber 11, and multiple notches 22 formed in the inner circle of the storage ring 21. The contouring component includes a bolt 23, a contouring head 24 (made of different materials to simulate different shapes and surface roughness) threaded to the top of the bolt 23, a baffle 25 fixedly fitted to the bottom of the bolt 23, a short post 26 connected to the bottom of the bolt 23 by a chain, and a magnetic block 27 slidably connected to the outside of the short post 26. The magnetic block 27 has a circular hole in its center, the size of which matches that of the short post 26. The bottom of the short post 26 has a skirt larger than its size. Through the design of the circular hole and the skirt, the magnetic block 27 can slide outside the short post 26 without detaching from it.

[0023] The baffle 25 is used to snap and limit the profile part at the top of the notch 22. The threaded connection between the bolt 23 and the profile head 24, and the snap-fit ​​limiting structure of the baffle 25 are conventional detachable connection and limiting designs in the mechanical field. This allows for quick replacement of the profile head 24 (to adapt to different test conditions) and stable storage of the profile part. Its connection strength and limiting accuracy can be guaranteed by conventional machining processes, which will not be elaborated on here.

[0024] See Figures 5-7 As shown, the test assembly includes a robotic arm 31 fixed in the middle of the hatch (in this embodiment, the robotic arm 31 is of the type with precise three-dimensional motion capability), an actuator end of the robotic arm 31 for assembling the dexterous hand to be tested, a vision recognition camera 32 fixed on both sides of the actuator end, a vertically mounted tension-compression bidirectional force sensor 33 fixed in the middle of the bottom of the hatch, and an electromagnet 34 fixed in the sensing area at the top of the tension-compression bidirectional force sensor 33. The vision recognition camera 32 is used for feature recognition of the contour head 24 and for positioning and alignment between the magnetic block 27 and the electromagnet 34. The tension-compression bidirectional force sensor 33 only senses forces along its own vertical axis. 1. The vertically mounted tension and compression bidirectional force sensor 33 is a standard component in the field of industrial measurement. Its "sensing only forces along its own vertical axis" is a conventional characteristic achieved through the sensor's structural design (such as strain gauge mounting direction and elastic body structure). It can effectively filter out interference from horizontal lateral forces, ensuring the accuracy of tension measurement without the need for additional modifications. 2. The feature recognition and positioning alignment function of the visual recognition camera 32 is based on the conventional image acquisition and matching technology of industrial machine vision. By pre-entering the appearance feature template of the profiling head 24, the alignment reference template of the magnetic block 27 and the electromagnet 34, target recognition and assisted positioning can be achieved, which is a well-known technology in this field.

[0025] See Figures 6-7As shown, the fall arrestor assembly includes a gear ring 41 rotatably connected to the bottom of the cabin, a gear 42 meshing with the gear ring 41, and a motor 43 for driving the gear 42 to rotate. In this embodiment, the gear ring 41 is of the internal gear type. The motor 43 is fixedly mounted to the bottom of the cabin. The sealed test chamber 11 also includes crossbeams 44 fixedly connected to the top two sides of the gear ring 41, slide cylinders 45 fixedly mounted in the middle of each crossbeam 44, and fall arrestor plates 46 fixedly connected to the slides of each slide cylinder 45. In this embodiment, the gear ring 41 is... The electromagnet 34 is set as the center, and the crossbeams 44, the slide table electric cylinder 45, and the fall arrestor 46 on both sides of its top are symmetrically arranged about the electromagnet 34. The meshing transmission of the gear ring 41 and the gear 42, the drive of the motor 43, and the linear drive of the slide table electric cylinder 45 are all conventional technical combinations in the field of mechanical transmission. Its core function is to drive the fall arrestor 46 to achieve precise opening, closing, and rotation through the transmission mechanism. The motion accuracy can be ensured by selecting standard transmission components and drive components, without the need for innovative design of the transmission principle.

