Robot environment perception data processing method, system, device and medium

Abstract

The present application provides a robot environment perception data processing method, system, device and medium. The method obtains environment perception data through each sensor in the environment perception sensor cluster, then integrates the collected environment perception data into an environment perception data set and sends it to a processor. The processor generates a robot operation data matrix based on the received environment perception data set. The processor determines the environment perception result based on the robot operation data matrix, realizes the collection and spatial position correction of multi-sensor data, solves the problem of multi-source heterogeneous data fusion, and after generating the robot operation data matrix, can also efficiently extract environmental features and determine the environment perception result, significantly improving the robot's perception accuracy and real-time performance of complex environments, and providing a reliable data foundation for subsequent navigation, obstacle avoidance and other tasks.

Description

Technical Field

[0001] The present application relates to data processing technology, and in particular to a method, system, device and medium for processing robot environment perception data. Background Art

[0002] With the continuous development of robotics, robots are increasingly being used in various fields, such as intelligent logistics, industrial automation, smart homes, medical assistance, and autonomous driving. In these applications, robots need to be able to accurately perceive their surroundings in order to make correct decisions and actions. Therefore, robot environmental perception technology has become a key research direction in the field of robotics.

[0003] Environmental perception sensors play a crucial role in robotic environmental perception systems. These sensors acquire information from the surrounding environment, such as the shape, position, distance, temperature, and humidity of objects, and convert this information into digital signals for the robot to process. However, due to the complexity and diversity of the robot's operating environment, a single sensor often struggles to obtain comprehensive and accurate environmental information. Therefore, robotic perception systems typically utilize a sensor cluster consisting of multiple sensors to achieve multi-dimensional, all-encompassing perception of the environment.

[0004] However, the data fusion problem caused by multiple environmental perception sensors has become a key factor restricting the accuracy and real-time performance of robots' environmental perception. Differences in operating principles, measurement ranges, accuracy, and noise characteristics among different sensors lead to inconsistencies and heterogeneity in the collected environmental perception data. These differences not only increase the complexity of data processing but also may introduce errors, affecting the robot's accurate perception of the environment.

[0005] Traditional data fusion methods often struggle to effectively address these issues. They either ignore the differences between sensors and directly fuse the raw data, resulting in inaccurate results; or they employ complex algorithms to preprocess and correct the data, which is computationally intensive and lacks real-time performance. Therefore, a more efficient and accurate data processing method is needed to achieve the fusion and spatial position correction of multi-source heterogeneous data. Summary of the Invention

[0006] The present application provides a robot environment perception data processing method, system, device and medium for realizing the fusion of multi-source heterogeneous environment perception data.

[0007] In a first aspect, the present application provides a robot environment perception data processing method, characterized in that it is applied to a robot perception system, wherein the robot perception system includes a processor and an environment perception sensor cluster composed of multiple environment perception sensors, and each environment perception sensor in the environment perception sensor cluster is communicatively connected to the processor; the method includes:

[0008] Acquire environmental perception data through each environmental perception sensor in the environmental perception sensor cluster to generate an environmental perception data set, and send the environmental perception data set to the processor, wherein the environmental perception data includes a spatial position;

[0009] The processor generates a robot operation data matrix according to the environment perception data set, wherein each robot operation feature vector in the robot operation data matrix includes an environment perception data sequence at the same correction space position;

[0010] The processor determines an environment perception result according to the robot operation data matrix.

[0011] In the above scheme, environmental perception data is obtained through each sensor in the environmental perception sensor cluster, and then the collected environmental perception data is integrated into an environmental perception data set and sent to the processor. The processor generates a robot operation data matrix based on the received environmental perception data set. The processor determines the environmental perception result based on the robot operation data matrix, realizes the collection and spatial position correction of multi-sensor data, and solves the problem of multi-source heterogeneous data fusion. After generating the robot operation data matrix, it can also efficiently extract environmental features and determine environmental perception results, significantly improving the robot's perception accuracy and real-time performance of complex environments, and providing a reliable data basis for subsequent navigation, obstacle avoidance and other tasks.

[0012] Optionally, each environment perception sensor in the environment perception sensor cluster is calibrated based on a different reference coordinate system.

[0013] In the above solution, the actual calibration of the environmental perception sensors on the robot is based on different reference coordinate systems. This process is mainly to reduce data inconsistencies caused by factors such as the sensor installation position and orientation. However, precisely because each environmental perception sensor is calibrated based on a different reference coordinate system, the environmental perception data obtained by each environmental perception sensor cannot be directly fused. In the above solution, the robot operation data matrix solves the problem of multi-source heterogeneous data fusion, enabling more accurate judgments by fusing multi-source data. For example, in the autonomous driving scenario of a logistics robot, the data fusion of radar and visual sensors can enable the robot to more accurately identify obstacles such as road markings, safety signs, and surrounding cargo, thereby improving the safety and reliability of autonomous navigation.

[0014] Optionally, each row vector in the robot operation data matrix corresponds to an environmental perception data sequence at the same corrected spatial position, and each column vector includes environmental perception data of the same environmental perception sensor in the environmental perception sensor cluster at different corrected spatial positions.

[0015] In the above scheme, the structure of the robot's operational data matrix helps enhance the robot's data analysis capabilities. Because the data in the matrix is organized according to spatial location and sensor type, the robot can easily perform in-depth data analysis using matrix operations, statistical analysis, and other methods. For example, the robot can evaluate the stability and reliability of environmental perception data by calculating statistics such as the mean and variance of specific rows or columns in the matrix. Alternatively, it can extract characteristic information from the data through matrix transformations, providing a richer basis for subsequent decision-making. Furthermore, the structural characteristics of the robot's operational data matrix enable the robot to make more accurate decisions. During autonomous navigation, the robot can quickly determine the current state and changing trends of the environment based on the data in the matrix, thereby developing a reasonable driving route and obstacle avoidance strategy. For example, when the robot detects an obstacle ahead, it can determine the obstacle's location, size, and motion state by analyzing the data at the corresponding position in the matrix, and adjust its driving direction and speed accordingly to avoid a collision. Furthermore, the structure of the robot's operational data matrix is highly scalable. As the number of sensors in the robot's perception system increases or new sensor types are introduced, the data structure can be easily expanded by simply adding corresponding columns to the matrix. This scalability enables the robot to flexibly adapt to different application scenarios and task requirements.

[0016] Optionally, the processor uses a preset spatial position correction model and generates a robot operation data matrix according to the environmental perception data set, including:

[0017] The processor uses a preset spatial position correction model and generates the robot operation data matrix according to the environmental perception data set and the calibration parameters of each environmental perception sensor in the environmental perception sensor cluster.

[0018] In the above solution, by utilizing a preset spatial position correction model, the processor can accurately correct the data acquired by each sensor, eliminating data deviations caused by factors such as sensor installation position and orientation. This significantly improves data accuracy and provides a reliable foundation for subsequent data analysis and processing. The structural characteristics of the robot's operational data matrix ensure that data from all sensors are compared and analyzed within the same reference coordinate system. This consistency not only simplifies the data analysis process but also improves its accuracy and reliability. For example, during autonomous navigation, the robot can accurately determine whether environmental information acquired by different sensors is consistent, thereby avoiding decision-making errors caused by inconsistent data. The orderly organization of the robot's operational data matrix makes the data analysis process more efficient. The processor can conveniently use matrix operations, statistical analysis, and other methods to conduct in-depth data analysis and extract valuable information. This efficient data analysis mechanism provides strong support for the robot's real-time environmental perception and decision-making. The robot's operational data matrix generated based on the preset spatial position correction model is highly robust. If a sensor fails or data anomalies occur, the robot can rely on other accurately calibrated sensors to continue perceiving the surrounding environment and make decisions based on the remaining data in the matrix. This redundant design significantly improves the stability and reliability of the system.

