A topographic classification method and system based on flat panel capacitance tomography sensor

Abstract

The application discloses a kind of topographic classification method and system based on flat plate capacitance tomography sensor, belong to robot application technical field.The application first designs a flat plate ECT sensor array comprising eight electrodes, and builds a mobile robot body perception data acquisition system, and then the mutual capacitance data and inertial data of terrain are collected.Second, by building a multi-layer perception machine classification model, the capacitance data and inertial data are extracted, fused and classified, and after training the multi-layer perception machine model, the robot moves to the known terrain, and the terrain classification is completed by the multi-layer perception machine.The application uses the two modal information of inertia and capacitance, by combining the interactive vibration information of robot and terrain and the change of dielectric constant in terrain, can automatically identify a variety of different material terrains, has strong universality and practical significance.

Description

Technical Field

[0001] This invention relates to the field of robotics application technology, and in particular to a terrain classification method and system based on a planar capacitive tomography sensor. Background Technology

[0002] With the rapid development of robotics technology, the autonomous navigation capability of robots in unknown environments has become a research hotspot. In complex environments, robots need to accurately identify and classify different terrain types in order to select appropriate paths or adjust their movement patterns, thereby improving passability and reducing bumps during movement. Robotic systems are typically equipped with multiple sensors, mainly including external perception sensors and proprioception sensors. External perception sensors (such as RGB cameras, LiDAR, etc.) can capture visual information about the terrain, reflecting the appearance differences between different terrains. However, these sensors are susceptible to interference from lighting conditions and atmospheric environments. For example, under strong or weak light conditions, visual data may contain a large amount of noise, leading to a decrease in terrain classification accuracy. Furthermore, terrain classification methods that rely solely on visual sensors are vulnerable in outdoor environments and struggle to meet the needs of practical applications.

[0003] To overcome the limitations of external sensing sensors, researchers have begun exploring the use of proprioceptive sensors for terrain classification. Proprioceptive sensors (such as force-torque sensors, inertial measurement units, and tactile sensors) identify terrain features by measuring internal parameters (such as force, vibration, and acceleration) during robot interaction. For example, some studies have collected tactile data from robot-terrain interactions by mounting accelerometers or tactile sensors on robots and encoding this data into contextual representations for terrain classification. Other studies have designed specific pressing and rubbing motions, combined with force-torque sensors to measure foot forces, further improving the accuracy of terrain classification. However, the performance of proprioceptive sensors is highly dependent on the quality of robot-terrain interaction; different motion patterns can significantly impact the perception results. In complex environments, robots struggle to maintain stable motion, which can lead to a decrease in the reliability of proprioceptive data.

[0004] In summary, existing technologies still face many challenges in terrain classification. Therefore, developing a robust terrain classification method that is insensitive to environmental changes and movement patterns is of significant research and application value. Summary of the Invention

[0005] The present invention aims to at least partially solve one of the technical problems in the related art.

[0006] Therefore, this invention proposes a terrain classification method based on a planar capacitance tomography sensor. The ECT sensor is a non-contact, real-time, and relatively low-cost measurement technology, and its capacitance measurement can reflect the internal dielectric constant distribution of the terrain. By integrating a customized planar ECT sensor into the bottom of a mobile robot, capacitance measurement data is collected in real time during navigation to reflect the inherent differences in dielectric constant between different terrain types, thereby achieving efficient and accurate terrain classification.

[0007] To achieve the above objectives, this invention proposes a terrain classification method based on a planar capacitive tomography sensor, comprising:

[0008] Construct a planar ECT sensor array containing eight electrodes;

[0009] A mobile robot body perception data acquisition system was built based on the aforementioned flat panel ECT sensor array;

[0010] A terrain ontology perception dataset is acquired using a mobile robot ontology perception data acquisition system. Based on the corresponding material labels provided in the dataset, all elements in the capacitance time-series feature sequence and the inertial time-series feature sequence are extracted and used as input data for a multilayer perceptron network. The terrain classification label is used as output data for network training to obtain a classifier model that can identify different terrain types. The robot is controlled to move to the terrain to be measured, and the inertial data and mutual capacitance data measurements are recorded three times consecutively within a preset time. The real-time capacitance time-series feature sequence and the real-time inertial time-series feature sequence are calculated, and the trained classifier model is used to classify the terrain to obtain the terrain classification label of the terrain to be measured, thus completing the terrain classification.