[0026] See Figures 6-7 As shown, the test assembly also includes two visual recognition cameras 36 fixed on both sides of the top of the toothed ring 41, used to collect the spatial coordinates of both ends of the chain. Specifically, in this embodiment, the visual recognition cameras 36 are installed at the tail ends of the two sliding stage electric cylinders 45. Thus, when the toothed ring 41 rotates, the plane on which the chain is tilted is located within the gap between the fall arrestor plates 46, and the two visual recognition cameras 36 located at the tail ends of the two sliding stage electric cylinders 45 are positioned on both sides of the chain. This avoids the chain itself from obstructing the image during the recognition process. The two visual recognition cameras 36 form a binocular vision system. Binocular vision for acquiring three-dimensional coordinates is a mature existing technology in the field of machine vision. Its core logic is to calculate the three-dimensional spatial position of the target by combining the parallax effect of the two cameras (i.e., the positional deviation of the same target on the imaging plane of the two cameras) with the camera's intrinsic and extrinsic parameters. This can be achieved by directly applying the conventional binocular vision calibration and calculation process.

[0027] For further details, please refer to [link / reference]. Figures 3-4 As shown, the contouring component also includes a receiving cavity 28 opened on the top of the magnetic block 27. When the baffle 25 is fastened to the top of the magnetic block 27, the receiving cavity 28 is used to accommodate the naturally hanging chain, so as to achieve a stable fastening between the baffle 25 and the magnetic block 27. The gravity of the contouring component is one of the parameters in calculating the grasping force. When the gravity is measured, stable placement is very important. By accommodating the chain through the receiving cavity 28, the fastening between the baffle 25 and the magnetic block 27 is more stable after the dexterous hand releases the contouring head 24, and the accumulated chain is prevented from overturning the baffle 25.

[0028] Furthermore, when the short post 26 slides to its highest point relative to the magnetic block 27, the connection between the top of the short post 26 and the chain will extend beyond the top of the receiving cavity 28, ensuring that the chain is not interfered with by the edge of the top of the receiving cavity 28 when tilted. This design is based on conventional structural optimization to avoid motion interference. By presetting the sliding stroke of the short post 26 and the height of the receiving cavity 28, the chain's freedom of movement within the entire test angle range is guaranteed. No complex mechanical simulation verification is required, which is a conventional design consideration for those skilled in the art.

[0029] For further details, please refer to [link / reference]. Figure 4 and Figure 7 As shown, the contouring assembly also includes a positioning ring 29 fixedly connected to the bottom of the magnetic block 27, and the test assembly also includes a limiting groove 35 opened on the top of the electromagnet 34. The positioning ring 29 and the limiting groove 35 are coaxially adapted. When the positioning ring 29 is embedded in the limiting groove 35, the lateral support of the positioning ring 29 by the limiting groove 35 restricts the lateral movement of the magnetic block 27 relative to the electromagnet 34. At the same time, when the chain hangs down, the short column 26 naturally sinks into the positioning ring 29 due to gravity, thus making room for the bottom of the bolt 23 and the chain in the receiving cavity 28.

[0030] See Figure 2 As shown, the system further includes an environmental simulation component, comprising a humidifier 51 fixedly installed on one side of the cabin interior, and a humidity sensor 52 fixedly installed on the side wall of the cabin. The environmental simulation component also includes a fan 53 fixedly installed on one side of the cabin interior and located next to the humidifier 51, a heating wire (conventional technology, not specifically shown in the figure) fixedly installed at the air inlet of the fan 53, a temperature sensor 54 fixedly installed on the side wall of the cabin and located next to the humidity sensor 52, and an exhaust fan 55 fixedly installed on one side of the top of the cabin cover. The humidifier... Humidifier 51, fan 53, heating wire, exhaust fan 55, and temperature and humidity sensor 52 are all industrial-grade conventional temperature and humidity control components. Their working principle and control logic are existing technologies: the temperature and humidity sensor 52 collects the environmental parameters inside the cabin in real time and feeds them back to the external control unit. The control unit adjusts the working status of humidifier 51, heating wire, and fan 53 to achieve precise control of temperature and humidity inside the cabin. The exhaust fan 55 is used to ensure air circulation inside the cabin and avoid uneven local temperature and humidity. The entire control system can be directly implemented based on conventional PID control algorithms without the need for innovative control logic design.

