Self-adaptive flexible pneumatic gripper based on intelligent sensing and mechanical arm

Abstract

The invention discloses a self-adaptive flexible pneumatic gripper based on intelligent sensing and a mechanical arm. The gripper comprises a plurality of bases, a plurality of flexible pneumatic fingers and a plurality of flexible pressure sensors. The multiple flexible pneumatic fingers are correspondingly installed on the multiple bases. A main driving cavity and an auxiliary adjusting cavity which are independent of each other are formed in each finger. The main driving cavity is inflated to drive the fingers to bend towards one side, and the auxiliary adjusting cavity is inflated to drive the fingers to bend towards the other side. The multiple flexible pressure sensors are correspondingly arranged on the inner side contact surfaces of the multiple fingers respectively and are made of piezoresistive materials, and flexible conductive silica gel channels are formed in the flexible pressure sensors. The gripper determines the characteristics of an object according to an electric signal generated when the pressure sensor is touched and pressed and generates a gripping strategy, and self-adaptive gripping is achieved by adjusting the air pressure of each cavity. Bidirectional bending and rigidity adjustment can be achieved, efficient, accurate and flexible grabbing of diversified objects in the unstructured environment is achieved, and the mechanical arm has wide application prospects and market value.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of robots, in particular to a self-adaptive flexible pneumatic gripper based on intelligent sensing, and to a robot arm based on intelligent sensing. BACKGROUND

[0002] With the expansion of robot technology to complex scenarios, traditional rigid grippers have obvious limitations in handling fragile objects, irregular objects and complex environments. In recent years, bionic flexible grippers have become a research hotspot due to their material compliance and structural adaptability.

[0003] The existing flexible gripper technology mainly faces the following problems: first, insufficient degrees of freedom. Most flexible grippers can only bend in one direction, making it difficult to adapt to the diverse grasping needs in unstructured environments, such as the ring-shaped grasping of objects with protrusions on the side. Second, lack of intelligent sensing. Existing solutions rely on pre-set programs and cannot sense contact force and object state changes in real time, which can easily lead to excessive grasping force damaging the object or insufficient grasping force causing the object to slip. Third, low integration level. Traditional designs separate sensors, controllers and actuators, resulting in complex structures, large volumes and difficulty in meeting lightweight and small size requirements. SUMMARY

[0004] To solve the technical problems of insufficient degrees of freedom and lack of intelligent sensing in existing flexible grippers, the present application provides a self-adaptive flexible pneumatic gripper based on intelligent sensing and a robot arm.

[0005] The present application adopts the following technical solutions: a self-adaptive flexible pneumatic gripper based on intelligent sensing, comprising: a plurality of bases; a plurality of flexible pneumatic fingers corresponding to the plurality of bases; each flexible pneumatic finger has a main driving cavity and an auxiliary adjusting cavity inside, which are independent of each other, and each cavity is connected to an external air pressure control system through an independent air path; the main driving cavity drives the corresponding finger to bend to one side when inflated, and the auxiliary adjusting cavity drives the corresponding finger to bend to the other side when inflated; a plurality of flexible pressure sensors corresponding to the inner contact surfaces of the plurality of fingers, made of piezoresistive material, and having a flexible conductive silicone channel inside; wherein the gripper determines the object characteristics and generates a grasping strategy based on the electrical signals generated by the pressure sensors when they are pressed, and adjusts the air pressure of each cavity to achieve adaptive grasping.

[0006] The application realizes bidirectional bending and stiffness adjustment by independently arranging a main driving cavity and an auxiliary adjusting cavity in the flexible pneumatic finger, connecting the air pressure control system through an independent air path, bending the finger to the inner side when the main driving cavity is inflated, adjusting the bending degree according to the inflation amount, realizing the embracing action on the object, slightly twisting the finger to the outer side when the auxiliary adjusting cavity is inflated alone, and adapting to the asymmetric shape of the object; when the two cavities are inflated cooperatively, the knuckles can complete the bending action, for example, when grabbing the object with a protrusion on the side, the double air chambers are coordinated to adjust the contact angle, and stable grabbing is ensured. The flexible conductive silica gel is injected into the flexible conductive silica gel channel to transmit the sensor electric signal, and after pre-treatment and adhesion, pressure distribution sensing, surface contour identification and sliding object detection can be realized, solving the technical problems of insufficient freedom and lack of intelligent sensing of the existing flexible gripper.

