Robot obstacle detection method and system in smart factory environment

Abstract

The application provides a robot obstacle detection method and system in a smart factory environment, and relates to the technical field of obstacles. The method comprises the following steps: determining strong light areas and high temperature areas of the factory according to light intensity data and temperature distribution data; regarding the overlapping area of the strong light areas and the high temperature areas as a perception blind area, and regarding the remaining area as a non-perception blind area; evaluating the perception ability level of each robot according to metal dust concentration data, and regarding the robot with a perception ability level greater than a preset level and located in the non-perception blind area as a collaborative robot; determining the obstacle position and motion state in the perception blind area by combining the environmental data sent by the collaborative robot and the robot in the perception blind area; constructing an obstacle avoidance map according to the obstacle position and motion state, and planning an obstacle avoidance path for a goods robot to be crossed through the perception blind area. The application has the technical effect of accurately detecting the position of the obstacle and reducing the risk of collision of the robot.

Description

Technical Field

[0001] This application relates to the field of obstacle detection technology, specifically to a robot obstacle detection method and system in a smart factory environment. Background Technology

[0002] With the rapid development of smart factories, robots are increasingly being used in production environments. In a smart factory environment, multiple robots work collaboratively to improve production efficiency and flexibility. However, due to the complexity of the factory environment and the presence of various obstacles, such as raw materials, semi-finished products, and finished products, robot movement is interfered with, affecting production efficiency and safety. Therefore, accurately detecting obstacle locations and planning reasonable obstacle avoidance paths has become a challenge for robot motion control in smart factory environments.

[0003] Currently, common obstacle avoidance methods for robots typically involve equipping them with devices such as lidar and visual sensors to perceive their surroundings and avoid obstacles. However, in harsh environments such as strong light and high temperatures, these methods can cause the robot's sensing devices to malfunction due to interference, resulting in the obstacle avoidance system failing to accurately detect the location of obstacles and increasing the risk of collisions between the robot and humans. Summary of the Invention

[0004] This application provides a robot obstacle detection method, system, device, and storage medium in a smart factory environment, which can accurately detect the position of obstacles and reduce the risk of robot collisions.

[0005] In a first aspect, this application provides a robot obstacle detection method in a smart factory environment. The method includes: acquiring temperature distribution data, metal dust concentration data, and light intensity data sent by multiple robots in the factory; determining the strong light area of ​​the factory based on the light intensity data, and determining the high temperature area of ​​the factory based on the temperature distribution data; designating the overlapping area of ​​the strong light area and the high temperature area as a perception blind zone, and the remaining area as a non-perception blind zone; evaluating the perception capability level of each robot based on the metal dust concentration data, and designating robots with a perception capability level greater than a preset level and located in the non-perception blind zone as collaborative robots; combining the first environmental perception data sent by the collaborative robots and the second environmental perception data sent by the robots in the perception blind zone, determining the position and movement state of obstacles in the perception blind zone; constructing an obstacle avoidance map based on the obstacle position and the movement state, and planning an obstacle avoidance path for a cargo robot to traverse the perception blind zone based on the obstacle avoidance map.

[0006] By adopting the above technical solution, and by acquiring data on temperature distribution, metal dust concentration, and light intensity of multiple robots in the factory, the overlapping area of ​​strong light and high temperature areas is identified as the perception blind zone. The perception ability level of each robot is evaluated, and robots with stronger perception capabilities in the non-perception blind zone are selected as collaborative robots. By combining the first environmental perception data of the collaborative robots and the second environmental perception data of the robots in the perception blind zone, the position and movement status of obstacles in the perception blind zone can be accurately obtained. This allows for the construction of an obstacle avoidance map and the planning of a reasonable obstacle avoidance path, accurately detecting the position of obstacles and reducing the risk of robot collisions.

[0007] Optionally, determining the strong light area of ​​the factory based on the light intensity data includes: performing time-domain filtering on the light intensity data to obtain target light intensity data; establishing a light intensity distribution matrix based on the target light intensity data; dividing each area in the factory into different light levels based on the light intensity distribution matrix; and determining the area with a light level greater than a preset level as a strong light area.