[0011] The terrain classification method based on a planar capacitive tomography sensor according to embodiments of the present invention may also have the following additional technical features:

[0012] In one embodiment of the present invention, a planar ECT sensor array comprising eight electrodes is constructed, including:

[0013] The eight electrodes of the planar ECT sensor array are arranged in a 2*4 grid and implemented on a PCB board. The external dimensions of the sensor are 80mm*85mm. FR-4 material is laid under the sensor plate to isolate external electromagnetic interference. The protection electrode separates adjacent sensing units and suppresses capacitive coupling through a unified grounding reference grid.

[0014] In one embodiment of the present invention, a mobile robot body perception data acquisition system includes: a mobile robot, a multiplexer, a capacitance measurement device, a flat panel ECT sensor array, a microcomputer, and an inertial measurement unit; wherein,

[0015] The inertial measurement unit is used to measure inertial data;

[0016] The flat-panel ECT sensor array is installed on the bottom of the mobile robot and connected to a multiplexer. A microcomputer is connected to the multiplexer to control the gating logic. A capacitance measuring device is connected to the multiplexer to collect capacitance data and then save it to the microcomputer. A laptop computer is connected to the microcomputer via a wireless local area network to control the movement of the mobile robot.

[0017] In one embodiment of the present invention, training a multilayer perceptron network includes:

[0018] (3-1) Collect the ontological perception dataset of the terrain:

[0019] (3-1-1) Suppose that there are N known terrain features;

[0020] (3-1-2) When the capacitive sensor array is unloaded, the mutual capacitance between the two electrodes collected by the capacitive sensor array is C. b =[c1,c2,…,c 28 After the mobile robot moves onto terrain O, it continuously collects inertial data and mutual capacitance data. The mutual capacitance between the two electrodes collected by the capacitance sensor array each time is denoted as C. t =[c 1,t ,c 2,t ,…,c 28,t Each inertial measurement unit (IMU) collects inertial data including the robot's angular velocity ω and acceleration a along the x, y, and z axes. t =[ω x,t ,ω y,t ,ω z,t ,a x,t ,a y,t ,a z,t Record the measured values ​​of inertial data C1, C2, C3 and mutual capacitance data I1, I2, I3 three times consecutively within a certain period, and calculate the capacitance time-series characteristic sequence C. m =[C1-C b C2-C b [C2-C1,C3-C2] and inertial time series feature sequence I m =[I1,I2,I2-I1,I3-I2], and k capacitance and inertial data are collected on terrain O to obtain k capacitance time series C of terrain O. M,O =[C m,1 C m,2 ,…,C m,k ] and k inertial time series I M,O =[I m,1 ,I m,2 ,…,Im,k ], and record the category label of terrain O;

[0021] (3-1-3) For the capacitance time series C M,O and inertial time series I M,O Preprocessing is performed for C M,O and I M,O All elements in the array are normalized using the min-max method;

[0022] (3-1-4) Traverse the N types of terrain in step (3-1-1) and repeat step (3-1-2) to obtain the ontology-aware dataset D of the terrain. IC =[C M,1 C M,2 ,…,C M,N ,I M,1 ,I M,2 ,…,I M,N ];

[0023] (3-1-5) Construct a multilayer perceptron network. For 24-dimensional inertial time-series data, use a two-layer multilayer perceptron network with 22 and 16 neurons per layer, respectively; for 112-dimensional capacitive time-series data, use a three-layer MLP with 64, 32, and 16 neurons per layer, respectively. Combine the features extracted from these two types of data into a 32-dimensional vector. Pass the vector through a two-layer multilayer perceptron with 16 and 10 neurons per layer, respectively, and a softmax function to calculate the likelihood value of each label. Select the label with the highest likelihood value as the final terrain classification result.

[0024] (3-2) Based on the terrain ontology perception dataset D collected in step (3-1-4) IC The corresponding material labels provided in the document utilize a multilayer perceptron network built according to (3-1-5) as a terrain classifier, with the capacitance temporal feature sequence C as the basis. M,O All elements C m,i and inertial time series characteristic sequence I M,O All elements I M,,i As input to the multilayer perceptron, the terrain classification label is used as the output of the multilayer perceptron to train the network and obtain a classifier model that can identify terrain types.