[0019] Optionally, the processor determines an environmental perception result according to the robot operation data matrix, including:

[0020] The processor determines the environmental feature confidence level of the target sensing position based on the robot operation data matrix;

[0021] The processor determines the target environmental feature state of the target perception position according to the environmental feature confidence and a preset environmental feature confidence threshold.

[0022] In this solution, by comprehensively considering data from multiple sensors, the processor can more accurately calculate the confidence level of environmental features at the target location, thereby improving the accuracy of environmental perception. Based on accurate environmental perception results, the robot can make decisions more quickly. For example, during autonomous navigation, the robot can use environmental perception results to plan a safe path to avoid obstacles. During obstacle avoidance tasks, the robot can promptly detect obstacles ahead and take appropriate avoidance measures.

[0023] Optionally, after determining the environmental feature confidence level of the target perception position according to the robot operation data matrix, the method further includes:

[0024] The processor generates an environment perception distribution grid according to the confidence level of the environment characteristics of each target perception position, and the environment perception distribution grid is used to perform grid processing on the perception projection plane of the target perception space along the perception direction of the environment perception sensor.

[0025] In the above solution, the environmental perception distribution grid intuitively displays the state of environmental features within the target perception space. Using the confidence values within the grid, the robot can clearly identify key information such as obstacles and traversable areas, providing strong support for subsequent decision-making. Because the environmental perception distribution grid comprehensively grids the target perception space, the robot can more comprehensively perceive its surroundings. Furthermore, by comprehensively considering data from multiple sensors and their varying reliability, the processor can more accurately calculate the confidence level of environmental features for each grid, thereby improving the accuracy of environmental perception. Based on the environmental perception distribution grid, the robot can make decisions more quickly. For example, during autonomous navigation, the robot can plan a safe path to avoid obstacles based on the confidence values within the grid. During obstacle avoidance tasks, the robot can promptly detect obstacles ahead and take appropriate avoidance measures. Furthermore, the environmental perception distribution grid provides important reference information for tasks such as path planning and object detection. By generating the environmental perception distribution grid, the robot can tolerate noise and uncertainty in sensor data to a certain extent. Even if the data from a certain sensor is abnormal or deviant, the processor can still rely on the data from other sensors to generate a relatively accurate environmental perception distribution grid, thereby improving the robustness and reliability of the system.

[0026] Optionally, after the processor generates the environment perception distribution grid according to the confidence level of the environment feature of each target perception position, the method further includes:

[0027] The processor determines a safety boundary according to the environmental awareness distribution grid, wherein the safety boundary is a grid in the environmental awareness distribution grid where only one adjacent grid corresponds to an environmental feature confidence level greater than a preset environmental feature confidence level threshold.

[0028] In the above solution, by accurately determining the safety margin, the robot can avoid potentially dangerous areas during path planning and choose a safer route. This helps improve the robot's driving safety and reduce the probability of accidents. Specifically, the determination of the safety margin provides a crucial basis for the robot's obstacle avoidance tasks. Based on the position and shape of the safety margin, the robot can accurately determine the location and range of obstacles, enabling it to take more effective avoidance measures. By reducing the probability of collisions and accidents, the robot can complete its tasks more stably and efficiently, thereby improving the overall performance and reliability of the system.

[0029] In a second aspect, the present application provides a robot perception system, comprising: a processor and an environment perception sensor cluster consisting of a plurality of environment perception sensors, wherein each environment perception sensor in the environment perception sensor cluster is communicatively connected to the processor;

[0030] Acquire environmental perception data through each environmental perception sensor in the environmental perception sensor cluster to generate an environmental perception data set, and send the environmental perception data set to the processor, wherein the environmental perception data includes a spatial position;

[0031] The processor generates a robot operation data matrix according to the environment perception data set, wherein each robot operation feature vector in the robot operation data matrix includes an environment perception data sequence at the same correction space position;

[0032] The processor maps the robot operation data matrix into a preset feature grid map to output an environmental perception processing result.

[0033] Optionally, each environment perception sensor in the environment perception sensor cluster is calibrated based on a different reference coordinate system.

[0034] Optionally, each row vector in the robot operation data matrix corresponds to an environmental perception data sequence at the same corrected spatial position, and each column vector includes environmental perception data of the same environmental perception sensor in the environmental perception sensor cluster at different corrected spatial positions.

[0035] Optionally, the processor generates a robot operation data matrix according to the environmental perception data set, including:

[0036] The processor uses a preset spatial position correction model and generates the robot operation data matrix according to the environmental perception data set and the calibration parameters of each environmental perception sensor in the environmental perception sensor cluster.

[0037] Optionally, the processor determines an environmental perception result according to the robot operation data matrix, including:

[0038] The processor determines the environmental feature confidence level of the target sensing position based on the robot operation data matrix;

[0039] The processor determines the target environmental feature state of the target perception position according to the environmental feature confidence and a preset environmental feature confidence threshold.

[0040] Optionally, after determining the environmental feature confidence level of the target perception position according to the robot operation data matrix, the method further includes:

[0041] The processor generates an environment perception distribution grid according to the confidence level of the environment characteristics of each target perception position, and the environment perception distribution grid is used to perform grid processing on the perception projection plane of the target perception space along the perception direction of the environment perception sensor.

[0042] Optionally, after the processor generates the environment perception distribution grid according to the confidence level of the environment feature of each target perception position, the method further includes:

[0043] The processor determines a safety boundary according to the environmental awareness distribution grid, wherein the safety boundary is a grid in the environmental awareness distribution grid where only one adjacent grid corresponds to an environmental feature confidence level greater than a preset environmental feature confidence level threshold.

[0044] In a third aspect, the present application provides an electronic device, comprising:

[0045] processor; and,

[0046] a memory for storing executable instructions of the processor;

[0047] The processor is configured to perform any possible method described in the first aspect by executing the executable instructions.

[0048] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement any possible method described in the first aspect.

[0049] The robot environment perception data processing method, system, device and medium provided in the present application obtain environment perception data through each sensor in the environment perception sensor cluster, and then integrate the collected environment perception data into an environment perception data set and send it to the processor. The processor generates a robot operation data matrix based on the received environment perception data set. The processor determines the environment perception result based on the robot operation data matrix, realizes the collection and spatial position correction of multi-sensor data, solves the problem of multi-source heterogeneous data fusion, and after generating the robot operation data matrix, can also efficiently extract environmental features and determine environmental perception results, significantly improving the robot's perception accuracy and real-time performance of complex environments, and providing a reliable data basis for subsequent navigation, obstacle avoidance and other tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0051] Figure 1 This is a flow chart of a method for processing robot environment perception data according to an example embodiment of the present application;

[0052] Figure 2 is a flow chart of a method for processing robot environment perception data according to another exemplary embodiment of the present application;

[0053] Figure 3 is a schematic structural diagram of a robot perception system according to an exemplary embodiment of the present application;

[0054] Figure 4 It is a structural diagram of an electronic device according to an exemplary embodiment of the present application.

