Natural tactile interaction method and assembly for data-driven bionic robots

Abstract

The application provides a natural tactile interaction method and component of a data-driven bionic robot, and the method comprises the following steps: collecting tactile interaction process data of a real organism corresponding to the bionic robot and a user and statistically aggregating the data to obtain a training data sequence; establishing a data-driven interaction prediction model and training the data-driven interaction prediction model by using the training data sequence; determining a tactile input signal generated by a tactile input system of the bionic robot acted on by the user; and controlling the natural tactile interaction of the bionic robot according to the tactile input signal by using the trained data-driven interaction prediction model. By the method of the application, the touch action of the user can be recognized by the tactile input system, the natural feedback that the bionic robot should give to the user can be calculated by the data-driven interaction prediction model, and the closed loop of the natural tactile interaction feedback can be realized by controlling the action of the bionic robot.

Description

Technical Field

[0001] This invention relates to the field of robotics, and in particular to a data-driven biomimetic robot's natural tactile interaction method and components. Background Technology

[0002] With the continuous development of technology, various types of robots are emerging, especially biomimetic robots, such as quadrupedal biomimetic robots, which are beginning to replace pets. However, biomimetic robots rely heavily on voice interaction in human-computer interaction, which can only convey purpose but lacks the ability to convey emotions. Their movements are stiff, failing to achieve a truly biomimetic effect and making effective interaction difficult. Touch is one of the most important channels for human-biomimetic robot interaction. Currently, tactile research in the field of robotics mainly focuses on local perception, such as in robotic arms, grippers, and localized robotic skin, but no specific implementation plan has been clearly provided for a data-driven natural tactile interaction method for biomimetic robots. Summary of the Invention

[0003] This invention provides a data-driven natural tactile interaction method and components for bionic robots, addressing the lack of research on data-driven natural tactile interaction methods for bionic robots in the prior art. It can identify the user's touch actions through a tactile input system, calculate the natural feedback that the bionic robot should give to the user through a data-driven interaction prediction model, and achieve a closed loop of natural tactile interaction feedback by controlling the actions of the bionic robot.

[0004] This invention provides a data-driven method for natural tactile interaction of a bionic robot, comprising: collecting tactile interaction data between a user and a real organism corresponding to the bionic robot and statistically summarizing the data to obtain a training data sequence; establishing a data-driven interaction prediction model and training the data-driven interaction prediction model using the training data sequence; determining the tactile input signal generated by the user acting on the tactile input system of the bionic robot; and controlling the natural tactile interaction of the bionic robot based on the tactile input signal using the trained data-driven interaction prediction model.

[0005] According to a data-driven natural tactile interaction method for a bionic robot provided by the present invention, the tactile input system includes a tactile sensing device and a tactile sensing signal processing device, wherein the tactile sensing device is communicatively connected to the tactile sensing signal processing device; determining the tactile input signal generated by the tactile input system acting on the bionic robot includes: controlling the tactile sensing device to detect the tactile sensation of the bionic robot and generate a tactile sensing signal; and controlling the tactile signal processing device to perform signal processing on the tactile sensing signal to obtain the tactile input signal.

[0006] According to the present invention, a data-driven natural tactile interaction method for a bionic robot is provided, wherein the tactile sensing device is a large-format, high-density flexible tactile sensor, which is disposed in a large-format, high-density manner at a predetermined location on the bionic robot and is attached to and connected to the bionic robot.

[0007] According to a data-driven bionic robot natural tactile interaction method provided by the present invention, an electromagnetic shielding structure is provided outside the tactile sensing device to reduce the noise of the tactile sensing signal.

[0008] According to a data-driven bionic robot natural tactile interaction method provided by the present invention, a tactile guidance structure is provided outside the electromagnetic shielding structure for tactile guidance of the user.

[0009] According to the present invention, a data-driven natural tactile interaction method for a biomimetic robot is provided, wherein the training data sequence includes the user's action tags and the action tags of the real organism, and the user's action tags and the action tags of the real organism have a mapping relationship.