[0025] (3-3) After the classifier model is trained in step (3-2), the robot is controlled to move to one of the known terrains O, and the inertial data and mutual capacitance data measurements are recorded three times in a row over a period of time. The corresponding capacitance time-series feature sequence C is then calculated. m and corresponding inertial time series feature sequence I mThen, the terrain is classified using a classifier model to obtain the terrain classification label of the terrain O to be tested, thus completing the terrain classification.

[0026] To achieve the above objectives, another aspect of the present invention proposes a terrain classification system based on a planar capacitive tomography sensor, comprising:

[0027] Sensor array building module for building a planar ECT sensor array containing eight electrodes;

[0028] The data acquisition system construction module is used to build a mobile robot body perception data acquisition system based on the flat panel ECT sensor array;

[0029] The classifier model training module is used to acquire the terrain ontology perception dataset using the mobile robot ontology perception data acquisition system, and extract all elements in the capacitance time-series feature sequence and all elements in the inertial time-series feature sequence based on the corresponding material labels provided in the terrain ontology perception dataset. These elements are used as input data for the multilayer perceptron network, and the terrain classification labels are used as output data for network training to obtain a classifier model that can identify different terrain types.

[0030] The terrain classification module is used to control the robot to move to the terrain to be measured, record the measurement values ​​of inertial data and mutual capacitance data three times in a preset time, calculate the real-time capacitance time series feature sequence and the real-time inertial time series feature sequence, and use the trained classifier model to classify the terrain to obtain the terrain classification label of the terrain to be measured, so as to complete the terrain classification.

[0031] The terrain classification method and system based on a planar capacitance tomography sensor of this invention first collects capacitance and inertial data of terrain samples, and then accurately classifies the terrain samples by constructing and training a multilayer perceptron network. Experiments based on this invention demonstrate that it is less affected by the robot's motion pattern, achieves good classification results for different motion models such as straight-line movement, curved movement, and movement at random speeds, and is unaffected by weather or lighting conditions, enabling all-weather terrain classification.

[0032] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0033] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0034] Figure 1 This is a flowchart of a terrain classification method based on a planar capacitive tomography sensor according to an embodiment of the present invention;

[0035] Figure 2 This is a design diagram of an eight-electrode planar ECT sensor array according to an embodiment of the present invention;

[0036] Figure 3 This is a schematic diagram illustrating the working principle of capacitive terrain data acquisition according to an embodiment of the present invention;

[0037] Figure 4 This is a physical image of a mobile robot terrain classification system according to an embodiment of the present invention;

[0038] Figure 5 This is a structural diagram of a terrain classification system based on a planar capacitive tomography sensor according to an embodiment of the present invention. Detailed Implementation

[0039] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0040] To enable those skilled in the art to better understand the present invention, 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0041] The following description, with reference to the accompanying drawings, illustrates a terrain classification method and system based on a planar capacitive tomography sensor according to an embodiment of the present invention.

[0042] Figure 1 This is a flowchart of a terrain classification method based on a planar capacitive tomography sensor according to an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes, but is not limited to, the following steps:

[0043] S1, Construct a planar ECT sensor array containing eight electrodes.

[0044] Specifically, the eight electrodes of the planar ECT sensor array are arranged in a 24-grid pattern and implemented on a PCB board. The external dimensions of the sensor are 80mm*85mm. In order to reduce environmental interference (such as external fields) in the planar array, FR-4 material is laid under the sensor plate to isolate external electromagnetic interference. In addition, the protection electrode separates the adjacent sensing units and suppresses capacitive coupling through a unified grounding reference grid.

[0045] S2, A mobile robot body perception data acquisition system is built based on the flat panel ECT sensor array.

[0046] Specifically, mobile robot body perception data acquisition system, such as Figure 4 As shown, it includes a mobile robot (Agilex LIMO in this embodiment), a multiplexer (MAX306), a capacitance measurement device (Ruby Development Board), and a planar ECT sensor array (PCB design as shown). Figure 2 The system consists of a microcomputer (Raspberry Pi 4B), an inertial measurement unit (MPU-6050), and a laptop computer. The working principle diagram for terrain data acquisition is shown below. Figure 3 As shown.

[0047] Among them, the inertial measurement unit is used to measure inertial data;

[0048] The flat panel ECT sensor array is installed on the bottom of the mobile robot and connected to a multiplexer. A microcomputer is connected to the multiplexer to control the gating logic. A capacitance measuring device is connected to the multiplexer to collect capacitance data and then save it to the microcomputer. A laptop computer is connected to the microcomputer via a wireless local area network to control the movement of the mobile robot.