[0055] The above drawings illustrate specific embodiments of the present application, which will be described in more detail below. These drawings and the textual description are not intended to limit the scope of the present application in any way, but rather to illustrate the concepts of the present application to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0056] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0057] To address the above-mentioned issues, the embodiments provided in this application address the problem of multi-source heterogeneous data fusion by integrating data from multiple environmental perception sensors and performing corrections using a preset spatial position correction model. This approach not only improves data accuracy and consistency but also provides a reliable data foundation for subsequent environmental perception and decision-making. The row vectors and column vectors in the matrix correspond to the environmental perception data sequence at the same correction spatial position and the data from the same sensor at different correction spatial positions, respectively. This orderly organized data structure enhances the robot's data analysis capabilities. The processor calculates the environmental feature confidence level at the target perception location based on the robot's operational data matrix and determines the target environmental feature status based on a preset environmental feature confidence threshold. Subsequently, the processor generates an environmental perception distribution grid based on the environmental feature confidence level at each target perception location, visually displaying the environmental feature status within the target perception space. Based on this environmental perception distribution grid, the processor further determines a safety boundary, providing an important basis for the robot's path planning and obstacle avoidance tasks. By accurately determining the safety boundary, the robot can avoid potentially dangerous areas during path planning and select a safer driving path.

[0058] The robot environment perception data processing method provided in the embodiment of the present application can be applied to a robot perception system, which includes a processor and an environment perception sensor cluster composed of multiple environment perception sensors, and each environment perception sensor in the environment perception sensor cluster is communicatively connected to the processor.

[0059] Specifically, the cluster includes various types of environmental perception sensors, including radar, visual, and infrared sensors. These sensors are responsible for acquiring environmental information in different dimensions, such as distance, shape, and temperature. Each sensor is calibrated based on its unique reference coordinate system to ensure data accuracy and reliability. As the core component of the system, the processor is responsible for receiving, processing, and analyzing data from the environmental perception sensor cluster. The processor possesses powerful computing and storage capabilities, enabling it to efficiently process large amounts of heterogeneous data and generate a data matrix for the robot's operations.

[0060] Each sensor in the environmental perception sensor cluster begins operating, acquiring its own environmental perception data, including information such as spatial location. This data is integrated into an environmental perception data set and sent to the processor for further processing. After receiving the environmental perception data set, the processor corrects the data using a preset spatial position correction model. This model takes into account factors such as the sensor's installation location and orientation, as well as calibration parameters, to ensure data consistency. The corrected data is organized into a robot operation data matrix. Each element in the matrix represents a sequence of environmental perception data at the same corrected spatial location. This structure makes the data analysis process more efficient and facilitates subsequent environmental feature extraction and decision-making. Based on the robot operation data matrix, the processor calculates the environmental feature confidence level for the target perception location. This confidence level reflects the likelihood that a specific environmental feature (such as an obstacle) is present at the target location.

[0061] Figure 1 FIG. 1 is a flow chart of a method for processing robot environment perception data according to an exemplary embodiment of the present application. Figure 1 As shown, the robot environment perception data processing method provided in this embodiment includes:

[0062] S101: Acquire environmental perception data through each environmental perception sensor in an environmental perception sensor cluster to generate an environmental perception data set.

[0063] In this step, environmental perception data is acquired by each environmental perception sensor in the environmental perception sensor cluster to generate an environmental perception data set, and the environmental perception data set is sent to the processor, wherein the environmental perception data includes a spatial position.

[0064] Specifically, environmental perception data is collected by the individual sensors in the environmental perception sensor cluster. These sensors may include radar, cameras, lidar, and other types, capable of capturing information about the surrounding environment from various angles and dimensions. The collected environmental perception data not only includes spatial location information but also may encompass various attributes such as an object's shape, size, and speed. After data collection is complete, each sensor integrates the generated environmental perception data into a single environmental perception data set. This set, encompassing data from various sensors, provides the foundation for subsequent data fusion and processing. This data set is then sent to a processor for further processing.

[0065] In addition, since the environmental perception sensors on the robot are actually calibrated based on different reference coordinate systems, this process is mainly to reduce data inconsistency caused by factors such as the sensor installation position and orientation. However, precisely because each environmental perception sensor is calibrated based on a different reference coordinate system, the environmental perception data obtained by each environmental perception sensor cannot be directly fused. Optionally, the above-mentioned robot operation data matrix solves the problem of multi-source heterogeneous data fusion, enabling more accurate judgments to be made by fusing multi-source data. For example, in the autonomous driving scenario of a logistics robot, the data fusion of radar and visual sensors can enable the robot to more accurately identify obstacles such as road markings, safety signs, and surrounding goods, thereby improving the safety and reliability of autonomous navigation.

[0066] Specifically, a cluster of environmental perception sensors may include various sensor types, such as radar, cameras, and lidar. In practice, these sensors are calibrated based on different reference coordinate systems. This calibration method is primarily intended to reduce data inconsistencies caused by factors such as sensor installation position and orientation. Each sensor has its own unique reference coordinate system, which defines the spatial position and orientation of the data it captures. During the calibration process, each sensor's reference coordinate system must be associated with the robot's execution coordinate system to ensure that all data can be compared and analyzed within the same reference framework.

[0067] First, a suitable calibration environment must be selected. This environment should provide ample space for the robot to move freely and contain a variety of obstacles to simulate the complexities of a real-world work environment. Each sensor in the environmental perception sensor cluster must be installed on the robot, ensuring that their position and orientation meet the design requirements. During installation, the position and orientation of each sensor must be recorded for subsequent data processing. In the calibration environment, the robot is controlled to move along a predetermined trajectory while simultaneously collecting environmental perception data from each sensor. This data will be used for subsequent calibration calculations.

[0068] Using the collected calibration data set, the least squares method is used for iterative calculation to determine the transformation matrix for each sensor. This transformation matrix is used to convert the sensor's reference coordinate system into the robot's execution coordinate system. During the calculation process, factors such as the amount of data in the calibration environment perception data set or the calibration robot operation data set, as well as the noise covariance matrix of the environment perception sensor, need to be considered. These factors have a significant impact on the accuracy and stability of the calculation results.

[0069] After completing sensor calibration, the resulting transformation matrix and weight values can be applied to subsequent processing of the robot's environmental perception data. When processing environmental perception data sets, these parameters can be used to correct and fuse the sensor data to generate a more accurate robot operation data matrix. Furthermore, the calibration results can be used to optimize the robot's path planning and obstacle avoidance strategies. By more accurately perceiving its surroundings, the robot can make decisions more quickly and avoid collisions with obstacles.

[0070] S102. The processor generates a robot operation data matrix according to the environment perception data set.

[0071] In this step, the processor generates a robot operation data matrix based on the environment perception data set, and each robot operation feature vector in the robot operation data matrix includes an environment perception data sequence at the same correction space position.

[0072] Specifically, after receiving the environmental perception data set, the processor begins to generate the robot's operational data matrix. The key to this step is to use a preset spatial position correction model to correct the data to eliminate data deviations caused by factors such as sensor installation position and orientation.

[0073] Specifically, the processor generates a robot operation data matrix based on the environmental perception data set and the calibration parameters of each sensor. In this matrix, each row vector corresponds to a sequence of environmental perception data at the same calibration spatial position, while each column vector contains environmental perception data from the same sensor at different calibration spatial positions. This matrix structure design helps enhance the robot's data analysis capabilities. Through matrix operations, statistical analysis, and other methods, the robot can easily conduct in-depth analysis of the data, extracting valuable information and providing strong support for subsequent decision-making.