[0007] As a further improvement of the above-mentioned scheme, the gripper further comprises: a plurality of rods corresponding to the plurality of bases, one end of each rod being rotatably connected to the corresponding base; a plurality of members corresponding to the plurality of rods, the other end of each rod being rotatably connected to one end of the corresponding member; a fixed disc connected to the other end of the plurality of members.

[0008] Further, the gripper further comprises: a stepper motor; a lead screw connected to the output shaft of the stepper motor; a lead screw sleeve screwed with the lead screw; a plurality of pull rods corresponding to the plurality of rods, one end of each pull rod being rotatably connected to the lead screw sleeve, and the other end being rotatably connected to the corresponding rod; wherein, when the stepper motor rotates, the lead screw and the lead screw sleeve move relatively, so that each pull rod pulls each finger to realize the grabbing action.

[0009] As a further improvement of the above-mentioned scheme, each finger comprises two air chamber outer walls with a concave-convex cross section and a partition plate between the two air chamber outer walls, and the two air chamber outer walls and the partition plate form two cavities.

[0010] As a further improvement of the above-mentioned scheme, the piezoresistive material is a conductive composite material made of carbon nanotubes and flexible polymers, and the finger is integrally formed by 3D printing using a silica gel-based material.

[0011] As a further improvement of the above-mentioned scheme, the gripper further comprises: An intelligent control system is used for processing a visual image taken based on a deep learning algorithm, identifying the category and state of a target object, generating an optimal grasping scheme according to the identification result and a preset knowledge base, and finally adjusting the air pressure value of each chamber in real time according to the pressure information detected by the pressure sensor, so that the grasping force of each finger is kept within a preset safe range.

[0012] As a further improvement of the above scheme, each finger joint adopts a non-equal cross-section design, and the root width is greater than the tip width, and a mechanical limiting structure is arranged at the root of the finger joint to limit the maximum bending angle.

[0013] As a further improvement of the above scheme, the flexible conductive silica gel channel is distributed in the pressure sensor in the shape of a heart, and the remaining part is flexible non-conductive silica gel.

[0014] The application also provides an intelligent perception-based mechanical arm, which comprises at least one intelligent perception-based adaptive flexible pneumatic gripper.

[0015] As a further improvement of the above scheme, the mechanical arm further comprises: a bottom plate; a plurality of fixed suction cups mounted at the bottom of the bottom plate; a visual recognition module mounted on the bottom plate and used for taking a visual image of a target object and identifying the category and state of the object; a fixed support mounted on the bottom plate; a plurality of rudders mounted on the fixed support and used for realizing the rotation and extension of the whole gripper.

[0016] Compared with the existing flexible gripper, the intelligent perception-based adaptive flexible pneumatic gripper and the mechanical arm have the following beneficial effects: 1. The intelligent perception-based adaptive flexible pneumatic gripper, by arranging independent main driving cavities and auxiliary adjusting cavities in the flexible pneumatic fingers, connecting the air pressure control system through independent air paths, can realize bidirectional bending and stiffness adjustment, the fingers bend inward when the main driving cavities are inflated, the bending degree can be adjusted according to the inflation amount, the ring action on the object is realized, when the auxiliary adjusting cavities are inflated alone, the fingers will slightly twist outward, which can adapt to the asymmetric shape of the object; when the two cavities are inflated cooperatively, the finger joints can complete the bending action, for example, when the object with a protrusion on the side is grasped, the contact angle is adjusted by the coordinated bending of the double air chambers, so as to ensure stable grasping. The flexible conductive silica gel is injected into the flexible conductive silica gel channel to transmit the sensor electrical signal, and after pre-treatment and fitting, the pressure distribution perception, surface contour identification and sliding object detection can be realized, solving the technical problems of insufficient degrees of freedom and lack of intelligent perception of the existing flexible gripper.

[0017] 2. The adaptive flexible pneumatic gripper based on intelligent sensing has excellent environmental adaptability and precise force control capability, can grasp various material objects such as fragile objects and hard objects, has low system response time, and lower power consumption than traditional solutions, realizes efficient, precise and flexible grasping of diversified objects in an unstructured environment, and has broad application prospects and market value.

[0018] 3. The adaptive flexible pneumatic gripper based on intelligent sensing can adapt to grasping of objects in complex shapes and positions in an unstructured environment through multi-degree-of-freedom motion and intelligent sensing, does not need accurate pre-positioning, has excellent environmental adaptability. The flexible tactile sensing system of the gripper cooperates with a closed-loop control algorithm to realize precise control of grasping force, can grasp various material objects from eggs (fragile objects) to metal parts (hard objects), has wide applicability. The gripper also has fast response capability and low power consumption design, realizes efficient, precise and flexible grasping of diversified objects in an unstructured environment through innovative structural design and intelligent control system, and has broad application prospects and market value.