[0049] In this embodiment of the invention, a planar ECT sensor array is mounted on the bottom of a mobile robot, a capacitance measuring device and a multiplexer are mounted on both sides of the mobile robot, a microcomputer is mounted on the top of the mobile robot, and an inertial measurement unit is mounted inside the mobile robot. The capacitance measuring device, the microcomputer, and the multiplexer together form a capacitance data acquisition unit. The multiplexer consists of two MAX306 chips. The first eight analog signal input ports CH1-CH8 of each MAX306 are connected to the eight channels of the planar array sensor. The address pins A0-A3 of each MAX306 are connected to the microcomputer, and the common terminal COM of each MAX306 is connected to the capacitance measuring device. The microcomputer is responsible for controlling the circuit selection between the planar array sensor and the capacitance measuring device, so as to realize the measurement of the mutual capacitance value between any two plates of the eight plates.

[0050] The capacitance measurement device includes a capacitance measurement chip, a signal generator, and a main control chip. The signal generator is responsible for generating periodic excitation voltages. The capacitance measurement chip converts the capacitance value collected by the sensor into 24-bit data and stores it in an internal register. The main control chip reads the data from the register and converts it into hexadecimal results. The results are then sent to the microcomputer in binary form via a serial port. The microcomputer then converts the data into the corresponding capacitance value.

[0051] In addition, the microcomputer is also responsible for controlling the movement of the mobile robot. The microcomputer can be connected to a laptop computer via network cable and local area network.

[0052] S3. A terrain ontology perception dataset is acquired using a mobile robot ontology perception data acquisition system. Based on the corresponding material labels provided in the terrain ontology perception dataset, all elements in the capacitance time-series feature sequence and all elements in the inertial time-series feature sequence are extracted and used as input data for a multilayer perceptron network. The terrain classification label is used as output data for network training to obtain a classifier model that can identify different terrain types. The robot is controlled to move to the terrain to be measured, and the measurement values ​​of inertial data and mutual capacitance data are recorded three times consecutively within a preset time. The real-time capacitance time-series feature sequence and the real-time inertial time-series feature sequence are calculated. The trained classifier model is used to classify the terrain to obtain the terrain classification label of the terrain to be measured, thus completing the terrain classification.

[0053] Specifically, the steps for training a terrain classification network model are as follows:

[0054] (3-1) Collect volume perception data for N types of known terrain, specifically including the following steps:

[0055] (3-1-1) N=10, the known terrain includes: cement road, grassland, stone road, artificial turf, PVC floor, ceramic tile, brick road, permeable bricks for parking lot, plastic running track, and land.

[0056] (3-1-2) First, measure the mutual capacitance between the two electrodes collected by the capacitive sensor array under no-load conditions. Specifically, remove the capacitive sensor array from the bottom of the mobile robot, lay it flat on the ground, keeping the area above the sensors empty, measure 200 mutual capacitance values ​​between the two electrodes, and calculate the average of all values. The mutual capacitance between the two electrodes collected by the capacitive sensor array under no-load conditions is C. b =[c1,c2,…,c 28After controlling the mobile robot to move onto terrain O, the mobile robot's movement modes include: linear movement (moving forward or backward at a certain speed v; in this embodiment, six speeds are included: v1 = 0.3 m / s, v2 = -0.3 m / s, v3 = 0.5 m / s, v4 = -0.5 m / s, v5 = 0.8 m / s, v6 = -0.8 m / s) and curvilinear movement (moving with a combination of constant linear velocity v and angular velocity ω, where v ∈ {0.3 m / s, 0 Ω}). 0.5m / s, 0.8m / s}, ω∈{0.8rad / s, 1.2rad / s}), random motion (the robot generates random linear and angular velocities, where the linear velocity v ranges from (0.3m / s, 1.2m / s) and the angular velocity ω ranges from (0.8rad / s, 1.2rad / s)), continuously collecting inertial and mutual capacitance data. The electrode plate has 8 electrodes, and the mutual capacitance between the two electrodes collected by the capacitance sensor array each time is denoted as C. t =[c 1,t ,c 2,t ,…,c 28,t Each collected inertial data includes the robot's angular velocity ω and acceleration a along the x, y, and z axes. t =[ω x,t ,ω y,t ,ω z,t ,a x,t ,a y,t ,a z,t The sampling frequencies of both the capacitance sensor and the inertial sensor were adjusted to 10 Hz. The measured values ​​of three consecutive inertial data points C1, C2, and C3, and mutual capacitance data I1, I2, and I3 were recorded over a period of time. The capacitance time-series characteristic sequence C was then calculated. m =[C1-C b C2-C b [C2-C1,C3-C2] and inertial time series feature sequence I m = [I1, I2, I2-I1, I3-I2], where the dimensions of the two time-series feature sequences are 112 and 24, respectively. K capacitance and inertial data are collected on terrain O, resulting in k capacitance time series C of terrain O. M,O =[C m,1 C m,2 ,…,C m,k ] and k inertial time series I M,O =[I m,1 ,I m,2 ,…,I m,k The system records the category label of terrain O. In this example, for terrain O, linear motion and curvilinear motion modes are used, and the capacitance time series C obtained for each velocity and velocity combination is recorded. M,O and inertial time series I M,OThe quantity k is 500, and a random motion mode is used to obtain the capacitance time series C. M,O and inertial time series I M,O Since the quantity k is 1500, for terrain O, after collecting data using all motion modes, a total of 7500 sets (500×6×2+1500) capacitance time series C are obtained. M,O and inertial time series I M,O ;