[0074] Optionally, each row vector in the robot operation data matrix corresponds to an environmental perception data sequence at the same correction space position, and each column vector includes environmental perception data of the same environmental perception sensor in the environmental perception sensor cluster at different correction space positions.

[0075] The structure of the robot's operational data matrix helps enhance the robot's data analysis capabilities. Because the data in the matrix is organized according to spatial location and sensor type, the robot can easily perform in-depth data analysis using matrix operations and statistical analysis. For example, the robot can evaluate the stability and reliability of environmental perception data by calculating statistics such as the mean and variance of specific rows or columns in the matrix. Alternatively, it can extract characteristic information from the data through matrix transformations, providing a richer basis for subsequent decision-making. Furthermore, the structural characteristics of the robot's operational data matrix enable the robot to make more accurate decisions. During autonomous navigation, the robot can quickly determine the current state and changing trends of the environment based on the data in the matrix, thereby developing a reasonable driving route and obstacle avoidance strategy. For example, when the robot detects an obstacle ahead, it can determine the obstacle's location, size, and motion state by analyzing the data at the corresponding position in the matrix, and adjust its driving direction and speed accordingly to avoid a collision. Furthermore, the structure of the robot's operational data matrix is highly scalable. As the number of sensors in the robot's perception system increases or new sensor types are introduced, the data structure can be easily expanded by simply adding corresponding columns to the matrix. This scalability enables the robot to flexibly adapt to different application scenarios and task requirements.

[0076] S103. The processor determines the environmental perception result according to the robot operation data matrix.

[0077] Specifically, after generating the robot's operational data matrix, the processor begins determining the environmental perception results based on this matrix. This step may involve multiple computational processes, including calculating the environmental feature confidence level and determining the target environmental feature state. First, the processor determines the environmental feature confidence level at the target perception location based on the robot's operational data matrix. This confidence level reflects the robot's level of confidence in the current environmental state and serves as an important basis for subsequent decision-making. The processor then determines the target environmental feature state at the target perception location based on the environmental feature confidence level and a preset environmental feature confidence threshold. If the confidence level is greater than or equal to the threshold, the target object (such as an obstacle) is considered to be present at the target perception location. If the confidence level is less than the threshold, the target object (such as an obstacle) is considered to be present at the target perception location.

[0078] In this embodiment, environmental perception data is acquired through each sensor in the environmental perception sensor cluster, and then the collected environmental perception data is integrated into an environmental perception data set and sent to the processor. The processor generates a robot operation data matrix based on the received environmental perception data set. The processor determines the environmental perception result based on the robot operation data matrix, thereby realizing the collection and spatial position correction of multi-sensor data and solving the problem of multi-source heterogeneous data fusion. After generating the robot operation data matrix, it can also efficiently extract environmental features and determine environmental perception results, significantly improving the robot's perception accuracy and real-time performance of complex environments, and providing a reliable data basis for subsequent navigation, obstacle avoidance and other tasks.

[0079] Figure 2 FIG. 1 is a flow chart of a method for processing robot environment perception data according to another exemplary embodiment of the present application. Figure 2 As shown, the robot environment perception data processing method provided in this embodiment includes:

[0080] S201: Acquire environmental perception data through each environmental perception sensor in an environmental perception sensor cluster to generate an environmental perception data set.

[0081] In this step, environmental perception data is acquired by each environmental perception sensor in the environmental perception sensor cluster to generate an environmental perception data set, and the environmental perception data set is sent to the processor, wherein the environmental perception data includes a spatial position.

[0082] Specifically, environmental perception data is collected by the individual sensors in the environmental perception sensor cluster. These sensors may include radar, cameras, lidar, and other types, capable of capturing information about the surrounding environment from various angles and dimensions. The collected environmental perception data not only includes spatial location information but also may encompass various attributes such as an object's shape, size, and speed. After data collection is complete, each sensor integrates the generated environmental perception data into a single environmental perception data set. This set, encompassing data from various sensors, provides the foundation for subsequent data fusion and processing. This data set is then sent to a processor for further processing.

[0083] S202. The processor generates a robot operation data matrix according to the environment perception data set.

[0084] In this step, the processor generates a robot operation data matrix based on the environment perception data set, and each robot operation feature vector in the robot operation data matrix includes an environment perception data sequence at the same correction space position.

[0085] It's worth noting that in robotic environmental perception systems, the use of multiple different types of environmental perception sensors (such as radar, visual sensors, and infrared sensors)—based on varying operating principles, measurement ranges, accuracy, and noise characteristics—results in inconsistent and heterogeneous collected environmental perception data. Traditional data fusion methods often struggle to effectively address these issues. They either ignore these differences and directly fuse the data, resulting in inaccurate results, or employ complex preprocessing algorithms that are computationally intensive and lack real-time performance. Furthermore, because the actual calibration of environmental perception sensors on robots is based on different reference coordinate systems, the collected environmental perception data exhibits spatial deviations. If these deviations are not corrected, they can severely impact subsequent data analysis and decision-making. Traditional data fusion and correction methods often employ complex algorithms for data preprocessing and correction, which not only increases data processing complexity but also compromises the system's real-time performance. Real-time performance is crucial in robotic real-time environmental perception and decision-making scenarios. In complex and changing environments, sensors can be affected by various interference factors, leading to anomalies or deviations in the collected data. Traditional data processing methods often struggle to effectively address these interferences, resulting in insufficient robustness and stability.

[0086] In one possible implementation, the processor generates a robot operation data matrix using Formula 1 based on the environmental perception data set and calibration parameters of each environmental perception sensor in the environmental perception sensor cluster, where Formula 1 is:

[0087]

[0088] in, Run the data matrix for the robot The first position under the correction space Robot operation data; The first in the environmental perception sensor cluster The first position under the correction space Environmental perception data from environmental perception sensors, is the transformation matrix between the reference coordinate system of the environmental perception sensor and the execution coordinate system of the robot, The first in the environmental perception sensor cluster The preset installation position offset of each environmental perception sensor.

[0089] In this solution, by generating a robot operation data matrix, the processor can efficiently organize and manage data from different sensors. Each row in the matrix represents a sequence of environmental perception data at the same calibrated spatial position, while each column contains data from the same sensor at different calibrated spatial positions. This structured data organization enables the processor to easily perform in-depth data analysis using methods such as matrix operations and statistical analysis, thereby extracting valuable information. Furthermore, generating the robot operation data matrix simplifies the data analysis process, improving the accuracy and efficiency of data processing.

[0090] Specifically, by applying Equation 1, the processor can transform the data from each sensor in the environmental perception sensor cluster from its respective reference coordinate system to the robot's execution coordinate system. The transformation matrix between the environmental perception sensor's reference coordinate system and the robot's execution coordinate system plays a key role in this process, ensuring the accuracy and consistency of data from different sensors within the unified coordinate system. The introduction of a preset installation position offset for the environmental perception sensors further improves data accuracy. In practical applications, sensors may experience some positional offset due to installation errors or mechanical wear. By compensating for these offsets, the processor can generate a robot operation data matrix that more closely reflects actual conditions. Furthermore, this solution not only considers data correction for individual sensors but also achieves multi-sensor data fusion through a unified processing framework. This fusion approach fully leverages the strengths of different sensors, such as the camera's image recognition capabilities and the lidar's high-precision ranging capabilities, thereby enhancing the robot's perception of complex environments. Precisely corrected and fused environmental perception data provides a more reliable foundation for the robot's target detection and tracking. The processor can more accurately identify target objects in the environment and track their position and dynamic changes in real time, providing strong support for subsequent decision-making and planning. Based on accurate environmental perception data, robots can more efficiently plan their paths. The processor can evaluate the feasibility and safety of different paths in real time and select the optimal path to avoid obstacles and hazardous areas, thereby improving the robot's operational efficiency and safety. By accurately perceiving its surroundings, the robot can more promptly detect potential obstacles and take appropriate avoidance measures. This capability is crucial for autonomous navigation and operation in complex environments, effectively preventing collisions and damage.