[0031] Furthermore, the control system of the robotic arm 31 incorporates a spatial coordinate recording module and a vector dot product algorithm module. The spatial coordinate recording module, with the center of the electromagnet 34 as the origin, outputs real-time XYZ three-dimensional coordinate data of the execution end. This data assists the visual recognition camera 32 in completing the contour-grabbing and positioning, and the alignment and positioning of the magnetic block 27 with the electromagnet 34. It also cross-verifies the coordinates of both ends of the chain collected by the visual recognition camera 36, ​​calibrating the angle calculation results and eliminating anomalies. Additionally, it enables real-time monitoring and overtravel protection of the dexterous hand's movement trajectory during testing. The vector dot product algorithm module calculates the spatial angle θ between the chain and the vertical direction. The control system also incorporates a hand-eye calibration unit, a coordinate verification and filtering unit, and a vertical reference self-calibration unit. (See reference...) Figure 8 As shown, the specific calculation process includes: Step 1: Establish the conversion relationship between the pixel coordinates of the visual recognition camera 2 36 and the world coordinates of the robotic arm 31 through the hand-eye calibration unit. The binocular vision system composed of two visual recognition cameras 2 36 identifies the connection between the bottom of the bolt 23 and the chain, and the connection between the top of the short column 26 and the chain. Calculate the true three-dimensional coordinates of the two by combining the parallax principle with the conversion relationship, and use them as the coordinates of the top and bottom of the chain, respectively. Among them, the hand-eye calibration technology corresponding to the hand-eye calibration unit is a well-known existing technology in the fields of machine vision and robotics. Its core purpose is to establish a rigid transformation relationship between the camera pixel coordinate system and the robot world coordinate system (i.e., to obtain the transformation matrix). This embodiment can directly adopt the mature Zhang's calibration method in this field. By taking pictures of the preset calibration board image, the camera intrinsic parameters (such as focal length and principal point coordinates) and extrinsic parameters (such as the relative posture of the camera and the robotic arm 31) are calculated, thereby completing the coordinate system unification without the need for innovative calibration algorithms. The parallax principle of binocular vision is also a mature existing technology. By combining the imaging differences of two cameras on the same target with the calibrated camera parameters, the three-dimensional spatial coordinates of the target can be calculated. This is a conventional means of obtaining three-dimensional position information in the industrial field.

[0032] Step 2: The coordinate verification and filtering unit performs mean filtering on the collected coordinates of the top and bottom ends of the chain to remove outliers. Among them, the coordinate verification (outlier identification) and mean filtering techniques corresponding to the coordinate verification and filtering unit are both common existing data processing technologies: outlier identification can be achieved by judging whether the coordinate data exceeds the preset reasonable range (such as based on the σ criterion), while mean filtering reduces random errors (such as noise from visual acquisition or deviations caused by mechanical vibration) by averaging the coordinates of the same target collected multiple times. Both are data preprocessing methods that can be implemented by those skilled in the art without creative effort and are widely used in various measurement systems.

[0033] Step 3: The vertical reference self-calibration unit calls the true vector when the chain is completely vertical as the vertical reference vector for angle calculation; Among them, the self-calibration logic corresponding to the vertical reference self-calibration unit belongs to the conventional reference calibration technology. Its core idea is "collecting known standard states → correcting system references". It is widely used in angle measurement and attitude calibration equipment. In this embodiment, the state when the cable is completely vertical is a standard state that can be accurately reproduced. By pre-collecting the cable vector in this state as a reference, the installation deviation between the 31 coordinate system of the robotic arm and the actual vertical direction can be corrected, ensuring the accuracy of the reference for angle calculation. This logic does not require innovation and can be directly applied to the conventional calibration process.

[0034] Step 4: Substitute the calibrated coordinates of the top and bottom ends of the chain into the vector dot product algorithm module to calculate the spatial angle θ between the chain and the vertical direction.

[0035] Among them, the vector dot product algorithm is a standard existing technology in the fields of mathematics and engineering for calculating the angle between two vectors. Its core formula is: cosθ=(vector a・vector b) / (|vector a|×|vector b|). In this embodiment, vector a is the cable vector formed by the coordinates of the two ends of the cable, and vector b is the vertical reference vector determined in step three. The included angle θ can be directly calculated through this formula. The algorithm itself is not novel. The value of this solution lies in integrating this conventional algorithm with technologies such as visual acquisition and coordinate calibration to adapt to the specific scenario of dexterous hand grasping force testing.