[0057] (3-1-3) For the capacitance time series C M,O and inertial time series I M,O Preprocessing is performed for C M,O and I M,O All elements in the array are normalized using the min-max method;

[0058] (3-1-4) Traverse the 10 terrain types in step (3-1-1) and repeat step (3-1-2) to obtain the ontology-aware dataset D of the terrain. IC =[C M,1 C M,2 ,…,C M,N ,I M,1 ,I M,2 ,…,I M,N ];

[0059] (3-1-5) A multilayer perceptron network is constructed for feature extraction, fusion, and classification of inertial and capacitive data. For 24-dimensional inertial time-series data, a two-layer multilayer perceptron network is used, with 22 and 16 neurons in each layer, respectively. For 112-dimensional capacitive time-series data, a three-layer MLP is used, with 64, 32, and 16 neurons in each layer, respectively. The features extracted from these two types of data are combined into a 32-dimensional vector. Each neuron in each layer contains an activation function. In this example, the ReLU function is used as the activation function. This 32-dimensional vector is passed through a two-layer multilayer perceptron, with 16 and 10 neurons in each layer, respectively, and a softmax function. The likelihood values ​​of 10 terrain labels are calculated, and the label with the highest likelihood value is selected as the final terrain classification result.

[0060] (3-2) Based on the terrain ontology perception dataset D collected in step (3-1-4) IC The corresponding material labels provided in the document utilize a multilayer perceptron network built according to (3-1-5) as a terrain classifier, with the capacitance temporal feature sequence C as the basis. M,O All elements C m,i and inertial time series characteristic sequence I M,O All elements I M,,iThe terrain classification labels, as input to the multilayer perceptron, are used as output to train the multilayer perceptron, resulting in a classifier capable of identifying terrain types. Specifically, the terrain ontology perception dataset D... IC The dataset D is divided into a training set and a validation set, with a ratio of 8:2. IC After the data is randomly shuffled, it is divided into training and validation sets according to the ratio. Training begins with a batch size of 64, a maximum training epoch of 200, and a learning rate of 0.01. Every 10 training epochs, the validation set is used for validation. If the classification accuracy is greater than the existing model, it is saved and overlaid on the existing model (the model is saved directly during the first validation).

[0061] (3-3) After training the classification model according to step (3-2), control the robot to move to one of the 10 known terrains, terrain O, record the measurement values ​​of inertial data and mutual capacitance data three times within a certain period of time, and calculate the corresponding capacitance time-series feature sequence C. m and corresponding inertial time series feature sequence I m Then, a multilayer perceptron classification model is used to classify the terrain, and the terrain classification label of the terrain O to be measured is obtained, thus completing the terrain classification.

[0062] This invention discloses a terrain classification method based on a planar capacitance tomography (ECT) sensor. It designs an eight-electrode planar ECT sensor array and constructs a mobile robot body perception data acquisition system to collect mutual capacitance and inertial data of the terrain. Next, a multilayer perceptron (MPP) classification model is built to extract, fuse, and classify features from the capacitance and inertial data. After training, the MPP model guides the robot to known terrain, and the MPP classifies the terrain. This invention utilizes both inertial and capacitance modal information, combining the interactive vibration information between the robot and the terrain with changes in the terrain's dielectric constant, to automatically identify terrains of various materials, demonstrating strong versatility and practical significance.