[0091] It's worth noting that in robotic environmental perception systems, transforming sensor data from different reference coordinate systems into a unified robot execution coordinate system requires calculating a transformation matrix. However, traditional transformation matrix calculation methods often suffer from insufficient accuracy and robustness. For example, directly applying a simple linear transformation may not accurately describe the complex spatial relationship between the sensor and the robot, especially in the presence of noise and interference, resulting in significant deviations in the calculated results. In complex robotic operating environments, sensor data is often affected by various noise and interference factors, such as electromagnetic interference, temperature fluctuations, and mechanical vibration. Traditional transformation matrix calculation methods are often overly sensitive to noise and interference, resulting in unstable calculation results, which in turn affects subsequent data fusion and decision-making. Furthermore, traditional transformation matrix calculation methods often use complex mathematical models and algorithms, resulting in high computational complexity and poor real-time performance. In scenarios involving real-time robotic environmental perception and decision-making, this can impact the overall system performance and response time. Furthermore, traditional transformation matrix calculation methods often lack sufficient flexibility and adaptability, making them difficult to adapt to different sensor types and operating environments. For example, when a robot changes sensors or enters a new operating environment, the calculation method and parameters may need to be redesigned.

[0092] Therefore, before the processor uses Formula 1 to generate the robot operation data matrix based on the environmental perception data set and the calibration parameters of each environmental perception sensor in the environmental perception sensor cluster, it can also use Formula 2 to determine the transformation matrix based on the calibrated environmental perception data set and the corresponding calibrated robot operation data set. , Formula 2 is:

[0093]

[0094] in, is the transformation matrix iterated by the least squares method, is the preset transformation Lie group in three-dimensional space, To calibrate the amount of data in the environment perception data set or the robot operation data set, To calibrate the robot running data set Calibrated robot operation data, To calibrate the first Calibrated environmental perception data, is the noise covariance matrix of the environmental perception sensor, To calibrate the first The weight value corresponding to the calibrated environmental perception data.

[0095] In the above scheme, Equation 2 uses the least squares method to iteratively find the optimal transformation matrix. This method ensures the accuracy and reliability of the transformation matrix by minimizing the sum of squared errors between the calibrated environmental perception data and the transformed data. Through iterative optimization, the system can gradually approximate the true transformation relationship, thereby improving the accuracy of data correction. Furthermore, the transformation matrix is confined to a predefined Lie group of transformations in three-dimensional space. This predefined Lie group encompasses all possible rigid body transformations, including rotations and translations, ensuring the comprehensiveness and applicability of the transformation matrix. By constraining the transformation matrix to this predefined Lie group, the system can more accurately describe the complex transformation relationship between the environmental perception sensor and the robot's execution coordinate system. Furthermore, Equation 2 incorporates the noise covariance matrix of the environmental perception sensor. This matrix reflects the noise characteristics and uncertainty of the sensor data. By accounting for the impact of noise, the system can better adapt to the actual environment during the data correction process, improving the data's robustness and anti-interference capabilities. Furthermore, by appropriately assigning weights in Equation 2, the system can perform targeted corrections based on the importance and reliability of different data, further improving the accuracy and effectiveness of data correction.

[0096] Furthermore, before using Equation 2 to determine the transformation matrix, the calibration environment perception data set and the calibration robot operation data set can also be preprocessed. This includes steps such as data cleaning, format conversion, and outlier detection to ensure the accuracy and consistency of the calibration data. Through preprocessing, the system can reduce errors and uncertainties in the subsequent data correction process and improve data processing efficiency. Furthermore, by determining the transformation matrix in advance, the system can directly apply this matrix when subsequently generating the robot operation data matrix using Equation 1, without the need for complex iterative calculations. This not only reduces the computational burden but also improves the system's real-time performance and response speed.

[0097] Furthermore, to set the noise covariance matrix for the aforementioned environmental perception sensor, a set of calibration data can be collected, including the sensor's measured values under different conditions and the corresponding true values (or high-precision reference values). This data should cover the sensor's operating range and the various environmental conditions it may encounter. For each calibration data point, the difference between the sensor's measured value and the true value is calculated to obtain a noise vector. This difference reflects the sensor's noise level at that data point.

[0098] Next, we calculate the covariance between all noise vectors. Covariance measures how much two variables (in this case, the noise at different data points) vary together. If the covariance between two noise vectors is positive, they tend to vary together; if it is negative, they tend to vary in opposite directions; and if it is zero, there is no linear correlation between them.

[0099] The covariance values between all noise vectors are organized into a matrix, the noise covariance matrix. The diagonal elements of this matrix represent the variance of each noise vector (i.e., their respective degree of fluctuation), while the off-diagonal elements represent the covariance between different noise vectors. The quality of the calibration data has a significant impact on the accuracy of the noise covariance matrix. Therefore, the accuracy, completeness, and representativeness of the calibration data should be ensured. The larger the number of samples, the more accurate the noise covariance matrix estimate. However, in practical applications, resource or time constraints may necessitate a trade-off between sample size and accuracy. Different types of environmental perception sensors have different noise characteristics. Therefore, when determining the noise covariance matrix, the sensor type, operating principle, and performance indicators should be fully considered. Furthermore, environmental conditions have a significant impact on sensor noise. Therefore, when collecting calibration data, it is important to cover as many environmental conditions as possible that the sensor may encounter to improve the applicability and accuracy of the noise covariance matrix.

[0100] Furthermore, in robotic environmental perception systems, because the data collected by different environmental perception sensors vary in quality, noise level, and information volume, traditional data processing methods often ignore these differences and treat all data equally. This simple processing method cannot fully utilize the advantages of high-quality data, nor can it effectively suppress the interference that low-quality data may cause, thereby affecting the accuracy and reliability of environmental perception. In existing data processing methods, weight allocation strategies are often based on pre-set methods and cannot be adjusted according to actual application scenarios and needs. As a result, in certain specific application scenarios, the weight allocation may not meet actual needs, thereby affecting the accuracy and reliability of environmental perception.

[0101] Therefore, using Formula 2, the transformation matrix is determined based on the calibration environment perception data set and the corresponding calibration robot operation data set. Previously, it also included:

[0102] Using formula 3, and according to the calibration environment perception data set, the weight value corresponding to each calibration environment perception data in the calibration environment perception data set is determined. Formula 3 is:

[0103]

[0104] in, is the trace of the noise covariance matrix, To calibrate the first The information entropy of the calibrated environment perception data, To calibrate the first The information entropy of the calibrated environment perception data, is the preset adjustment factor, .