[0036] A testing method for a dexterous hand grasping force testing device includes the following steps: Step 1: The temperature and humidity inside the sealed test chamber 11 are adjusted to the preset test conditions through the environmental simulation component, and the temperature sensor 54 and humidity sensor 52 provide real-time feedback to maintain stable conditions. Step 2: The robotic arm 31 moves the dexterous hand to be tested to the placement ring 21. The visual recognition camera 32 identifies the appearance features of the contouring head 24 to determine the target contouring part. Then, the dexterous hand grabs the contouring head 24 and takes the contouring part out from the positioning notch 22. The robotic arm 31 moves the contouring part to directly above the electromagnet 34. The visual recognition camera 32 assists in positioning so that the positioning ring 29 at the bottom of the magnetic block 27 is embedded in the limiting groove 35 of the electromagnet 34. The electromagnet 34 is activated and the magnetic block 27 is attracted and fixed to the top of the electromagnet 34. The gravity value G of the contouring part is measured. Step 3: The dexterous hand grabs the contouring head 24 of the contouring part again, and the robotic arm 31 drives the dexterous hand to lift vertically upward until the chain is taut vertically. The slide table electric cylinder 45 drives the two anti-fall plates 46 to move closer to each other above the magnetic block 27, with a safe gap reserved between the anti-fall plates 46 so as not to touch the chain. Step 4: The robotic arm 31 continues to pull the dexterous hand vertically upwards. The bidirectional force sensor 33 collects the vertical force data in real time. When the profiling head 24 detaches from the dexterous hand, the force detected by the sensor instantly drops to zero. The maximum force value Fmax before detachment is recorded. After the profiling head 24 falls, it is caught by the fall arrestor 46 to prevent damage to the equipment. This is achieved using vector mechanics formulas. F_max = G + |Fmax|; The maximum vertical gripping force of the dexterous hand, F_straight_grip_max, was calculated. Step 5: The dexterous hand grabs the contouring head 24 of the contouring part again. The robotic arm 31 drives the dexterous hand to continue to pull vertically upward until the chain is taut. The motor 43 drives the gear ring 41 to rotate at a preset angle. At this time, the direction of the gap between the fall arrestor plates 46 is the direction of the dexterous hand's lateral movement. Then, the robotic arm 31 drives the dexterous hand to move along the gap direction, keeping the chain taut during the movement. At this time, the chain will form a certain angle θ with the vertical direction. Then, the dexterous hand is kept still, and the visual recognition camera 36 obtains the three-dimensional coordinate data of the top and bottom of the chain. The vector dot product algorithm module built into the control system of the robotic arm 31 calculates the angle and obtains the spatial angle θ value between the chain and the vertical direction. Step 6: The robotic arm 31 drives the dexterous hand to continue pulling along the extension direction of the chain until the contour head 24 falls off. The tension and compression bidirectional force sensor 33 records the maximum vertical tension value Fmax. After the contour head 24 falls, it is caught again by the fall arrestor plate 46 to prevent damage to the equipment. Step 7: Using vector mechanics formulas; Fpull-up = |Fmax| / cosθ; To reconstruct the true diagonal tension transmitted by the chain, we then use vector mechanics formulas: Foblique grip max = G*cosθ + Fpull oblique; The maximum gripping force of the dexterous hand in the direction of the included angle θ is calculated, namely F_oblique_grip_max.

[0037] It should be noted that the "force superposition and decomposition calculation" (formulas in steps four and seven) involved in the above testing methods are all based on the principles of classical vector mechanics, especially the application of action and reaction forces. For example, the tensile force detected by the sensor is actually the force exerted by the contouring part on the sensor, with its vector direction being from the sensor to the contouring part. However, when actually calculating the grasping force, what is needed is the tensile force exerted by the sensor on the contouring part, with its vector direction being from the contouring part to the sensor. The two are equal in absolute value, which is the manifestation of action and reaction forces. All of the above belong to the well-known mechanical calculation methods in this field, and their accuracy has been widely verified. This solution only applies it to the calculation scenario of dexterous hand grasping force and does not require the innovation of mechanical principles.