[0063] To achieve the above embodiments, such as Figure 5 As shown, this embodiment also provides a terrain classification system 10 based on a flat-panel capacitive tomography sensor, including:

[0064] Sensor array building module 100 is used to build a planar ECT sensor array containing eight electrodes;

[0065] The data acquisition system construction module 200 is used to build a mobile robot body perception data acquisition system based on the flat panel ECT sensor array;

[0066] The classifier model training module 300 is used to acquire the terrain ontology perception dataset using the mobile robot ontology perception data acquisition system, and extract all elements in the capacitance time-series feature sequence and all elements in the inertial time-series feature sequence based on the corresponding material labels provided in the terrain ontology perception dataset. These elements are used as input data for the multilayer perceptron network, and the terrain classification labels are used as output data for network training to obtain a classifier model that can identify different terrain types.

[0067] The terrain classification module 400 is used to control the robot to move to the terrain to be measured, record the measurement values ​​of inertial data and mutual capacitance data three times in a preset time, calculate the real-time capacitance time series feature sequence and the real-time inertial time series feature sequence, and use the trained classifier model to classify the terrain to obtain the terrain classification label of the terrain to be measured, so as to complete the terrain classification.

[0068] This invention discloses a terrain classification system based on a planar capacitance tomography (ECT) sensor. It designs an eight-electrode planar ECT sensor array and constructs a mobile robot body perception data acquisition system to collect mutual capacitance and inertial data of the terrain. Next, a multilayer perceptron (MPP) classification model is built to extract, fuse, and classify features from the capacitance and inertial data. After training, the MPP model guides the robot to known terrain, and the MPP classifies the terrain. This invention utilizes both inertial and capacitance modal information, combining the interactive vibration information between the robot and the terrain with changes in the terrain's dielectric constant, to automatically identify terrains of various materials, demonstrating strong versatility and practical significance.

[0069] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0070] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A terrain classification method based on a planar capacitance tomography sensor, characterized in that, include: Construct a planar ECT sensor array containing eight electrodes; A mobile robot body perception data acquisition system was built based on the aforementioned flat panel ECT sensor array; A terrain ontology perception dataset is acquired using a mobile robot ontology perception data acquisition system. Based on the corresponding material labels provided in the terrain ontology perception dataset, all elements in the capacitance time-series feature sequence and all elements in the inertial time-series feature sequence are extracted and used as input data for a multilayer perceptron network. The terrain classification label is used as output data for network training to obtain a classifier model that can identify different terrain types. The robot is controlled to move to the terrain to be measured, and the measurement values ​​of inertial data and mutual capacitance data are recorded three times consecutively within a preset time. The real-time capacitance time-series feature sequence and the real-time inertial time-series feature sequence are calculated. The trained classifier model is then used to classify the terrain to obtain the terrain classification label of the terrain to be measured, thus completing the terrain classification. Training a multilayer perceptron network includes: (3-1) Collecting a terrain ontology-aware dataset: (3-1-1) Suppose that it contains N Known terrain features; (3-1-2) The mutual capacitance between the two electrodes of the capacitive sensor array under no-load conditions is: C b =[ c 1, c 2,…, c 28 [Measurement and control of the mobile robot's movement to the terrain] O After that, inertial data and mutual capacitance data are continuously collected, and the mutual capacitance between the two electrodes collected by the capacitance sensor array each time is recorded as . C t =[ c 1,t , c 2,t , …, c 28,t Each inertial measurement unit collects inertial data containing information about the robot's movement. x , y , z angular velocity on three axes ω and acceleration a , I t =[ ω x,t , ω y,t , ω z,t , a x,t , a y,t , a z,t Record three consecutive inertial data points within a certain period of time. C 1. C 2. C 3 and mutual capacitance data I 1. I 2. I 3. Measure the value and calculate the capacitance timing characteristic sequence. C m =[ C 1- C b , C 2- C b , C 2- C 1, C 3- C 2] and inertial time series characteristic sequences I m =[ I 1, I 2, I 2- I 1, I 3- I 2], in the terrain O Completed k The acquisition of secondary capacitance and inertial data yields terrain data. O of k Capacitance time series C M,O =[ C m,1 , C m,2 , …, C m,k ]and k An inertial time series I M,O =[ I m,1 , I m,2 , …, I m,k ], and record the terrain. O Category tags; (3-1-3) Capacitance time series C M,O and inertial time series I M,O Preprocessing is performed for C M,O and I M,O All elements in the array are normalized using the min-max method; (3-1-4) Traversal step (3-1-1) N For each terrain type, repeat step (3-1-2) to obtain the ontology-aware dataset of the terrain. D IC =[ C M,1 , C M,2 , …, C M,N , I M,1 , I M,2 , …, I M,N ]; (3-1-5) Construct a multilayer perceptron network. For 24-dimensional inertial time-series data, use a two-layer multilayer perceptron network with 22 and 16 neurons per layer, respectively; for 112-dimensional capacitive time-series data, use a three-layer MLP with 64, 32, and 16 neurons per layer, respectively. Combine the features extracted from these two types of data into a 32-dimensional vector. Pass the vector through a two-layer multilayer perceptron with 16 and 10 neurons per layer, respectively, and a softmax function to calculate the likelihood value of each label. Select the label with the highest likelihood value as the final terrain classification result. (3-2) Based on the terrain ontology perception dataset collected in step (3-1-4) D IC The corresponding material labels provided in the text utilize a multilayer perceptron network built in (3-1-5) as a terrain classifier, with capacitance temporal feature sequences as the basis. C M,O all elements C m,i and inertial time series characteristic sequences I M,O all elements I M,,i As input to the multilayer perceptron, the terrain classification label is used as the output of the multilayer perceptron to train the network and obtain a classifier model that can identify terrain types. (3-3) After the classifier model is trained in step (3-2), control the robot to move to one of the various known terrains. O Record the measurements of inertial data and mutual capacitance data three times consecutively within a certain period, and calculate the corresponding capacitance time-series characteristic sequence. C m and corresponding inertial time series feature sequences I m Then, a classifier model is used to classify the terrain to obtain the terrain to be measured. O Use terrain classification tags to complete terrain classification.