[0008] By adopting the above technical solution, the target light intensity data is obtained by performing time-domain filtering on the light intensity data, and a light intensity distribution matrix is ​​established. The various areas in the factory are divided into different light levels, and the areas with light levels greater than the preset level are identified as strong light areas. This can effectively eliminate noise interference in the light intensity data and improve the accuracy of strong light area identification.

[0009] Optionally, determining the high-temperature region of the factory based on the temperature distribution data includes: performing spatial interpolation on the temperature distribution data to obtain a temperature distribution heat map of the factory area; and determining the region with a temperature higher than a preset temperature threshold and a temperature gradient greater than a preset gradient threshold as a high-temperature region based on the temperature distribution heat map.

[0010] By adopting the above technical solution, a temperature distribution heatmap of the factory area can be obtained by spatial interpolation of the temperature distribution data. Based on the temperature distribution heatmap, areas with temperatures higher than a preset temperature threshold and temperature gradients greater than a preset gradient threshold are identified as high-temperature areas. This can achieve a continuous expression of the temperature distribution of the factory area and improve the accuracy of high-temperature area identification.

[0011] Optionally, evaluating the perception capability level of each robot based on the metal dust concentration data includes: determining the dust concentration at the location of each robot based on the metal dust concentration data; and determining the perception capability level of each robot based on the dust concentration, wherein the perception capability level is inversely proportional to the dust concentration.

[0012] By adopting the above technical solution, the dust concentration at each robot's location can be determined based on metal dust concentration data, and an inverse relationship between perception level and dust concentration can be established. This allows for an accurate assessment of the impact of metal dust on the robot's perception ability, thus enabling a more rational selection of collaborative robots with stronger perception capabilities.

[0013] Optionally, determining the position and motion state of obstacles within the perception blind zone by combining the first environmental perception data sent by the collaborative robot and the second environmental perception data sent by the robot in the perception blind zone includes: using the first environmental perception data as primary perception data and the second environmental perception data as auxiliary perception data; determining the initial position and initial motion state of obstacles within the perception blind zone based on the primary perception data; correcting the initial position and initial motion state based on the auxiliary perception data to obtain the corrected target obstacle position and target motion state; and using the target obstacle position and target motion state as the obstacle position and motion state within the perception blind zone.

[0014] By adopting the above technical solution, the initial position and initial motion state of the obstacle are determined by using the first environmental perception data of the collaborative robot as the main perception data and the second environmental perception data of the robot in the perception blind zone as the auxiliary perception data. The position and motion state of the target obstacle are obtained by using the auxiliary perception data for correction. This achieves multi-source data fusion perception of obstacles in the perception blind zone and improves the reliability of obstacle position and motion state recognition.

[0015] Optionally, constructing an obstacle avoidance map based on the obstacle's location and motion state includes: predicting the obstacle's trajectory within a preset time period based on the obstacle's location and motion state; mapping the trajectory to the factory's regional coordinate system to obtain the obstacle's influence area; determining the risk level of each area within the factory by combining the obstacle's influence area with the distribution of multiple devices in the factory; and generating an obstacle avoidance map based on the risk level.

[0016] By adopting the above technical solution, the obstacle's trajectory is predicted within a preset time period and mapped onto the factory area coordinate system to obtain the obstacle's influence area. Combined with the distribution of factory equipment, the risk level of each area is determined and an obstacle avoidance map is generated. This achieves an accurate grasp of the obstacle's dynamic characteristics and a reasonable division of risk areas, improving the timeliness and practicality of the obstacle avoidance map.

[0017] Optionally, after planning an obstacle avoidance path for the cargo robot to cross the perception blind zone based on the obstacle avoidance map, the method further includes: establishing a communication network among multiple cargo robots within the perception blind zone; and allocating a time window for each cargo robot to cross the perception blind zone according to the task priority of each cargo robot.