[0010] According to the present invention, a data-driven natural tactile interaction method for a bionic robot is provided. The method involves controlling the natural tactile interaction of the bionic robot based on the tactile input signal using a trained data-driven interaction prediction model. The method includes: extracting the user's action features from the tactile input signal using a convolutional neural network and performing classification prediction to obtain the user's action classification prediction result; modeling and predicting the natural tactile interaction feedback of the bionic robot based on the user's action classification prediction result using a sequence prediction neural network to obtain the robot's action prediction result; and controlling the natural tactile interaction of the bionic robot based on the robot's action prediction result.

[0011] The present invention also provides a data-driven natural tactile interaction system for a bionic robot, comprising: a data acquisition module for acquiring tactile interaction data between a user and a real biological entity corresponding to the bionic robot and statistically summarizing the data to obtain a training data sequence; a training module for establishing a data-driven interaction prediction model and training the data-driven interaction prediction model using the training data sequence; a determination module for determining the tactile input signal generated by the user acting on the tactile input system of the bionic robot; and a control module for controlling the natural tactile interaction of the bionic robot according to the tactile input signal using the trained data-driven interaction prediction model.

[0012] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement a natural tactile interaction method for a data-driven bionic robot as described above.

[0013] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a natural tactile interaction method for a data-driven bionic robot as described above.

[0014] This invention provides a data-driven method and components for natural tactile interaction of bionic robots, comprising: collecting tactile interaction data between a user and a real biological entity corresponding to the bionic robot and statistically summarizing this data to obtain a training data sequence; establishing a data-driven interaction prediction model and training the model using the training data sequence; determining the tactile input signal generated by the user's action on the bionic robot's tactile input system; and controlling the natural tactile interaction of the bionic robot based on the tactile input signal using the trained data-driven interaction prediction model. Through this data-driven natural tactile interaction method for bionic robots, the user's touch actions can be recognized through the tactile input system, and the natural feedback that the bionic robot should provide to the user can be calculated through the data-driven interaction prediction model. By controlling the actions of the bionic robot, a closed loop of natural tactile interaction feedback can be achieved, enabling the bionic robot to have a more realistic bionic effect, smoother interaction actions, and a more natural and fluid human-computer interaction process. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating the natural tactile interaction method for a data-driven bionic robot provided by the present invention.

[0017] Figure 2 This is a schematic diagram of the original model structure of the bionic robot provided by the present invention;

[0018] Figure 3 This is a schematic diagram of the structure of the bionic robot provided by the present invention after being equipped with a tactile sensing device;

[0019] Figure 4 This is a schematic diagram of the algorithm processing flow of the bionic robot provided by the present invention;

[0020] Figure 5 This is a schematic diagram of the structure of the natural tactile interaction system of the data-driven bionic robot provided by the present invention;

[0021] Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0023] The following is combined with Figures 1-6 This invention describes a data-driven natural tactile interaction method and components for a biomimetic robot.

[0024] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating a data-driven natural tactile interaction method for a biomimetic robot provided by the present invention.

[0025] Please refer to Figure 2 , Figure 2 This is a schematic diagram of the original model structure of the bionic robot provided by the present invention.

[0026] This invention provides a data-driven method for natural tactile interaction in biomimetic robots, comprising:

[0027] 101: Collect tactile interaction data between users and the real organisms corresponding to the bionic robot, and statistically summarize the data to obtain a training data sequence;

[0028] 102: Establish a data-driven interactive prediction model and train the data-driven interactive prediction model using training data sequences;

[0029] 103: Determine the tactile input signals generated by the user's tactile input system on the bionic robot;

[0030] 104: Using a trained data-driven interaction prediction model to control the natural tactile interaction of a bionic robot based on tactile input signals.