[0105] It's worth noting that Equation 3 assigns appropriate weights to each calibrated environmental perception data point by combining the trace of the noise covariance matrix and the information entropy of the calibrated environmental perception data. The trace of the noise covariance matrix reflects the overall noise level of the sensor data, while the information entropy measures the uncertainty and information content of the data. By comprehensively considering these two factors, the system can more accurately assess the quality and reliability of each data point. During the data correction process, the quality and reliability of different data points often vary. By assigning higher weights to high-quality, low-noise data points, the system can more accurately utilize these data points to estimate the transformation matrix, thereby improving the accuracy of data correction.

[0106] Furthermore, the preset adjustment factors in Equation 3 allow users to flexibly adjust the weight allocation strategy based on actual needs. By adjusting the preset adjustment factors, the system can balance the influence of the trace of the noise covariance matrix and the information entropy in weight allocation, further optimizing the weight allocation effect. In different application scenarios, the noise characteristics and information content of sensor data may vary. By adjusting the preset adjustment factors, the system can adapt to different application scenarios and assign more appropriate weights to different data points, thereby improving the accuracy and robustness of environmental perception. By assigning higher weights to high-quality data points, the system can more stably handle noise and uncertainty in sensor data. This helps reduce system errors and failures caused by data quality issues and improves system stability and reliability.

[0107] The trace of the noise covariance matrix is the sum of the matrix's diagonal elements, that is, the sum of all the matrix's eigenvalues. The trace of the noise covariance matrix represents a measure of the overall noise level of the sensor data. Alternatively, the trace of the noise covariance matrix can be calculated for each diagonal element and then summed up to obtain the trace.

[0108] Information entropy, on the other hand, is used to measure the uncertainty of data. If the calibration environment perception data is continuous, it is usually necessary to use a probability density function to estimate it. However, in practical applications, in order to simplify the calculation, the continuous data can also be discretized before calculating the information entropy. For robot environment perception data, since the data is usually multidimensional (such as position, velocity, acceleration, etc.), the uncertainty in all dimensions needs to be considered when calculating the information entropy. This usually involves estimating the joint probability distribution of multidimensional data and then calculating the information entropy based on this distribution. In actual operation, the calibration environment perception data can be divided into several discrete intervals or categories, the number of data points in each interval or category can be counted, and the corresponding probability distribution can be calculated. Then, based on the probability distribution, the information entropy can be calculated using the above information entropy formula.

[0109] S203. The processor determines the environmental feature confidence level of the target sensing position based on the robot operation data matrix.

[0110] Specifically, the robot's operational data matrix forms the foundation for the processor's environmental perception. This matrix contains environmental data collected by various sensors (such as visual sensors and lidar) during the robot's operation, as well as the robot's own motion state information (such as position, velocity, and acceleration). This data is organized into a multidimensional matrix, where each row represents sampled data at a specific point in time, and each column represents a different type of environmental or motion state information.

[0111] To determine the confidence level of the environmental features at the perceived target location, the processor first extracts environmental features relevant to the target location from the robot's operational data matrix. These features may include obstacle distribution, terrain characteristics, and lighting conditions surrounding the target location. The extracted environmental features are then quantized for subsequent confidence calculation. This quantization process may involve converting continuous environmental feature values into discrete category labels or calculating feature statistics (such as mean and variance).

[0112] Environmental feature confidence is the processor's assessment of the accuracy of the environmental features at the target's perceived location. It reflects the processor's confidence in the authenticity of the environmental features at the target's location. Confidence can be calculated using a variety of methods, such as Bayesian estimation and neural network prediction. The processor may employ a statistical model-based approach to calculate environmental feature confidence. Specifically, the processor constructs a statistical model of the relationship between environmental features and confidence based on historical data and prior knowledge. The processor then inputs the environmental features at the target's perceived location into the statistical model and calculates the corresponding confidence value.

[0113] Specifically, the processor uses Formula 4 and determines the environmental feature confidence of the target perception position based on the robot operation data matrix , Formula 4 is:

[0114]

[0115] in, is the number of environment perception sensors in the environment perception sensor cluster, To correct the number of spatial positions, The first in the environmental perception sensor cluster The fusion weight of the environment perception sensors, The first in the environmental perception sensor cluster The environmental perception sensor The spatial correlation coefficient of the corrected spatial position, is the Sigmoid activation function, For the A bias term to correct the spatial position.

[0116] The spatial correlation coefficient is used in the formula to represent the spatial correlation of the environmental perception sensor with respect to the correction spatial position. It measures the closeness of the spatial relationship between the sensor data and the environmental characteristics of the target perception position. A high spatial correlation coefficient means that the sensor data is highly correlated with the environmental characteristics of the target position, and vice versa. Optionally, the determination of the spatial correlation coefficient can be based on statistical methods such as the Pearson correlation coefficient and the Spearman rank correlation coefficient. By calculating the spatial correlation coefficient between each sensor and each correction spatial position, a matrix can be obtained, in which each element represents the corresponding spatial correlation coefficient. In addition, the calculated spatial correlation coefficient can also be manually adjusted and optimized according to the needs of the actual application scenario and the performance requirements of the robot. For example, the optimal spatial correlation coefficient between different sensors and different correction spatial positions can be determined through experiments or empirical data to improve the accuracy and robustness of environmental perception.

[0117] The bias term is primarily used to adjust the calculated environmental feature confidence level to compensate for systematic errors or deviations in sensor data, making it more realistic. The bias term is an adjustable parameter used to adjust the calculated environmental feature confidence level to account for factors such as systematic errors in sensor data, environmental interference, or model imperfections. By properly setting the bias term, the calculated environmental feature confidence level can be made closer to reality, improving the accuracy and robustness of environmental perception.

[0118] Before setting the bias term, you need to conduct a preliminary analysis of the sensor data and environmental characteristics to understand factors such as the distribution characteristics of the sensor data, the complexity of the environmental characteristics, and possible systematic errors. Based on this preliminary analysis, you can make a preliminary estimate of the bias term and determine a rough range of values.

[0119] Collect sensor data and environmental characteristics through experiments, and use the initially estimated bias term to calculate the environmental characteristic confidence. Compare the calculated environmental characteristic confidence with the actual situation to evaluate the rationality and accuracy of the bias term setting. If a significant deviation is found between the environmental characteristic confidence and the actual situation, adjust the bias term and repeat the experiment to verify.

[0120] Based on experimental verification, the bias term can also be fine-tuned using optimization algorithms (such as gradient descent and genetic algorithms). The optimization goal is to minimize the error between the calculated confidence level of environmental features and the actual situation. Through repeated iterations and optimization, a set of optimal bias term values can be obtained.

[0121] Apply the optimized bias term to the actual robot's environmental perception system to collect more sensor data and environmental characteristics. Evaluate and verify the application results to ensure that the bias term setting meets the needs of the actual application scenario and the robot's performance requirements.

[0122] S204: The processor determines the target environment feature state of the target perception position according to the environment feature confidence and a preset environment feature confidence threshold.

[0123] Specifically, the processor determines the confidence level of the environmental features according to the confidence level of the environmental features. And a preset environmental feature confidence threshold is used to determine the target environmental feature state of the target perception position.

[0124] Among them, if the environmental feature confidence If the value of the environmental feature confidence threshold is greater than or equal to the preset environmental feature confidence threshold, it is determined that the target object exists at the target sensing position. If the confidence value is less than the preset environmental feature confidence threshold, it is determined that there is no obstacle at the target perception location.