[0018] By adopting the above technical solution, establishing a communication network between multiple cargo robots within the perception blind zone, and allocating time windows for crossing the perception blind zone according to the task priority of each cargo robot, the orderly scheduling of cargo robots within the perception blind zone is realized, avoiding the congestion and collision risks that may occur when multiple robots cross the perception blind zone at the same time.

[0019] Secondly, this application provides a robot obstacle detection system in a smart factory environment, the system comprising: an acquisition module, a determination module, an output module, an evaluation module, a combination module, and a generation module; wherein,

[0020] The acquisition module is used to acquire temperature distribution data, metal dust concentration data, and light intensity data sent by multiple robots in the factory; the determination module is used to determine the strong light area of ​​the factory based on the light intensity data, and the high temperature area of ​​the factory based on the temperature distribution data; the output module is used to define the overlapping area of ​​the strong light area and the high temperature area as the perception blind zone, and the remaining area as the non-perception blind zone; the evaluation module is used to evaluate the perception capability level of each robot based on the metal dust concentration data, and designate the robot with a perception capability level greater than a preset level and located in the non-perception blind zone as a collaborative robot; the combination module is used to combine the first environmental perception data sent by the collaborative robot and the second environmental perception data sent by the robot in the perception blind zone to determine the position and movement state of obstacles in the perception blind zone; the generation module is used to construct an obstacle avoidance map based on the obstacle position and the movement state, and plan an obstacle avoidance path for the cargo robot to cross the perception blind zone based on the obstacle avoidance map.

[0021] Thirdly, this application provides an electronic device that adopts the following technical solution: it includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to enable the electronic device to execute a computer program such as the robot obstacle detection method in any of the above-mentioned smart factory environments.

[0022] Fourthly, this application provides a computer-readable storage medium, which adopts the following technical solution: storing a computer program that can be loaded by a processor and executed by any of the above-mentioned robot obstacle detection methods in a smart factory environment.

[0023] In summary, this application includes at least one of the following beneficial technical effects:

[0024] By acquiring data on temperature distribution, metal dust concentration, and light intensity of multiple robots within the factory, the overlapping area between strong light and high temperature areas is identified as a perception blind zone. The perception capability level of each robot is then evaluated, and robots with stronger perception capabilities within the non-perception blind zone are selected as collaborative robots. By combining the first environmental perception data of the collaborative robots with the second environmental perception data of the robots in the perception blind zone, the position and movement status of obstacles within the perception blind zone can be accurately obtained. This allows for the construction of an obstacle avoidance map and the planning of a reasonable obstacle avoidance path, accurately detecting the position of obstacles and reducing the risk of robot collisions. Attached Figure Description

[0025] Figure 1 This is a flowchart illustrating a robot obstacle detection method in a smart factory environment provided in an embodiment of this application;

[0026] Figure 2 This is a schematic diagram of the structure of a robot obstacle detection system in a smart factory environment provided in an embodiment of this application;

[0027] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0028] Explanation of reference numerals in the attached figures: 1000, electronic device; 1001, processor; 1002, communication bus; 1003, user interface; 1004, network interface; 1005, memory. Detailed Implementation

[0029] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0030] In the description of the embodiments in this application, words such as "illustrative," "for example," or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "illustrative," "for example," or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Rather, the use of words such as "illustrative," "for example," or "for example" is intended to present the relevant concepts in a specific manner.

[0031] Figure 1 This is a flowchart illustrating a robot obstacle detection method in a smart factory environment provided in an embodiment of this application. Figure 1 As shown, the method includes S101-S106:

[0032] S101 acquires temperature distribution data, metal dust concentration data, and light intensity data sent by multiple robots within the factory.

[0033] In a smart factory environment, factors such as high temperatures, strong light, and metal dust generated during production and processing can affect the performance of robot sensors, leading to a decrease in obstacle detection accuracy. To accurately identify perception blind spots in the factory environment and improve the reliability of obstacle detection, this embodiment first acquires environmental perception data sent by multiple robots within the factory.