[0019] 4. The mechanical arm based on intelligent sensing has the same beneficial effects as the adaptive flexible pneumatic gripper based on intelligent sensing described above, and will not be described here. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 FIG. 1 is a structural schematic diagram of the adaptive flexible pneumatic gripper based on intelligent sensing of embodiment 1 of the present application.

[0021] Figure 2 FIG. 2 is a structural schematic diagram of a finger of the adaptive flexible pneumatic gripper based on intelligent sensing in FIG. 1. Figure 1

[0022] Figure 3 FIG. 3 is a structural schematic diagram of a pressure sensor of the adaptive flexible pneumatic gripper based on intelligent sensing in FIG. 1. Figure 1

[0023] Figure 4 FIG. 4 is a structural schematic diagram of the mechanical arm based on intelligent sensing of embodiment 2 of the present application.

[0024] SYMBOL EXPLANATION 1 Pressure sensor 15 Inner cavity 2 Finger 16 Touch surface 3 Base 17 Mounting position 4 Rod 18 Signal input and output end 5 Member 19 Flexible conductive silicone channel 6 Fixed disc 20 Flexible non-conductive silicone 7 Motor housing 21 Screw sleeve 8 Fixed housing 22 Servo 9 Stepping motor 23 Fixed support 10 Pull rod 24 Main control board 11 Connection part 25 Bottom plate 12 Screw 26 Fixed suction cup 13 Air inlet 27 Visual identification module 14 Air chamber outer wall DETAILED DESCRIPTION

[0025] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0026] Embodiment 1​​ Please refer to Figure 1 , Figure 2 and Figure 3 , the embodiment provides a kind of based on intelligent perception's adaptive flexible pneumatic gripper, which can be widely applied in automation production, logistics sorting, medical auxiliary, rescue and so on Unstructured environment under the precise flexible grabbing scene, especially suitable for the precise flexible grabbing of different shape, material object under unstructured environment. Among them, the gripper includes multiple bases 3, multiple flexible pneumatic fingers 2 and multiple flexible pressure sensors 1, can also include multiple rods 4, multiple components 5, fixed disc 6, stepper motor 9, lead screw 12, lead screw sleeve 21, multiple pull rods 10, fixed shell 8, motor housing 7 and intelligent control system.

[0027] It needs to be explained here that the gripper of the embodiment is mainly composed of four modules of multi-degree-of-freedom flexible actuator module, flexible tactile sensing system module, intelligent control module and lightweight integrated module, the structure included in the multi-degree-of-freedom flexible actuator module mainly realizes the grabbing function of the gripper, such as flexible pneumatic finger 2. Flexible tactile sensing system module is a sensing module, including flexible pressure sensor 1, intelligent control module is intelligent control system, lightweight integrated module adopts flexible printed circuit board technology to integrate flexible tactile sensing system module, intelligent control module and driving circuit on flexible substrate material, and each functional module is connected through standardized interface.

[0028] Multiple flexible pneumatic fingers 2 are installed on multiple bases 3. In the embodiment, the number of flexible pneumatic fingers 2 and base 3 is preferably 4. The inside of each flexible pneumatic finger 2 is provided with a mutually independent main drive cavity and an auxiliary adjusting cavity, and each cavity is connected to the external air pressure control system through independent air path. Finger 2 can realize bidirectional bending and stiffness adjustment, and the root is provided with mechanical limiting structure to limit the maximum bending angle. The main drive cavity drives the corresponding finger 2 to bend to one side when inflated, and the auxiliary adjusting cavity drives the corresponding finger 2 to bend to the other side when inflated, so as to realize bidirectional bending. When the main drive cavity is inflated, the flexible knuckle bends inward, and the stiffness of the knuckle can be changed by adjusting the cavity pressure value.

[0029] When the main driving cavity is connected to the air pressure control system, the knuckles bend inward during inflation, and the bending degree can be adjusted according to the inflation amount to achieve the surrounding action on the object. When the auxiliary adjusting cavity is inflated alone, the knuckles will produce slight bending, which can adapt to the asymmetric shape of the object. When the two cavities are inflated together, the knuckles can complete the composite action of bending, for example, when grabbing an object with a protrusion on the side, the contact angle is adjusted by the coordinated bending of the double air chambers to ensure stable grabbing. To avoid excessive deformation, a mechanical limiting structure is provided at the root of the knuckles to limit the maximum bending angle and protect the internal air cavity and sensing elements. In addition, the air pressure control system is connected to a miniature electromagnetic valve to adjust the inflation amount of each air cavity; each module is connected through a flexible circuit board and is packaged in the base 3 to reduce the occupied space.