[0031] Touch is one of the most primitive sensory channels in living organisms. Studies have found that emotional bonds with pet dogs can reduce loneliness in humans, and tactile contact with dogs is beneficial to their owners' health. Touch is also one of the most important channels for human-machine interaction. Today, pet robots, such as quadrupedal bionic robots, can exhibit vivid animal-like behaviors when interacting with humans, such as stretching their bodies, shaking hands, and rolling over. With the development of emotional design in pet robots, they are increasingly capable of recognizing human emotions and responding appropriately. However, at the technological level, tactile-based interaction is explored relatively less than visual and auditory-based interaction. Emotional touch plays a crucial role in human-computer interaction. To fill this gap, this invention provides a data-driven method for natural tactile interaction of bionic robots. First, tactile interaction data between the user and the real organism corresponding to the bionic robot is collected and statistically summarized to obtain a training data sequence. This training data sequence can include touch actions performed by the user on the real organism corresponding to the bionic robot, such as gestures, as well as feedback actions from the real organism to the user's touch actions, such as full-body movements, foot movements, hip movements, or head movements. A data-driven interaction prediction model, which can be a deep learning interaction model, is then established. This model is then trained based on the training data sequence. When the user's touch action is applied to the bionic robot, the robot's tactile input system recognizes the user's tactile action to obtain a tactile input signal. Finally, the tactile input signal is input into the trained data-driven interaction prediction model, which predicts the natural tactile feedback behavior of the bionic robot, thereby obtaining the tactile interaction actions of the bionic robot for control of its tactile interaction.

[0032] In summary, the data-driven natural tactile interaction method for bionic robots of the present invention can recognize the user's touch actions through a tactile input system, calculate the natural feedback that the bionic robot should give to the user through a data-driven interaction prediction model, and realize a closed loop of natural tactile interaction feedback by controlling the actions of the bionic robot. This enables the bionic robot to have a more realistic bionic effect, smoother interaction actions, and a more natural and fluid human-computer interaction process.

[0033] Based on the above embodiments:

[0034] Please refer to Figure 3 , Figure 3 A schematic diagram of the structure of the bionic robot provided by the present invention after equipping it with a tactile sensing device.

[0035] In a preferred embodiment, the tactile input system includes a tactile sensing device and a tactile sensing signal processing device, wherein the tactile sensing device and the tactile sensing signal processing device are communicatively connected; determining the tactile input signal generated by the tactile input system acting on the bionic robot includes: controlling the tactile sensing device to detect the tactile sensation of the bionic robot and generate a tactile sensing signal; and controlling the tactile signal processing device to process the tactile sensing signal to obtain a tactile input signal.

[0036] In this embodiment, the tactile input system may include a tactile sensing device and a tactile sensing signal processing device. The tactile sensing device detects the tactile sensation of the bionic robot and generates a tactile sensing signal, which is then transmitted to the tactile signal processing device. The tactile signal processing device summarizes and processes the tactile sensing signal to obtain a tactile input signal. The overall system has high reliability and data accuracy.

[0037] Furthermore, the tactile sensing signal processing device can be a data acquisition board, such as a sensor reading board. The tactile sensing device can transmit the tactile signal to the data acquisition board via a ribbon cable, and multiple data acquisition boards (determined by the number of sensors) can aggregate and process the tactile signals. Piezoelectric thin-film sensors are formed by transferring force-sensitive materials, silver paste, etc., onto a flexible thin-film substrate using a precision printing process, followed by drying and curing. When pressure is applied to the sensor, the resistance decreases with increasing pressure; the reciprocal of the resistance has an approximately linear relationship with the pressure.

[0038] In a preferred embodiment, the tactile sensing device is a large-format, high-density flexible tactile sensor, which is disposed in a large-format, high-density manner at a predetermined location on the bionic robot and is attached to and connected to the bionic robot.

[0039] In order to more accurately detect the tactile sensation of the bionic robot, in this example, the tactile sensing device can be a large-format, high-density flexible tactile sensor. It is set in a large-format, high-density manner at a preset location on the bionic robot and is attached to the bionic robot, so as to detect the tactile sensation of the bionic robot more accurately and sensitively.