[0034] Specifically, each robot in the factory is equipped with a temperature sensor, a metal dust concentration sensor, and a light intensity sensor. The temperature sensor is an infrared array type, with a measurement range of -40℃ to 150℃ and a measurement accuracy of ±0.5℃. The metal dust concentration sensor is a laser scattering type, with a detection range of 0-1000 mg / m³ and a resolution of 0.1 mg / m³. The light intensity sensor is a digital light sensor, with a measurement range of 1-65535 lux and a response time of less than 1 ms.

[0035] Each robot transmits the collected environmental perception data to the central control system in real time via a wireless communication network. To ensure the real-time performance and reliability of data transmission, an industrial IoT communication protocol based on a 5G network is adopted, with a data transmission latency of less than 10ms and a packet loss rate of less than 0.01%. During data transmission, the collected raw data is timestamped and tagged with a device ID for subsequent data processing and analysis.

[0036] The frequency of environmental sensing data acquisition is dynamically adjusted according to the factory's production conditions. Under normal operating conditions, the data acquisition frequency is 10Hz; when a sudden change in environmental parameters is detected, the acquisition frequency is automatically increased to 50Hz to capture the detailed features of the environmental change. Simultaneously, to reduce the burden of data storage and transmission, a data preprocessing mechanism is employed to reduce noise and compress the raw data, retaining only the valid data for transmission.

[0037] By acquiring this environmental perception data, a comprehensive understanding of temperature distribution, metal dust concentration distribution, and light intensity distribution within the factory can be obtained, laying a data foundation for subsequent identification of perception blind spots, evaluation of robot perception capabilities, and planning of obstacle avoidance paths. This multi-dimensional environmental perception data acquisition method can effectively improve the system's perception capability of the factory environment, providing a guarantee for achieving efficient and reliable robot obstacle detection.

[0038] S102, based on the light intensity data, determine the strong light areas of the factory, and based on the temperature distribution data, determine the high temperature areas of the factory.

[0039] In factory environments, strong light and high temperatures can significantly impact the performance of sensing devices such as robot vision sensors and LiDAR. For example, strong light can cause image overexposure and abnormal laser reflection, while high temperatures can cause thermal disturbances that lead to ranging errors. Therefore, accurately identifying these areas is crucial for subsequent obstacle detection.

[0040] First, the acquired illumination intensity data undergoes temporal filtering. Specifically, a bandpass filter is used, with a cutoff frequency set between 0.1Hz and 10Hz to filter out the effects of ambient light flicker and sensor noise. The filtered illumination intensity data more accurately reflects the steady-state illumination distribution of the factory environment. Based on the target illumination intensity data, an MxN dimensional illumination intensity distribution matrix is ​​established, where M and N correspond to the number of grid divisions in the X and Y axes of the factory area, respectively, with each grid measuring 0.5m x 0.5m.

[0041] Based on the light intensity distribution matrix, the factory area is divided into four light levels: low illuminance area (0-1000 lux), normal illuminance area (1000-3000 lux), high illuminance area (3000-5000 lux), and strong light area (>5000 lux). When the light level of a certain area is greater than the preset level (set as high illuminance area in this embodiment), it is marked as a strong light area. To improve the accuracy of area division, a bilinear interpolation algorithm is used at the grid boundaries to smoothly transition the light intensity.

[0042] Simultaneously, spatial interpolation was performed on the temperature distribution data, and a temperature distribution heatmap of the factory area was constructed using the Kriging interpolation algorithm. This algorithm considers the spatial correlation of the temperature data and can generate continuous temperature distribution estimates between sampling points. The spatial resolution of the temperature distribution heatmap is consistent with that of the illumination intensity distribution matrix, both being 0.5m x 0.5m.

[0043] Based on the temperature distribution heatmap, a temperature threshold of 45℃ and a temperature gradient threshold of 5℃ / m are set. When the temperature of a region exceeds the temperature threshold and the temperature gradient exceeds the gradient threshold, it is marked as a high-temperature region. The temperature gradient is calculated using the central difference method to capture regions with rapid temperature changes. These regions are typically adjacent to high-temperature industrial equipment such as heat processing equipment and smelting equipment.