[0030] In this embodiment, each finger 2 includes two air chamber outer walls 14 with a concave-convex cross-section and a partition plate between the two air chamber outer walls 14, and the two air chamber outer walls 14 and the partition plate form two cavities. The fingers 2 are integrally formed by 3D printing using a silicone-based material, each finger 2 knuckle adopts a non-equal cross-section design, the root width is greater than the tip width, and a mechanical limiting structure is provided at the knuckle root to limit the maximum bending angle. This limiting structure is tightly bonded with the flexible actuator, and the bonding part has high sealing performance, which plays an auxiliary air supply role. During printing, a gradual thickness design is adopted at the knuckle tip and bending part to ensure structural strength and improve bending flexibility. After molding, each part is sealed and assembled by special silicone glue to ensure that the air cavity does not leak. The air path interface adopts a quick plug-in design, which is convenient to connect with the air pressure control system and ensures the stability of air pressure transmission. The flexible pneumatic finger 2 is internally provided with an air chamber inner cavity 15 with a cavity width of 1.5 mm, and the cavity is connected to the air pressure control system through the air inlet 13 by a quick driving mechanism. After the gas is introduced, the flexible pneumatic finger 2 bends. In this embodiment, high-elasticity silicone material is selected as the main body of the actuator, which has good flexibility and recovery, and can be repeatedly bent without fatigue.

[0031] A plurality of flexible pressure sensors 1 are respectively arranged on the inner contact surface of the plurality of fingers 2, installed at the installation position 17, and made of piezoresistive material, and internally provided with a flexible conductive silicone channel 19. Similarly, the number of flexible pressure sensors 1 is preferably 4. In this embodiment, the piezoresistive material is a conductive composite material made of carbon nanotubes and flexible polymers. The resistance of the conductive composite material changes when it is deformed under stress, and the pressure can be sensed by the resistance change. The composite material is made into a thin sheet-shaped sensing unit and distributed in an array form on the inner contact surface of each knuckle to form a sensing network covering the entire gripping area.

[0032] The flexible conductive silicone channel 19 is distributed in the pressure sensor 1 in the form of a heart curve, and the remaining part is a flexible non-conductive silicone 20, that is, a flexible non-conductive silicone 20 shell. The flexible pressure sensor 1 is distributed on the contact surface (that is, the side in contact with the object) of the flexible pneumatic finger 2, covering the main stress area when the knuckles are bent. Among them, the signal input and output end 18 realizes the conduction of the sensor electric signal by introducing flexible conductive silicone into the flexible conductive silicone channel 19. The pressure distribution sensing is realized by the different electric signals generated by the different resistance changes of the sensing unit caused by the different pressures of the object in the contact area.

[0033] The sensing unit is connected to the signal processing module through flexible wires. The wires are made of soft material and deform with the bending of the knuckles, without affecting the movement. The signal processing module converts the resistance change of the sensing unit into an electric signal, which is transmitted to the main controller after filtering noise, and reflects the distribution of the contact pressure in real time. For example, when grabbing a spherical object, the pressure difference of different parts of the knuckles can be detected to determine whether the contact is uniform.

[0034] Gas is introduced into the air chamber inner cavity 15 (cavity width 1.5 mm) through the air inlet 13. The inner cavity 15 includes a bridge structure inside the air chamber for convenient printing. After the inner cavity 15 is inflated, the air chamber outer wall 14 (wall thickness 1.5 mm) is bent, realizing the grabbing function of the pneumatic finger 2. When the chamber is inflated, the silicone inside the knuckles expands, realizing the inward bending action. At the same time, the screw rod 12 structure realizes the opening and closing of the gripper, which can complete the ring-shaped grabbing of objects of different diameters. Auxiliary adjustment of the inflation of the cavity can adapt to asymmetric objects or adjust the grabbing posture. In this way, by adjusting the air pressure value of each chamber, the stiffness of the knuckles can be changed, and low stiffness mode is adopted for fragile objects and high stiffness mode is adopted for heavy objects.