[0040] Tactile interaction is crucial for natural and seamless human-computer interaction, and accurate and efficient tactile image recognition methods are essential to ensuring its quality. Tactile images can be generated by pressure-sensitive materials, smart fabrics, piezoelectric thin-film sensors, etc. They are typically interpreted as a pressure distribution matrix on the tactile sensing surface. There are many methods for processing the sensing matrix, and the recognition of the sensing matrix is ​​largely the same as that of digital images. For example, a quadrupedal bionic robot dog uses large-format, high-density flexible pressure sensors to sensitively and accurately perceive various natural human tactile gestures on the robot dog and perform corresponding actions like a real pet dog. This is achieved by attaching a layer of piezoelectric thin-film sensors to the robot dog's skin, covering a large and uniquely shaped area, including the head, back, hips, and legs. Sensors are installed on the robot dog's head, back, hips, and forelimbs, with the number of sensors determined by the robot's surface area.

[0041] In a preferred embodiment, an electromagnetic shielding structure is provided on the outside of the tactile sensing device to reduce noise in the tactile sensing signal.

[0042] When a tactile sensing device detects the touch of a bionic robot, noise may be present. In this embodiment, an electromagnetic shielding structure can be set on the outside of the tactile sensing device. For example, electromagnetic shielding materials such as copper foil and aluminum foil can be attached to the surface of the tactile sensing device to reduce tactile signal noise and improve the accuracy of tactile signal detection.

[0043] In a preferred embodiment, a tactile guidance structure is provided outside the electromagnetic shielding structure for tactile guidance of the user.

[0044] To facilitate tactile guidance for users, in this embodiment, a tactile guidance structure can be set outside the electromagnetic shielding structure. For example, the color and material of the robot's exterior can be used to guide users to subconsciously touch the robot's tactile sensing devices to input tactile input signals. Specifically, for aesthetic appearance and comfortable touch, without affecting the sensing performance of the pressure sensor, the head, back, feet, or buttocks of the robot dog, where the tactile sensing devices are located, can be covered with fur or other materials. The tactile guidance structure allows users to unconsciously touch the locations of the tactile sensing devices on the bionic robot, improving the accuracy and efficiency of tactile input signal acquisition. Users can touch the robot dog as if touching a real dog, and the robot dog can react similarly to a real dog by sensing the gesture. For example, touching the dog's head will trigger the robot dog to tilt its head back and turn its head left and right. The relationship between the touch gesture and the robot dog's feedback is predicted by modeling real human-dog interaction data.

[0045] In a preferred embodiment, the training data sequence includes user action labels and real-world animal action labels, with a mapping relationship between the user action labels and the real-world animal action labels.

[0046] Emotional communication between humans and tactile organisms involves two processes. Human emotions change in response to the touch of the tactile organism, and the tactile organism reacts accordingly to express its emotions upon sensing touch. In this embodiment, by collecting and statistically summarizing data on the tactile interaction between the user and the corresponding real organism of the bionic robot, a training data sequence can be obtained. This training data sequence can include the user's action labels and the real organism's action labels, and there is a mapping relationship between the user's action labels and the real organism's action labels.

[0047] Taking a quadrupedal bionic dog robot as an example, tactile gestures are defined as the combination of human tactile gestures and interactive parts of the dog's body, such as stroking its head. These tactile gestures are dynamic, meaning they are accompanied by changes in the position of the contact surface over a certain period of time, forming a series of tactile images. When humans touch a dog, they usually inject their own feelings or intentions, and the dog responds to human emotions, thus creating an emotional interaction. To simulate the real human-dog interaction experience using data-driven deep learning methods, we collected data on dog owners' gesture preferences when interacting with their pet dogs in real life, as well as the dogs' different reactions to different gestures. By collecting and statistically analyzing data on the tactile interaction process between humans and real dogs, we summarized the following labels: there are 13 categories of user-robot dog interaction gestures, and 11 different interaction parts, totaling 81 different types of gesture labels. Examples include touching, patting, and stroking. Typical feedback actions of the robot dog are divided into four categories: whole-body movements, foot movements, hip movements, and head movements, with 44 different types. There is a mapping relationship between gesture labels and tactile feedback labels of the robot dog. Therefore, the method of this embodiment can provide training data for training the data-driven interaction prediction model, so that the trained data-driven interaction prediction model can be used to control the natural tactile interaction of the bionic robot according to the tactile input signal, thereby realizing a closed loop of natural tactile interaction feedback.

[0048] Please refer to Figure 4 , Figure 4 This is a schematic diagram of the algorithm processing flow for the bionic robot provided by the present invention.