[0044] Based on the above embodiments, as an optional implementation, in S102, determining the strong light area of ​​the factory according to the light intensity data specifically includes S21-S24:

[0045] S21, perform time-domain filtering on the illumination intensity data to obtain the target illumination intensity data.

[0046] In factory environments, the spatiotemporal distribution of light intensity directly affects the environmental perception performance of robots. To accurately identify areas of strong light that affect the normal operation of robots, it is necessary to systematically process and analyze the collected light intensity data. Since lighting conditions inside factories are affected by various factors, such as natural light, artificial lighting, and equipment reflections, the raw light intensity data often contains noise and instantaneous fluctuations, requiring temporal filtering to obtain stable and reliable light intensity characteristics.

[0047] In the time-domain filtering stage, a sliding median filter is used to process the raw illumination intensity data. The sliding window size is set to 10 sampling points, and the sampling frequency is 10Hz. This effectively suppresses interference from sudden changes in illumination while maintaining the long-term trend of illumination intensity. The target illumination intensity data obtained after filtering can more accurately reflect the actual illumination conditions in various areas of the factory.

[0048] S22, Based on the target light intensity data, establish a light intensity distribution matrix.

[0049] Based on the target illumination intensity data, an illumination intensity distribution matrix for the factory environment is established. The factory space is divided into several grid cells, each with a size of 0.5 meters × 0.5 meters. For each grid cell, the average illumination intensity value at its corresponding location is calculated and stored in the illumination intensity distribution matrix. This matrix representation method not only facilitates subsequent data processing but also intuitively displays the illumination distribution characteristics of the factory space.

[0050] S23, based on the light intensity distribution matrix, divides the various areas in the factory into different light levels.

[0051] S24, define areas with light levels greater than the preset level as strong light areas.

[0052] To achieve accurate segmentation of illuminated areas, the factory environment is divided into different illumination levels based on the illumination intensity distribution matrix. Specifically, a multi-threshold segmentation method is used to divide the illumination intensity into five levels: low illumination area, normal illumination area, medium intensity area, high intensity area, and extremely high intensity area. The segmentation thresholds for each level are determined based on actual industrial field measurement data and the performance indicators of the robot vision system, ensuring that the classification results match the actual working requirements of the robot.

[0053] Finally, areas with illumination levels higher than a preset level are designated as strong light areas. The preset level is set as medium-intensity light areas, meaning that high-intensity and extremely high-intensity light areas are uniformly marked as strong light areas. This classification method fully considers the dynamic range and anti-interference capability of the robot vision system, and can effectively identify areas that may lead to a decrease in the robot's perception performance.

[0054] Based on the above embodiments, as an optional implementation, in S102, determining the high-temperature zone of the factory according to the temperature distribution data specifically includes S31-S32:

[0055] S31, spatial interpolation is performed on the temperature distribution data to obtain a temperature distribution heat map of the factory area.

[0056] S32, based on the temperature distribution heat map, the area with a temperature higher than the preset temperature threshold and a temperature gradient greater than the preset gradient threshold is identified as a high temperature area.

[0057] In factory environments, the distribution of high-temperature equipment and heat sources can affect the normal operation and safety of robots. Due to the limited deployment locations of temperature sensors, it is impossible to directly obtain the continuous temperature distribution of the entire factory space. Therefore, it is necessary to construct a complete temperature distribution heat map through spatial interpolation methods and identify potential hazardous areas based on temperature values ​​and temperature gradient characteristics.

[0058] In the spatial interpolation stage, Kriging interpolation is used to process discrete temperature distribution data. This method comprehensively considers the correlation of spatial location and the weight distribution of measurement points, enabling it to reflect the spatial continuity of the temperature field while ensuring interpolation accuracy. Specifically, a spatial variogram model of the temperature data is first established to describe the variation of temperature values ​​with spatial distance. Then, based on this model, the optimal linear unbiased estimation of temperature values ​​at unmeasured points in the factory space is performed. The resolution of the interpolation grid is set to 0.2 meters, ensuring both the detailed representation of the temperature distribution and meeting the requirements of real-time computation.