[0035] In this embodiment, flexible conductive silicone is injected into the flexible conductive silicone channel 19 to transmit sensor electric signals. After pretreatment and adhesion, pressure distribution sensing, surface contour recognition, sliding object detection and material preliminary judgment can be realized, and a thin silicone protective film is covered on the surface. The sensing array is attached to the inner contact surface of the flexible knuckles. Before adhesion, the knuckle surface is wiped with alcohol to remove dirt and coated with silicone glue. The surface of the sensing unit is covered with a thin silicone protective film to ensure that the sensing array and the knuckle surface are tightly bonded without air gap, realizing pressure distribution sensing. In this embodiment, carbon nanotube / polymer composite material is directly injected into the sensor for preparation, with a thickness of 50-100 microns, good flexibility and tensile properties.

[0036] The gripper determines the object characteristics and generates a grasping strategy according to the positive correlation between the electric signal generated by the pressure sensor 1 when it is pressed (i.e. the touch pressure surface 16 is contacted by the target object to generate touch pressure) and the pressure, and adjusts the air pressure of each chamber to achieve adaptive grasping. In this embodiment, based on the pressure change gradient algorithm, the convex features on the surface of the object can be identified to provide a basis for grasping strategy adjustment and achieve surface contour recognition. In this embodiment, by monitoring the pressure change rate (threshold value 0.5N / s), feedback adjustment can be triggered within 0.1 seconds when the object shows a sliding trend, and sliding object detection can be achieved.

[0037] A plurality of rods 4 correspond to a plurality of bases 3, and one end of each rod 4 is rotatably connected to the corresponding base 3. A plurality of members 5 correspond to a plurality of rods 4, and the other end of each rod 4 is rotatably connected to one end of the corresponding member 5. A fixed disc 6 is connected to the other end of the plurality of members 5. In this embodiment, the number of rods 4 is preferably 4.

[0038] The lead screw 12 is connected to the output shaft of the stepper motor 9, and the lead screw sleeve 21 is screwed with the lead screw 12. A plurality of pull rods 10 correspond to a plurality of rods 4, and one end of each pull rod 10 is rotatably connected to the lead screw sleeve 21 through the connecting part 11, and the other end is rotatably connected to the corresponding rod 4. In this embodiment, the number of pull rods 10 is preferably 4. When the stepper motor 9 rotates, it drives the lead screw 12 and the lead screw sleeve 21 to move relative to each other, causing each pull rod 10 to pull each finger 2 to achieve a grasping action. The motor housing 7, the fixed housing 8, and the stepper motor 9 together constitute the driving part of the gripper, which drives the linear motion of the member 5 by rotating the lead screw 12, and further drives the member 5 connected to the flexible pneumatic finger 2 to move linearly.

[0039] The intelligent control system is used to process the visual images captured based on a deep learning algorithm, identify the category and state of the target object, generate an optimal grasping scheme according to the identification result and a preset knowledge base, and finally adjust the air pressure value of each chamber in real time according to the pressure information detected by the pressure sensor 1, so that the grasping force of each finger 2 is kept within a preset safe range.

[0040] In this embodiment, the intelligent control system includes a microcontroller, a visual recognition module (which uses the YOLOV11 deep learning algorithm to realize target recognition and positioning), and an algorithm processing unit. The algorithm processing unit can complete the generation of a grasping strategy, force feedback control based on a PID algorithm, and abnormality processing. The visual recognition module can recognize target objects within a certain diameter range and achieve precise positioning. The visual recognition module collects images of the surrounding environment through a camera, identifies the position and approximate shape of the target object, and simultaneously, the visual recognition module has an upper limit of 8400 on the number of bounding boxes, the maximum input image size is 840*840 pixels, the recognition range of the desktop at the same time is 10m*10m, and the highest accuracy is 2cm*2cm*2cm.

[0041] After the system is started, the visual recognition module first scans the scene, identifies the target object through a deep learning algorithm, determines the category (such as fruits, parts, tools, etc.) and basic characteristics (such as size, shape, etc.) of the target object. The controller selects a suitable grasping scheme from a pre-set strategy library based on this information, for example, a three-point ring grasping scheme for circular objects and a side wrapping grasping scheme for long strip-shaped objects. When executing the grasping, the air pressure control module first drives the fingers to move along a pre-set trajectory, and after contacting the object, the tactile sensing system feeds back pressure data in real time. The main controller dynamically adjusts the air pressure based on the pressure size - if the pressure is too small, the inflation amount is increased to enhance the grasping force; if the pressure is too large, the inflation amount is reduced to avoid damaging the object, forming a closed-loop control.

[0042] The force feedback control adjusts the air pressure value in real time based on the PID algorithm to control the pressure fluctuation range, which can achieve non-destructive grasping of fragile objects such as eggs. The abnormality processing mechanism triggers an emergency strategy when detecting a pressure mutation or continuous sliding, such as increasing the air pressure or adjusting the posture.