[0125] The preset environmental feature confidence threshold is an important basis for the processor to determine the state of environmental features at the target sensing location. This threshold is set based on the actual application scenario and the robot's performance requirements. If the threshold is set too high, the processor may be overly conservative and miss some important environmental features; if the threshold is set too low, it may introduce excessive noise and cause misjudgment. Therefore, in actual applications, the threshold needs to be adjusted according to the specific situation.

[0126] After calculating the confidence level of the target's perceived environmental feature, the processor compares it to a preset environmental feature confidence threshold. If the confidence level is higher than the threshold, the processor deems the target's perceived environmental feature reliable and outputs it as the target environmental feature status. If the confidence level is lower than the threshold, the processor deems the target's perceived environmental feature to be uncertain or noisy, and may need further verification or be disregarded.

[0127] Finally, the processor outputs the determined target environment feature status to the robot's control system or other related modules. This status information can be used in the robot's decision-making processes such as path planning, obstacle avoidance, and task execution, improving the robot's autonomy and intelligence.

[0128] In S203-S204, Formula 4 uses the fusion of data from multiple environmental perception sensors to achieve multi-dimensional perception of the environmental characteristics of the target location. Each sensor has its own unique perception capabilities and limitations. Weighted fusion leverages the strengths of each sensor, improving the comprehensiveness and accuracy of environmental perception. The spatial correlation coefficient in the formula accounts for the spatial relationship between sensor data and the environmental characteristics of the target location. By introducing the spatial correlation coefficient, the contribution of sensor data to the target location can be more accurately assessed, further improving the accuracy of environmental perception.

[0129] Furthermore, the Sigmoid activation function is used in Equation 4. This function maps input values to the interval (0, 1), thereby obtaining a probability distribution of the environmental feature confidence. The smoothness of the Sigmoid function makes the calculation of environmental feature confidence insensitive to small changes in the input data, thus enhancing the robustness of environmental perception.

[0130] Furthermore, the introduction of a bias term can be used to adjust the calculated confidence level of environmental features to make them more consistent with reality. The introduction of a bias term can compensate for systematic errors or deviations in sensor data, further improving the accuracy and robustness of environmental perception.

[0131] The processor determines the state of the environmental features at the target sensing location based on the calculated environmental feature confidence level and a preset environmental feature confidence threshold. If the environmental feature confidence level is greater than or equal to the preset threshold, the target object (such as an obstacle or pedestrian) is determined to be present at the target sensing location. If the environmental feature confidence level is less than the preset threshold, the target object is determined to be absent. This confidence-based state determination method is simple, intuitive, and highly accurate. The preset environmental feature confidence threshold can be flexibly set based on the needs of the actual application scenario and the robot's performance requirements. This flexible threshold setting mechanism enables the robot to adapt to different environments and task requirements, improving its adaptability and practicality.

[0132] S205: The processor determines the confidence level of the environmental characteristics of each target's perceived position. Generates a context-aware distribution raster.

[0133] Specifically, the processor determines the confidence level of the environmental features of each target's perceived location. Generate an environment-aware distribution grid, where the confidence level of environmental features in is the row number of the environment perception distribution grid, is the column number of the environment perception distribution grid, which is used to perform gridding processing on the perception projection plane of the target perception space along the perception direction of the environment perception sensor.

[0134] In this solution, the target perception space is gridded along the perception projection plane of the environmental perception sensor's sensing direction to generate an environmental perception distribution grid, achieving planarization of environmental perception information. Each cell (i.e., grid point) in the grid corresponds to a specific environmental feature confidence value, allowing the robot to intuitively understand its perception of the surrounding environment. The environmental perception distribution grid graphically displays the distribution of environmental feature confidences within the target perception space. By analyzing the confidence values in the grid, the robot can quickly identify the location and range of potential target objects (such as obstacles and pedestrians), providing strong support for subsequent decision-making and action. The environmental perception distribution grid divides the target perception space into multiple small grid cells, achieving a finer division of environmental perception information. This fine division helps improve the accuracy of environmental perception, enabling the robot to more accurately identify the location and shape of target objects.

[0135] By generating an environmental perception distribution grid, the robot can efficiently process large amounts of environmental perception data. This grid-based processing allows the robot to quickly traverse and analyze the confidence values at the grid points, thereby improving environmental perception processing efficiency. The environmental perception distribution grid provides important environmental information support for the robot's autonomous navigation. Based on the confidence values in the grid, the robot can plan a safe path, avoid collisions with obstacles, and improve the safety and reliability of autonomous navigation.

[0136] By analyzing the distribution of confidence values within the environmental perception distribution grid, robots can make more intelligent decisions. For example, when a potential obstacle is detected, the robot can choose to detour or slow down to avoid it; when a pedestrian is detected, the robot can choose to yield or adjust its direction. By incorporating environmental perception distribution grid technology, the robot's processing of environmental perception information becomes more robust. Even in the presence of noise or errors in sensor data, the robot can still make reasonable decisions and actions by analyzing the distribution of confidence values within the grid.

[0137] S206: The processor determines the safety boundary of the obstacle according to the environment perception distribution grid.

[0138] The processor determines a safety boundary of the obstacle based on the environmental perception distribution grid, wherein the safety boundary is a grid in the environmental perception distribution grid where only one adjacent grid corresponds to an environmental feature confidence level greater than a preset environmental feature confidence level threshold.

[0139] In the above solution, by accurately determining the safety margin, the robot can avoid potentially dangerous areas during path planning and choose a safer route. This helps improve the robot's driving safety and reduce the probability of accidents. Specifically, the determination of the safety margin provides a crucial basis for the robot's obstacle avoidance tasks. Based on the position and shape of the safety margin, the robot can accurately determine the location and range of obstacles, enabling it to take more effective avoidance measures. By reducing the probability of collisions and accidents, the robot can complete its tasks more stably and efficiently, thereby improving the overall performance and reliability of the system.

[0140] Figure 3 FIG. 1 is a schematic diagram of a robot perception system according to an exemplary embodiment of the present application. Figure 3 As shown, the robot perception system 300 provided in this embodiment includes:

[0141] A processor 310 and an environment perception sensor cluster 320 composed of a plurality of environment perception sensors, wherein each environment perception sensor in the environment perception sensor cluster 320 is communicatively connected to the processor 310;

[0142] Acquire environmental perception data through each environmental perception sensor in the environmental perception sensor cluster 320 to generate an environmental perception data set, and send the environmental perception data set to the processor 310, wherein the environmental perception data includes a spatial position;

[0143] The processor 310 generates a robot operation data matrix according to the environment perception data set, wherein each robot operation feature vector in the robot operation data matrix includes an environment perception data sequence at the same correction space position;

[0144] The processor 310 maps the robot operation data matrix into a preset feature grid map to output an environment perception processing result.

[0145] Optionally, each environment perception sensor in the environment perception sensor cluster 320 is calibrated based on a different reference coordinate system.

[0146] Optionally, each row vector in the robot operation data matrix corresponds to an environmental perception data sequence at the same corrected spatial position, and each column vector includes environmental perception data of the same environmental perception sensor in the environmental perception sensor cluster 320 at different corrected spatial positions.

[0147] Optionally, the processor 310 generates a robot operation data matrix according to the environmental perception data set, including:

[0148] The processor 310 uses a preset spatial position correction model and generates the robot operation data matrix according to the environmental perception data set and the calibration parameters of each environmental perception sensor in the environmental perception sensor cluster 320.