2. The method according to claim 1, characterized in that, Constructing a planar ECT sensor array containing eight electrodes includes: The eight electrodes of the planar ECT sensor array are arranged in a 2 The sensor is arranged in a 4-grid pattern and implemented on a PCB board. The sensor's external dimensions are 80mm. 85mm, FR-4 material is laid under the sensor plate to isolate external electromagnetic interference; the protection electrode separates adjacent sensing units and suppresses capacitive coupling through a uniform ground reference grid.

3. The method according to claim 1, characterized in that, A mobile robot body perception data acquisition system includes: a mobile robot, a multiplexer, a capacitance measurement device, a flat panel ECT sensor array, a microcomputer, and an inertial measurement unit; among which, The inertial measurement unit is used to measure inertial data; The flat-panel ECT sensor array is installed on the bottom of the mobile robot and connected to a multiplexer. A microcomputer is connected to the multiplexer to control the gating logic. A capacitance measuring device is connected to the multiplexer to collect capacitance data and then save it to the microcomputer. A laptop computer is connected to the microcomputer via a wireless local area network to control the movement of the mobile robot.

4. A terrain classification system based on a planar capacitance tomography sensor, used to execute the terrain classification method based on a planar capacitance tomography sensor as described in claim 1, characterized in that, include: Sensor array building module for building a planar ECT sensor array containing eight electrodes; The data acquisition system construction module is used to build a mobile robot body perception data acquisition system based on the flat panel ECT sensor array; The classifier model training module is used to acquire the terrain ontology perception dataset using the mobile robot ontology perception data acquisition system, and extract all elements in the capacitance time-series feature sequence and all elements in the inertial time-series feature sequence based on the corresponding material labels provided in the terrain ontology perception dataset. These elements are used as input data for the multilayer perceptron network, and the terrain classification labels are used as output data for network training to obtain a classifier model that can identify different terrain types. The terrain classification module is used to control the robot to move to the terrain to be measured, record the measurement values ​​of inertial data and mutual capacitance data three times in a preset time, calculate the real-time capacitance time series feature sequence and the real-time inertial time series feature sequence, and use the trained classifier model to classify the terrain to obtain the terrain classification label of the terrain to be measured, so as to complete the terrain classification.

Images