[0049] As a preferred embodiment, a trained data-driven interaction prediction model is used to control the natural tactile interaction of the bionic robot based on tactile input signals. This includes: extracting the user's action features from the tactile input signals through a convolutional neural network and performing classification prediction to obtain the user's action classification prediction results; modeling and predicting the natural tactile interaction feedback of the bionic robot based on the user's action classification prediction results through a sequence prediction neural network to obtain the robot's action prediction results; and controlling the natural tactile interaction of the bionic robot based on the robot's action prediction results.

[0050] Based on training data sequences, this embodiment trains a data-driven interaction prediction model and uses this model to control the natural tactile interaction of the bionic robot according to tactile input signals. Specifically, a gesture recognition algorithm based on convolutional neural networks is used to recognize the user's tactile gestures. The tactile image sequence is used as input, and the gesture classification result is used as output. A sequence prediction neural network is then used to model and predict the natural tactile interaction feedback of the bionic robot based on the user's action classification prediction result. Testing shows that this scheme achieves gesture classification accuracy and action prediction accuracy both above 98%. Sequence prediction is a classic problem in natural language processing. The tactile gesture classification backbone model can be any video / image classification algorithm, such as ResNet, Swing Transformer, or ViT (Vision Transformer). The interactive feedback prediction backbone model can be any sequence prediction model, such as Transformer, LSTM (Long Short-Term Memory), or RNN (Recurrent Neural Network), which improves computational efficiency while maintaining good predictive ability. In the tactile interaction between humans and robot dogs, it can predict the dog's movements well from human gestures.

[0051] Please refer to Figure 5 , Figure 5 A schematic diagram of the structure of the natural tactile interaction system of the data-driven bionic robot provided by the present invention.

[0052] The present invention also provides a data-driven natural tactile interaction system for a bionic robot, comprising: a data acquisition module 501, used to acquire tactile interaction process data between a user and a real organism corresponding to the bionic robot and to statistically summarize the data to obtain a training data sequence; a training module 502, used to establish a data-driven interaction prediction model and train the data-driven interaction prediction model using the training data sequence; a determination module 503, used to determine the tactile input signal generated by the user's action on the tactile input system of the bionic robot; and a control module 504, used to control the natural tactile interaction of the bionic robot according to the tactile input signal using the trained data-driven interaction prediction model.

[0053] For a description of the natural tactile interaction system for a data-driven bionic robot provided by this invention, please refer to the above method embodiments; the invention itself will not be repeated here.

[0054] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6 As shown, the electronic device may include a processor 601, a communications interface 602, a memory 603, and a communication bus 604. The processor 601, communications interface 602, and memory 603 communicate with each other via the communication bus 604. The processor 601 can call logical instructions in the memory 603 to execute a data-driven natural tactile interaction method for a bionic robot. This method includes: collecting tactile interaction data between the user and the bionic robot from a real biological entity and statistically summarizing this data to obtain a training data sequence; establishing a data-driven interaction prediction model and training the model using the training data sequence; determining the tactile input signal generated by the user's action on the bionic robot's tactile input system; and controlling the natural tactile interaction of the bionic robot based on the tactile input signal using the trained data-driven interaction prediction model.

[0055] Furthermore, the logical instructions in the aforementioned memory 603 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0056] On the other hand, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements a data-driven natural tactile interaction method for a bionic robot provided by the methods described above. The method includes: collecting tactile interaction process data between a user and a real organism corresponding to the bionic robot and statistically summarizing it to obtain a training data sequence; establishing a data-driven interaction prediction model and training the data-driven interaction prediction model using the training data sequence; determining the tactile input signal generated by the user's action on the tactile input system of the bionic robot; and controlling the natural tactile interaction of the bionic robot based on the tactile input signal using the trained data-driven interaction prediction model.