[0149] Optionally, the processor 310 determines an environmental perception result according to the robot operation data matrix, including:

[0150] The processor 310 determines the environmental feature confidence level of the target sensing position based on the robot operation data matrix;

[0151] The processor 310 determines the target environmental feature state of the target perception position according to the environmental feature confidence and a preset environmental feature confidence threshold.

[0152] Optionally, after determining the environmental feature confidence level of the target perception position according to the robot operation data matrix, the method further includes:

[0153] The processor 310 generates an environment perception distribution grid according to the confidence level of the environment characteristics of each target perception position, and the environment perception distribution grid is used to perform grid processing on the perception projection plane of the target perception space along the perception direction of the environment perception sensor.

[0154] Optionally, after the processor 310 generates the environment perception distribution grid according to the confidence level of the environment feature of each target perception location, the process further includes:

[0155] The processor 310 determines a safety boundary according to the environmental awareness distribution grid, wherein the safety boundary is a grid in the environmental awareness distribution grid where only one adjacent grid corresponds to an environmental feature confidence level greater than a preset environmental feature confidence level threshold.

[0156] Figure 4 FIG. 1 is a schematic diagram of the structure of an electronic device according to an exemplary embodiment of the present application. Figure 4 As shown, this embodiment provides an electronic device 400 including: a processor 401 and a memory 402; wherein:

[0157] The memory 402 is used to store computer programs. The memory may also be a flash memory.

[0158] The processor 401 is configured to execute the execution instructions stored in the memory to implement each step in the above method. For details, please refer to the relevant description in the above method embodiment.

[0159] Optionally, the memory 402 may be independent or integrated with the processor 401 .

[0160] When the memory 402 is a device independent of the processor 401, the electronic device 400 may further include:

[0161] The bus 403 is used to connect the memory 402 and the processor 401 .

[0162] This embodiment further provides a readable storage medium, in which a computer program is stored. When at least one processor of an electronic device executes the computer program, the electronic device executes the methods provided in the various aforementioned embodiments.

[0163] This embodiment further provides a program product, which includes a computer program stored in a readable storage medium. At least one processor of an electronic device can read the computer program from the readable storage medium, and at least one processor can execute the computer program to cause the electronic device to implement the methods provided in the various embodiments described above.

[0164] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present application that follow the general principles of the present application and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered merely as exemplary, and the true scope and spirit of the present application are indicated by the claims.

[0165] It should be understood that the present application is not limited to the exact structure described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A robot environment perception data processing method, characterized in that: Applied to a robot perception system, the robot perception system includes a processor and an environment perception sensor cluster consisting of multiple environment perception sensors, each environment perception sensor in the environment perception sensor cluster being communicatively connected to the processor; the method includes: Acquire environmental perception data through each environmental perception sensor in the environmental perception sensor cluster to generate an environmental perception data set, and send the environmental perception data set to the processor, wherein the environmental perception data includes a spatial position and is information about the environment surrounding the robot; The processor generates a robot operation data matrix according to the environmental perception data set, wherein each robot operation feature vector in the robot operation data matrix includes an environmental perception data sequence at the same correction space position, each row vector in the robot operation data matrix corresponds to an environmental perception data sequence at the same correction space position, and each column vector includes environmental perception data of the same environmental perception sensor in the environmental perception sensor cluster at different correction space positions; The processor maps the robot operation data matrix to a preset feature grid map to output an environmental perception processing result, including: The processor uses the following formula and determines the environmental feature confidence of the target perception position based on the robot operation data matrix: , the formula is: in, is the number of environment perception sensors in the environment perception sensor cluster, To correct the number of spatial positions, is the first in the environment perception sensor cluster The fusion weight of the environmental perception sensors, is the first in the environment perception sensor cluster The environmental perception sensor The spatial correlation coefficient of the corrected spatial position, is the Sigmoid activation function, For the A bias term to correct the spatial position; The robot runs the data matrix The first position under the correction space Robot operation data; The processor determines the target environmental feature state of the target perception location based on the environmental feature confidence and a preset environmental feature confidence threshold, wherein if the environmental feature confidence is greater than or equal to the preset environmental feature confidence threshold, it is determined that there is a target object at the target perception location; if the environmental feature confidence is less than the preset environmental feature confidence threshold, it is determined that there is no obstacle at the target perception location.

2. The robot environment perception data processing method according to claim 1, characterized in that: Each environment perception sensor in the environment perception sensor cluster is calibrated based on a different reference coordinate system.

3. The robot environment perception data processing method according to claim 1 or 2, characterized in that: The processor generates a robot operation data matrix according to the environmental perception data set, including: The processor uses a preset spatial position correction model and generates the robot operation data matrix according to the environmental perception data set and the calibration parameters of each environmental perception sensor in the environmental perception sensor cluster.

4. The robot environment perception data processing method according to claim 1, characterized in that: After determining the environmental feature confidence level of the target perception position according to the robot operation data matrix, the method further includes: The processor generates an environment perception distribution grid according to the confidence level of the environment characteristics of each target perception position, and the environment perception distribution grid is used to perform grid processing on the perception projection plane of the target perception space along the perception direction of the environment perception sensor.

5. The robot environment perception data processing method according to claim 4, characterized in that: After the processor generates an environment perception distribution grid according to the confidence level of the environment characteristics of each target perception position, the method further includes: The processor determines a safety boundary according to the environmental awareness distribution grid, wherein the safety boundary is a grid in the environmental awareness distribution grid where only one adjacent grid corresponds to an environmental feature confidence level greater than a preset environmental feature confidence level threshold.

6. A robot perception system, characterized in that: include: A processor and an environment perception sensor cluster consisting of a plurality of environment perception sensors, wherein each environment perception sensor in the environment perception sensor cluster is communicatively connected to the processor; Acquire environmental perception data through each environmental perception sensor in the environmental perception sensor cluster to generate an environmental perception data set, and send the environmental perception data set to the processor, wherein the environmental perception data includes a spatial position and is information about the environment surrounding the robot; The processor generates a robot operation data matrix according to the environmental perception data set, wherein each robot operation feature vector in the robot operation data matrix includes an environmental perception data sequence at the same correction space position, each row vector in the robot operation data matrix corresponds to an environmental perception data sequence at the same correction space position, and each column vector includes environmental perception data of the same environmental perception sensor in the environmental perception sensor cluster at different correction space positions; The processor maps the robot operation data matrix to a preset feature grid map to output an environmental perception processing result, including: The processor uses the following formula and determines the environmental feature confidence of the target perception position based on the robot operation data matrix: , the formula is: in, is the number of environment perception sensors in the environment perception sensor cluster, To correct the number of spatial positions, is the first in the environment perception sensor cluster The fusion weight of the environmental perception sensors, is the first in the environment perception sensor cluster The environmental perception sensor The spatial correlation coefficient of the corrected spatial position, is the Sigmoid activation function, For the A bias term to correct the spatial position; The robot runs the data matrix The first position under the correction space Robot operation data; The processor determines the target environmental feature state of the target perception location based on the environmental feature confidence and a preset environmental feature confidence threshold, wherein if the environmental feature confidence is greater than or equal to the preset environmental feature confidence threshold, it is determined that there is a target object at the target perception location; if the environmental feature confidence is less than the preset environmental feature confidence threshold, it is determined that there is no obstacle at the target perception location.

7. An electronic device, characterized in that: include: processor; as well as, a memory for storing executable instructions of the processor; The processor is configured to perform the method according to any one of claims 1 to 5 by executing the executable instructions.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 5 when executed by a processor.

Images