[0059] The temperature distribution heatmap generated based on the interpolation results uses a pseudo-color display method, with different colors corresponding to different temperature ranges, making it easy to intuitively identify areas of temperature anomalies. The color mapping of the heatmap adopts a gradient scheme from blue (low temperature) to red (high temperature), and the color intensity of each grid point has a linear relationship with its temperature value.

[0060] High-temperature areas are identified using a dual-threshold method, considering both the absolute temperature value and the temperature gradient. The preset temperature threshold is determined based on the temperature resistance level of the robot components, typically set to 45°C. The preset temperature gradient threshold is determined based on a thermal damage risk assessment and is used to identify areas with drastic temperature changes. When the temperature value of a region exceeds the preset temperature threshold, and the temperature gradient (i.e., the rate of temperature change per unit distance) of that region exceeds the preset gradient threshold, it is marked as a high-temperature area.

[0061] S103 defines the overlapping area of ​​the strong light area and the high temperature area as the perception blind zone, and the remaining area as the non-perception blind zone.

[0062] In factory environments, the simultaneous presence of strong light and high temperatures creates a cumulative effect, significantly reducing a robot's perception capabilities. For example, strong light can cause overexposure in visual sensor images, while high temperatures can alter the air's refractive index, leading to inaccuracies in lidar ranging. When these two interfering factors coexist, traditional single-environment compensation methods become ineffective, creating areas with severely limited perception capabilities. Therefore, it is necessary to identify these overlapping areas and define them as perception blind spots in order to implement specialized perception strategies.

[0063] In practice, the spatial distribution data of the strong light area and high temperature area obtained in the aforementioned steps are first processed to unify the data format. Since both types of data are represented using a 0.5m x 0.5m grid, spatial overlap analysis can be performed directly. Under a unified factory coordinate system, binary matrices A and B are established, where matrix A represents the distribution of the strong light area (strong light area is denoted as 1, and non-strong light area is denoted as 0), and matrix B represents the distribution of the high temperature area (high temperature area is denoted as 1, and non-high temperature area is denoted as 0).

[0064] The overlapping region matrix C is obtained through matrix operation C = A⊙B (where ⊙ represents element-wise multiplication). When an element in matrix C has a value of 1, it indicates that the grid location is simultaneously affected by strong light and high temperature interference, and this location is marked as a sensing blind zone. To avoid abrupt changes in the boundary of the sensing blind zone, a morphological processing method is used to smooth it. Specifically, a structuring element with a radius of 1m is used for dilation to ensure the continuity and safety margin of the sensing blind zone.

[0065] For regions where the element value in matrix C is 0, i.e., regions where there is no overlap between strong light and high temperature, these are marked as non-perception blind zones. Although these regions may have strong light or high temperature interference on their own, the robot can still maintain basic perception capabilities through a single environmental compensation strategy. For ease of subsequent processing, the factory environment is divided into three types of regions: perception blind zones (regions where strong light and high temperature overlap), partially restricted regions (regions where only strong light or high temperature exists), and normal perception regions (regions where there is neither strong light nor high temperature).

[0066] S104. Based on the metal dust concentration data, evaluate the perception ability level of each robot, and designate the robot with a perception ability level greater than the preset level and located in the non-perception blind zone as the collaborative robot.

[0067] Metal dust in factory environments can obstruct and scatter the vision sensors, lidar, and other sensing devices of robots, reducing their accuracy. The higher the concentration of metal dust, the greater the impact on sensing devices. Therefore, it is necessary to evaluate the sensing capabilities of each robot based on metal dust concentration data in order to select suitable collaborative robots for obstacle detection tasks.

[0068] First, a time window analysis was performed on the metal dust concentration data at each robot's location. The time window length was set to 10 seconds, and statistical characteristics were calculated for the dust concentration data within the window, including the mean μ, standard deviation σ, and coefficient of variation CV (CV = σ / μ). These statistical characteristics reflect the degree of dust pollution and its stability in the robot's environment.