[0043] The visual recognition module uses the YOLOV11 deep learning algorithm, which can recognize target objects within a diameter range of 10m*10m with a positioning accuracy of 2cm*2cm*2cm. The grasping strategy automatically matches the optimal grasping scheme based on the object shape (spherical, cylindrical, irregular), hardness (soft, hard), and weight level (light, medium, heavy).

[0044] The lightweight integrated module uses a flexible printed circuit board, which integrates sensing, control, and driving circuits through a laser etching process, realizes high-density integration, flexible interconnection, and modular connection, and a micro electromagnetic valve array ensures precise air pressure control. The flexible printed circuit board of the lightweight integrated module uses a polyimide base material, and the circuit pattern is prepared through a laser etching process. It can withstand a certain angle of bending without affecting the electrical performance. It integrates multiple sensor channels, air pressure control channels, and complete control circuits, and the micro electromagnetic valve array uses piezoelectric driving technology to achieve precise air pressure control, which can achieve precise air pressure control.

[0045] In this embodiment, high-density integration can be achieved. Here, examples are given to illustrate, such as 32 sensor channels, 16 air pressure control channels and complete control circuit are integrated in a volume of 100mmx80mmx20mm, so the weight is only 280g. The polyimide substrate FPCB with a thickness of 0.1mm is used, which can withstand ±60° bending without affecting the electrical performance, ensuring the reliable connection of each component during movement. It should be noted that the size described here is only an implementation to facilitate understanding, and in actual design, the size can be adjusted as needed. Each functional module (sensing, control, driving) is connected through a standardized interface, supporting quick replacement and upgrade. For example, after the sensor unit wears out, there is no need to replace the entire gripper, but only to replace the sensor module, reducing maintenance costs.

[0046] The overall shell of the gripper of this embodiment uses lightweight high-strength composite materials, which not only protects the internal components, but also reduces weight, making it easy to install on various robot arms. The surface of the shell is sealed to prevent dust and moisture from entering the interior and affecting the operation of the circuit, suitable for complex environments such as humidity and dust. Internal components are arranged according to function, and controller, air pressure valve, power supply and other modules are fixed through flexible supports to reduce the impact of vibration on precision components and improve system stability.

[0047] This embodiment takes the fruit and vegetable picking scene as an example, and the system workflow is as follows: 1. Target identification: the vision module scans the fruit tree, identifies the position and shape of the ripe fruit, and distinguishes between fruit and leaves; 2. Strategy planning: the main controller plans the motion path of the finger according to the size and position of the fruit, avoiding obstacles such as branches and leaves; 3. Flexible grasping: the finger bends to contact the fruit according to the planned path, and the sensing system detects the pressure and adjusts to the appropriate force to ensure that the fruit is held without being squeezed; 4. Stable handling: after grasping is stable, the robot arm moves to the collection basket, releases the finger and places the fruit, completing a grasp. In the precision parts assembly scene, the system precisely locates the position of the parts through vision, and controls the small force by combining with the tactile feedback, achieving gentle grasping and precise placement, avoiding damage to the parts.

[0048] The gripper of the embodiment has the following advantages: 1. Multi-modal perception and cooperative control: Through the fusion of visual and tactile perception, a comprehensive understanding of the environment and target objects is achieved. The visual system provides macro position and shape information, while the tactile system provides fine feedback during contact. The two systems work together to enable the gripper to adapt to complex and unstructured environments. 2. Flexible materials and structural innovation: The use of special silicone materials and 3D printing technology enables the actuator to have sufficient flexibility while maintaining a certain structural strength. The non-constant cross-section design and multi-chamber structure further enhance the flexibility and adaptability of the motion, enabling a variety of complex grasping actions. 3. Lightweight and integrated design: Through the use of flexible circuit boards, miniaturized components, and lightweight materials, the system weight is significantly reduced while the integration level is improved. Modular design enhances the maintainability and expandability of the system, making it easy to customize according to different application requirements. 4. Intelligent learning and optimization capabilities: The system has data storage and analysis capabilities, which can record relevant parameters and results of each grasping. Through learning and analysis of a large amount of data, the grasping strategy and control algorithm are continuously optimized to improve the success rate and efficiency of grasping, and to adapt to more complex scenarios.