[0057] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0058] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0059] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A data-driven method for natural tactile interaction in biomimetic robots, characterized in that, include: Data on the tactile interaction between users and the real organisms corresponding to the bionic robots were collected and statistically summarized to obtain a training data sequence. The bionic robot is a robot dog; A data-driven interaction prediction model is established and trained using the training data sequence. The training data sequence is obtained by collecting and statistically summarizing the tactile interaction data between the user and the real animal corresponding to the bionic robot. The training data sequence includes the user's gesture labels and the tactile feedback labels of the real dog, and the user's gesture labels and the real dog's tactile feedback labels have a mapping relationship. Determine the tactile input signal generated by the user's tactile input system on the bionic robot; The trained data-driven interaction prediction model is used to control the natural tactile interaction of the bionic robot based on the tactile input signal; The step of using the trained data-driven interaction prediction model to control the natural tactile interaction of the bionic robot based on the tactile input signal includes: The user's action features are extracted from the tactile input signal by a convolutional neural network and classified and predicted to obtain the user's action classification prediction result. The bionic robot's natural tactile interaction feedback is modeled and predicted based on the user's action classification prediction results using a sequence prediction neural network, thereby obtaining the bionic robot's action prediction results and controlling the bionic robot's natural tactile interaction based on the bionic robot's action prediction results. The tactile input system includes a tactile sensing device and a tactile sensing signal processing device, and the tactile sensing device is communicatively connected to the tactile sensing signal processing device; The determination of the tactile input signal generated by the user's tactile input system on the bionic robot includes: The tactile sensing device is controlled to detect the tactile sensation of the bionic robot and generate tactile sensing signals; The tactile sensing signal processing device is controlled to process the tactile sensing signal to obtain the tactile input signal.

2. The data-driven natural tactile interaction method for bionic robots according to claim 1, characterized in that, The tactile sensing device is a large-format, high-density flexible tactile sensor, which is arranged in a large-format, high-density manner at a predetermined location on the bionic robot and is attached to and connected to the bionic robot.

3. The data-driven natural tactile interaction method for bionic robots according to claim 2, characterized in that, An electromagnetic shielding structure is provided on the outside of the tactile sensing device to reduce noise in the tactile sensing signal.

4. The natural tactile interaction method for a data-driven bionic robot according to claim 3, characterized in that, A tactile guidance structure is provided outside the electromagnetic shielding structure for tactile guidance of the user.

5. A data-driven natural tactile interaction system for a biomimetic robot, characterized in that, include: The data acquisition module is used to collect data on the tactile interaction process between the user and the real biological organism corresponding to the bionic robot, and to statistically summarize the data to obtain a training data sequence. The bionic robot is a robot dog; The training module is used to establish a data-driven interaction prediction model and train the data-driven interaction prediction model using the training data sequence. The training data sequence is obtained by collecting and summarizing the tactile interaction data between the user and the real animal corresponding to the bionic robot. The training data sequence includes the user's gesture labels and the tactile feedback labels of the real dog. The user's gesture labels and the tactile feedback labels of the real dog have a mapping relationship. A determining module is used to determine the tactile input signal generated by the user's tactile input system on the bionic robot. The control module is used to control the natural tactile interaction of the bionic robot based on the tactile input signal using the trained data-driven interaction prediction model. The step of using the trained data-driven interaction prediction model to control the natural tactile interaction of the bionic robot based on the tactile input signal includes: The user's action features are extracted from the tactile input signal by a convolutional neural network and classified and predicted to obtain the user's action classification prediction result. The bionic robot's natural tactile interaction feedback is modeled and predicted based on the user's action classification prediction results using a sequence prediction neural network, thereby obtaining the bionic robot's action prediction results and controlling the bionic robot's natural tactile interaction based on the bionic robot's action prediction results. The tactile input system includes a tactile sensing device and a tactile sensing signal processing device, and the tactile sensing device is communicatively connected to the tactile sensing signal processing device; The determination of the tactile input signal generated by the user's tactile input system on the bionic robot includes: The tactile sensing device is controlled to detect the tactile sensation of the bionic robot and generate tactile sensing signals; The tactile sensing signal processing device is controlled to process the tactile sensing signal to obtain the tactile input signal.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the natural tactile interaction method of the data-driven bionic robot as described in any one of claims 1 to 4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the natural tactile interaction method of the data-driven bionic robot as described in any one of claims 1 to 4.

Images