[0069] Based on these statistical characteristics, a formula for evaluating robot perception capabilities is established. This formula employs a weighted scoring method, specifically calculated as follows: Perception capability score S = w1 × (1 - μ / μmax) + w2 × (1 - σ / σmax) + w3 × (1 - CV / CVmax); where w1, w2, and w3 are weighting coefficients, optimized experimentally to 0.5, 0.3, and 0.2 respectively; μmax, σmax, and CVmax are the maximum values ​​from historical data. The closer the environmental parameters are to the ideal state (i.e., the smaller the values ​​of each indicator), the higher the score.

[0070] Based on the perception ability score S, the robot's perception ability is divided into four levels: Excellent (S≥0.8): The performance of the perception device is almost unaffected; Good (0.6≤S<0.8): The performance of the perception device is slightly affected; Average (0.4≤S<0.6): The performance of the perception device is significantly affected; Poor (S<0.4): The performance of the perception device is severely affected.

[0071] In this embodiment, the preset level is set to "Good", which requires the collaborative robot's perception ability score to be no less than 0.6. Simultaneously, based on the perception blind spot distribution map obtained in the preceding steps, robots that simultaneously meet the following conditions are selected as collaborative robots: perception ability level greater than or equal to "Good" (S≥0.6), located in a non-perception blind spot (including partially restricted areas and normal perception areas), and currently not occupied by other tasks.

[0072] To ensure the real-time performance and reliability of collaborative perception, the system updates the robot's perception capability assessment results every 1 second and dynamically adjusts the selection of collaborative robots. When the perception capability level of a collaborative robot decreases or enters a perception blind zone, the system automatically selects another qualified robot to replace it.

[0073] Based on the above embodiments, as an optional implementation, in S104, evaluating the perception capability level of each robot based on the metal dust concentration data specifically includes S41-S42:

[0074] S41, Based on the metal dust concentration data, determine the dust concentration at the location of each robot.

[0075] S42, Determine the perception level of each robot based on the dust concentration, wherein the perception level is inversely proportional to the dust concentration.

[0076] In factory environments, metal dust is a significant environmental factor affecting the perception performance of robots. Metal dust not only interferes with the ranging accuracy of lidar but also degrades the imaging quality of vision sensors. Therefore, it is necessary to dynamically assess the robot's perception capabilities based on the dust concentration at its location in order to rationally allocate perception tasks during subsequent collaborative perception processes.

[0077] To determine the dust concentration at the robot's location, a nearest neighbor interpolation method is used to process the dust concentration data. First, the measurements from the nearest dust sensors to the robot are obtained. Then, based on the spatial distance between the sensor locations and the robot's location, the dust concentration at the robot's location is calculated. To improve the accuracy of the assessment, a distance weighting factor is introduced into the calculation process, giving greater influence to the measurements from closer sensors.

[0078] Based on the obtained dust concentration values, a perception capability level evaluation model was established. This model classifies the robot's perception capability into five levels: Excellent, Good, Average, Poor, and Very Poor. The specific level classification criteria are as follows: when the dust concentration is below 0.1 mg / m³, the perception capability level is Excellent; when the dust concentration is in the range of 0.1-0.5 mg / m³, it is Good; in the range of 0.5-1.0 mg / m³, it is Average; in the range of 1.0-2.0 mg / m³, it is Poor; and when it exceeds 2.0 mg / m³, it is Very Poor. This classification method fully considers the sensitivity of different sensors to dust interference, ensuring that the evaluation results accurately reflect the actual perception performance.

[0079] Since the perception level is inversely proportional to the dust concentration, the evaluation model uses an exponential decay function to describe this relationship. Specifically, the perception level score S can be expressed as: S = A × e^(-k × C), where A is the full score coefficient (set to 100), k is the decay coefficient (determined based on experimental calibration), and C is the dust concentration value. This mathematical model reflects both the basic law that perception ability decreases with increasing dust concentration and the rapid deterioration of perception ability in high concentration ranges.