[0038] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0039] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A dexterous hand grasping force testing device, characterized by, include: Sealed test chamber (11); The contouring assembly includes a contouring component for simulating different working conditions to grasp the workpiece, a storage ring (21) fixedly connected to the inner wall of the sealed test chamber (11), and multiple notches (22) opened in the inner circle of the storage ring (21). The contouring component includes a bolt (23), a contouring head (24) threaded to the top of the bolt (23), a baffle (25) fixedly fitted to the bottom of the bolt (23), a short post (26) connected to the bottom of the bolt (23) by a chain, and a magnetic block (27) slidably connected to the outside of the short post (26). The baffle (25) is used to snap the contouring component into place at the top of the notch (22). The test assembly includes a robotic arm (31) fixed in the middle of the hatch, an actuator end of the robotic arm (31) for assembling the dexterous hand to be tested, a visual recognition camera (32) fixed on both sides of the actuator end, a vertically mounted tension-compression bidirectional force sensor (33) fixed in the middle of the bottom of the hatch, and an electromagnet (34) fixed in the sensing area at the top of the tension-compression bidirectional force sensor (33). The test assembly also includes two visual recognition cameras (36) fixed on both sides of the top of the gear ring (41).

2. The dexterous hand grasping force testing device of claim 1, wherein: The contouring component also includes a receiving cavity (28) opened on the top of the magnetic block (27). When the baffle (25) is fastened to the top of the magnetic block (27), the receiving cavity (28) is used to accommodate the naturally hanging chain, so as to achieve a stable fastening between the baffle (25) and the magnetic block (27).

3. The dexterous hand grasping force testing device of claim 2, wherein: When the short post (26) slides to the top relative to the magnetic block (27), the connection between the top of the short post (26) and the chain will extend out of the top of the receiving cavity (28), ensuring that the chain is not interfered with by the edge of the top of the receiving cavity (28) when it is tilted.

4. The dexterous hand grasp force testing device of claim 3, wherein: The contouring assembly also includes a positioning ring (29) fixedly connected to the bottom of the magnetic block (27), and the test assembly also includes a limiting groove (35) opened on the top of the electromagnet (34). The positioning ring (29) and the limiting groove (35) are coaxially adapted. When the positioning ring (29) is embedded in the limiting groove (35), the lateral movement of the magnetic block (27) relative to the electromagnet (34) is restricted by the lateral support of the positioning ring (29) by the limiting groove (35).

5. The dexterous hand grasping force testing device of claim 1, wherein: It also includes a fall arrestor assembly, which includes a gear ring (41) rotatably connected to the bottom of the cabin, a gear (42) meshing with the gear ring (41), a motor (43) for driving the gear (42) to rotate, a crossbeam (44) fixedly connected to both sides of the top of the gear ring (41), a slide electric cylinder (45) fixedly mounted in the middle of each crossbeam (44), and a fall arrestor plate (46) fixedly connected to the slide of each slide electric cylinder (45).

6. The dexterous hand grasp force testing device of claim 1, wherein: It also includes an environmental simulation component, which includes a humidifier (51) fixedly installed on one side of the cabin interior, and a humidity sensor (52) fixedly installed on the side wall of the cabin.

7. The dexterous hand grasp force testing device of claim 6, wherein: The environmental simulation assembly further comprises a fan (53) fixedly installed on one side of the inside of the cabin body and located on one side of the humidifier (51), an electric heating wire fixedly installed at an air inlet of the fan (53), a temperature sensor (54) fixedly installed on a side wall of the cabin body and located on one side of the humidity sensor (52), and an exhaust fan (55) fixedly installed on one side of the top of the cabin cover.

8. The dexterous hand grasp force testing device of claim 1, wherein: The control system of the mechanical arm (31) is built-in with a space coordinate recording module and a vector dot product algorithm module, the space coordinate recording module takes the center of the electromagnet (34) as the space coordinate origin, and outputs the X-Y-Z three-dimensional coordinate data of the execution end in real time, which is used to assist the visual recognition camera one (32) to complete the profiling piece grabbing positioning and the alignment positioning of the magnetic block (27) and the electromagnet (34), and is used to cross-verify the two end coordinates of the cable chain collected by the visual recognition camera two (36), realize the calibration and abnormal elimination of the included angle calculation result, and realize the real-time monitoring and overtravel protection of the motion trajectory of the dexterous hand in the test process, the vector dot product algorithm module is used to calculate the space included angle θ of the cable chain and the vertical direction, and the control system is further built-in with a hand-eye calibration unit, a coordinate verification filtering unit and a vertical reference self-calibration unit, and the specific calculation process comprises: Step one, the conversion relationship between the pixel coordinates of the visual recognition camera two (36) and the world coordinates of the mechanical arm (31) is established through the hand-eye calibration unit, the binocular vision system composed of two visual recognition cameras two (36) identifies the connection between the bottom end of the bolt (23) and the cable chain and the connection between the top end of the short column (26) and the cable chain, and calculates the real three-dimensional coordinates of the two through the parallax principle combined with the conversion relationship, which are respectively taken as the top end coordinate and the bottom end coordinate of the cable chain; Step two, the coordinate verification filtering unit performs mean filtering on the collected top end and bottom end coordinates of the cable chain to eliminate abnormal values; Step three, the vertical reference self-calibration unit calls the real vector when the cable chain is completely vertical as the vertical reference vector for included angle calculation; Step four, the calibrated top end and bottom end coordinates of the cable chain are substituted into the vector dot product algorithm module to calculate the space included angle θ of the cable chain and the vertical direction.