[0049] The gripper of the embodiment can be applied in the following fields: 1. Agricultural automation: In addition to fruit and vegetable picking, it can also be applied to crop sorting, flower transplanting, etc., reducing manual labor intensity and improving agricultural production efficiency. 2. Logistics and warehousing: In the scenarios of express sorting and warehouse goods handling, it can quickly and accurately grasp packages of different shapes and weights, improving the level of logistics automation. 3. Medical assistance: It can be designed as a minimally invasive surgical tool, achieving gentle operation on human tissues through flexible structure and precise force control, reducing surgical trauma. 4. Rescue in disaster: In disaster sites such as earthquakes and fires, it can replace manual handling of dangerous goods and search for debris, reducing the risk to rescue personnel.

[0050] In summary, compared with existing flexible pneumatic grippers, the self-adaptive flexible pneumatic gripper based on intelligent perception of the embodiment has the following advantages: 1. The adaptive flexible pneumatic gripper based on intelligent sensing, which realizes bidirectional bending and stiffness adjustment by connecting the independent main driving cavity and auxiliary adjusting cavity in the flexible pneumatic finger 2 to the air pressure control system through independent air paths. When the main driving cavity is inflated, the finger 2 bends inward, and the bending degree can be adjusted according to the inflation amount to realize the encircling action on the object. When the auxiliary adjusting cavity is inflated alone, the finger 2 will slightly twist outward to adapt to the asymmetric shape of the object. When both cavities are inflated, the knuckles can complete the bending action, for example, when grabbing an object with a protrusion on the side, the double-chamber coordinated bending adjusts the contact angle to ensure stable grabbing. The flexible conductive silicone is injected into the flexible conductive silicone channel 19 to transmit sensor electrical signals, and after pre-treatment and fitting, it can realize pressure distribution sensing, surface contour recognition, and sliding object detection, solving the technical problems of insufficient degrees of freedom and lack of intelligent sensing in existing flexible grippers.

[0051] 2. The adaptive flexible pneumatic gripper based on intelligent sensing has excellent environmental adaptability and precise force control capability, can grasp various material objects such as fragile objects and hard objects, has low system response time and reduced power consumption compared to traditional solutions, realizes efficient, precise, and flexible grabbing of diversified objects in unstructured environments, and has broad application prospects and market value.

[0052] 3. The adaptive flexible pneumatic gripper based on intelligent sensing can adapt to the grabbing of objects with complex shapes and positions in unstructured environments through multi-degree-of-freedom motion and intelligent sensing, without the need for precise pre-positioning, and has excellent environmental adaptability. The flexible tactile sensing system of the gripper cooperates with a closed-loop control algorithm to realize precise control of the grabbing force, can grasp objects of various materials such as eggs (fragile objects) and metal parts (hard objects), and has wide applicability. The gripper also has fast response capability and low power consumption design, and through innovative structural design and intelligent control system, it realizes efficient, precise, and flexible grabbing of diversified objects in unstructured environments, and has broad application prospects and market value.

[0053] Embodiment 2 Please refer to Figure 4 The embodiment provides a mechanical arm based on intelligent sensing, which includes the adaptive flexible pneumatic gripper based on intelligent sensing in embodiment 1, and can also include a bottom plate 25, a plurality of fixed suction cups 26, a visual recognition module 27 (which can be the visual recognition module in embodiment 1), a fixed support 23, and a plurality of rudders 22.

[0054] A plurality of fixed suction cups 26 are installed at the bottom of the bottom plate 25, and the number of fixed suction cups 26 is preferably 6 in this embodiment. A visual recognition module 27 is installed on the bottom plate 25 and is used to shoot a visual image of a target object and identify the category and state of the object. A fixed support 23 is installed on the bottom plate 25, and a plurality of rudders 22 are installed on the fixed support 23 and are used to realize the rotation and extension of the whole gripper. The rudders 22 are connected to each other through the fixed support 23 to form a driving part, and the three rudders 22 realize the rotation and extension function of the whole mechanical gripper. The sensor interface and the PWM rudder 22 control interface are placed on the main control board 24. Among them, the visual recognition module 27 is fixed on the bottom plate 25 to realize the visual recognition function of the static object.

[0055] Embodiment 3 The embodiment provides a mechanical arm based on intelligent sensing, and the mechanical arm comprises the adaptive flexible pneumatic gripper based on intelligent sensing in the embodiment 1. It should be noted that, in the mechanical arm, structures other than the gripper part can adopt existing structures, and of course, the existing mechanical arm can be directly replaced with the gripper in the embodiment 1. The effect of the mechanical arm has been described in the embodiment 1, and will not be repeated here.

[0056] Embodiment 4 The embodiment provides a grabbing method based on intelligent sensing, which is applied to the adaptive flexible pneumatic gripper based on intelligent sensing in the embodiment 1 and can also be applied to the mechanical arm based on intelligent sensing in the embodiment 2. The grabbing method comprises the following steps: (1) processing a shot visual image based on a deep learning algorithm to identify the category and state of a target object; meanwhile, determining the object characteristics according to the pressure signal generated by the pressure sensor 1 when the pressure sensor 1 is pressed; (2) generating an optimal grabbing scheme (grabbing strategy) according to the identification result, the object characteristics and a preset knowledge base; and (3) adjusting the air pressure values of the chambers in real time according to the grabbing strategy, so that the grabbing force of each finger 2 is kept within a preset safe range.

[0057] The above merely describes the preferred embodiments of the present application and is not used to limit the present application, and any modification, equivalent replacement and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A smart-sensing based adaptive flexible pneumatic gripper, characterized in that, It comprises: a plurality of bases; a plurality of flexible pneumatic fingers corresponding to be installed on the plurality of bases; the inside of each flexible pneumatic finger is provided with a main driving cavity and an auxiliary adjusting cavity which are independent of each other, and each cavity is connected to an external air pressure control system through an independent air path; the main driving cavity drives the corresponding finger to bend to one side when inflated, and the auxiliary adjusting cavity drives the corresponding finger to bend to the other side when inflated; a plurality of flexible pressure sensors which are respectively arranged on the inner side contact surface of the plurality of fingers and are made of piezoresistive material, and are internally provided with a flexible conductive silica gel channel; wherein the gripper determines the object characteristics and generates a grasping strategy according to the electric signal generated by the pressure sensor when it is pressed, and adjusts the air pressure of each cavity to achieve adaptive grasping.

2. The smart-sensing based adaptive flexible pneumatic gripper of claim 1, wherein, The gripper further comprises: a plurality of rods corresponding to the plurality of bases, one end of each rod being rotatably connected to the corresponding base; a plurality of members corresponding to the plurality of rods, the other end of each rod being rotatably connected to one end of the corresponding member; a fixed disc connected to the other end of the plurality of members.

3. The smart-sensing based adaptive flexible pneumatic gripper of claim 2, wherein, The gripper further comprises: a stepper motor; a lead screw connected to the output shaft of the stepper motor; a lead screw sleeve screwed with the lead screw; a plurality of pull rods corresponding to the plurality of rods; one end of each pull rod is rotatably connected to the lead screw sleeve, and the other end is rotatably connected to the corresponding rod; wherein when the stepper motor rotates, it drives the lead screw and the lead screw sleeve to move relative to each other, so that each pull rod pulls each finger to achieve a grasping action.

4. The smart-sensing based adaptive flexible pneumatic gripper of claim 1, wherein, Each finger comprises two air chamber outer walls with a concave-convex cross section and a partition plate between the two air chamber outer walls, and the two air chamber outer walls and the partition plate form two cavities.

5. The smart-sensing based adaptive flexible pneumatic gripper of claim 1, wherein, The piezoresistive material is a conductive composite material made of carbon nanotubes and flexible polymers, and the finger is integrally formed by 3D printing using a silica gel-based material.

6. The smart-sensing based adaptive flexible pneumatic gripper of claim 1, wherein, The gripper further comprises: an intelligent control system for processing the visual image by deep learning algorithm, identifying the category and state of the target object, generating the optimal grasping scheme according to the identification result and the preset knowledge base, and finally adjusting the air pressure value of each cavity according to the pressure information detected by the pressure sensor, so that the grasping force of each finger remains within a preset safe range.

7. The smart-sensing based adaptive flexible pneumatic gripper of claim 1, wherein, Each finger joint adopts a non-equal cross-section design, and the root width is greater than the tip width, and a mechanical limiting structure is arranged at the root of the finger joint to limit the maximum bending angle.

8. The smart-sensing based adaptive flexible pneumatic gripper of claim 1, wherein, The flexible conductive silica gel channel is distributed in the pressure sensor in the shape of a cardioid, and the remaining part is flexible non-conductive silica gel.

9. A smart perception based robotic arm, characterized in that, It comprises at least one adaptive flexible pneumatic gripper based on intelligent sensing according to any one of claims 1-8.

10. The smart perception based robotic arm as claimed in claim 9, wherein, The mechanical arm further comprises: a bottom plate; a plurality of fixed suction cups mounted on the bottom of the bottom plate; a visual recognition module mounted on the bottom plate and used for shooting visual images of target objects and identifying the category and state of the objects; a fixed support mounted on the bottom plate; a plurality of rudders mounted on the fixed support and used for realizing the rotation and stretching of the whole gripper.

Images