9. A testing method of a dexterous hand grasping force testing device, characterized by: The following steps are included: Step one, the temperature and humidity in the sealed test cabin (11) are regulated to the preset test working condition through the environmental simulation assembly, and the working condition is maintained stable through real-time feedback of the temperature sensor and the humidity sensor; Step two, the mechanical arm (31) drives the dexterous hand to be tested to move to the object placing ring (21), the visual recognition camera one (32) identifies the appearance features of the profiling head (24) to determine the target profiling piece, then the dexterous hand grabs the profiling head (24) and takes out the profiling piece from the positioning gap (22), the mechanical arm (31) drives the profiling piece to move directly above the electromagnet (34), the visual recognition camera one (32) assists in positioning, so that the positioning ring (29) at the bottom of the magnetic block (27) is embedded into the limiting groove (35) of the electromagnet (34), the electromagnet (34) is started and the magnetic block (27) is attracted and fixed on the top of the electromagnet (34), and the profiling piece gravity value G is measured; Step three, the dexterous hand again grabs the profiling head (24) of the profiling part, the mechanical arm (31) drives the dexterous hand to pull vertically upward until the cable chain is vertically taut, the slide cylinder (45) drives the two anti-falling plates (46) to approach each other above the magnetic suction block (27), and a safety gap is reserved between the anti-falling plates (46) without touching the cable chain; Step four, the mechanical arm (31) drives the dexterous hand to continue pulling vertically upward, and the tensile and compressive force sensor (33) collects real-time pulling force data in the vertical direction. When the profiling head (24) falls off from the dexterous hand, the pulling force detected by the sensor is zero instantaneously, and the maximum pulling force value Fmax1 before falling off is recorded. After the profiling head (24) falls, it is received by the anti-falling plate (46) to avoid damaging the equipment. Through vector mechanics formula: Fstraight grab max=G+|Fmax1|; The maximum vertical grabbing force of the dexterous hand is calculated, that is, Fstraight grab max; Step five, the dexterous hand again grabs the profiling head (24) of the profiling part, the mechanical arm (31) drives the dexterous hand to continue pulling vertically upward until the cable chain is vertically taut, the motor (43) drives the gear ring (41) to rotate by a preset angle. At this time, the gap direction between the anti-falling plates (46) is the lateral movement direction of the dexterous hand. Then the mechanical arm (31) drives the dexterous hand to move along the gap direction, keeping the cable chain taut during movement. At this time, the cable chain will form a certain angle θ with the vertical direction. Then the dexterous hand is kept still, and the three-dimensional coordinate data of the top and bottom of the cable chain are obtained by the visual recognition camera two (36). The vector dot product algorithm module built-in the control system of the mechanical arm (31) is used to calculate the angle, and the spatial angle θ value of the cable chain and the vertical direction is obtained; Step six, the mechanical arm (31) drives the dexterous hand to continue pulling in the direction of the cable chain extension until the profiling head (24) falls off, and the tensile and compressive force sensor (33) records the maximum vertical pulling force value Fmax2. After the profiling head (24) falls, it is received by the anti-falling plate (46) again to avoid damaging the equipment; Step seven, through vector mechanics formula: Foblique pull=|Fmax2| / cosθ; The real oblique pulling force transmitted by the cable chain is restored, and through vector mechanics formula: Foblique grab max=G*cosθ+Foblique pull; The maximum grabbing force of the dexterous hand in the angle θ direction is calculated, that is, Foblique grab max.

Images