[0080] S105, combining the first environmental perception data sent by the collaborative robot and the second environmental perception data sent by the robot in the perception blind zone, determine the position and motion state of obstacles in the perception blind zone.

[0081] In factory environments, robots in blind spots are subject to significant uncertainties in their environmental perception data due to interference from both strong light and high temperatures. To accurately detect obstacles in these blind spots, it is necessary to fuse and analyze the environmental perception data from collaborative robots and robots in the blind spots. This approach leverages the complementary advantages of different perspectives and sensors to improve the reliability of obstacle detection.

[0082] First, the initial environmental perception data sent by the collaborative robot is preprocessed. This initial environmental perception data includes LiDAR point cloud data (scanning frequency 10Hz, angular resolution 0.25°) and visual image data (resolution 1920×1080, frame rate 30fps). Spatial filtering and downsampling are performed on the point cloud data to remove outliers and reduce data redundancy; distortion correction and illumination compensation are performed on the visual images to improve image quality.

[0083] Simultaneously, the system processes the second environmental perception data sent by the robot in the perception blind spot. Although this second environmental perception data is significantly affected by environmental interference, it still contains valuable information, mainly including degraded point cloud data and blurred visual images. An adaptive thresholding method is used on this data to extract key information that may contain obstacle features.

[0084] To achieve spatiotemporal alignment of multi-source data, a unified spatiotemporal reference system is established. In the time dimension, timestamp alignment technology is used to synchronize data from different sensors to a unified time base; in the spatial dimension, robot pose estimation and coordinate transformation are used to convert all perceived data to the factory's global coordinate system.

[0085] In the data fusion stage, an improved Probabilistic Hypothesis Density (PHD) filtering algorithm is used for multi-sensor data fusion. This algorithm can simultaneously handle multiple sources of uncertainty, including sensing noise, changes in detection probability, and false alarms. The specific fusion process includes the following steps:

[0086] Feature extraction: Extract the geometric features (size, shape) and motion features (velocity, acceleration) of obstacles from the first environmental perception data as highly reliable observations.

[0087] State estimation: An obstacle state estimation model is established by combining fuzzy features extracted from the second environmental perception data. The state vector includes position coordinates (x, y), velocity components (vx, vy), and acceleration components (ax, ay).

[0088] Dynamic update: The obstacle state estimate is updated recursively using the PHD filter. The measurement update weights of the filter are dynamically adjusted according to the data reliability. The weight of the first environmental perception data is higher (usually 0.7-0.9), and the weight of the second environmental perception data is lower (usually 0.1-0.3).

[0089] To improve real-time performance, a parallel computing architecture is adopted to implement the data fusion algorithm. Feature extraction and state estimation tasks are distributed to multiple processing cores, and GPUs are used to accelerate matrix operations, keeping the latency of a single fusion processing operation within 50ms.

[0090] Based on the above embodiments, as an optional implementation, in S105, combining the first environmental perception data sent by the collaborative robot and the second environmental perception data sent by the robot in the perception blind spot, determining the position and motion state of obstacles in the perception blind spot specifically includes S51-S54:

[0091] S51, the first environmental perception data is used as the primary perception data, and the second environmental perception data is used as the auxiliary perception data.

[0092] S52, based on the main perception data, determine the initial position and initial motion state of obstacles in the perception blind zone.

[0093] S53, based on the auxiliary sensing data, correct the initial position and initial motion state to obtain the corrected target obstacle position and target motion state.

[0094] S54, the position of the target obstacle and the target's motion state are taken as the position and motion state of the obstacle within the perception blind zone.

[0095] In factory environments with multiple robots working collaboratively, individual robots often have perception blind spots due to factors such as strong light, high temperatures, and metal dust, making it impossible for them to accurately perceive obstacles in the environment. To ensure the safety and reliability of the robot system, it is necessary to use a multi-robot collaborative perception approach, comprehensively utilizing the perception data of different robots to achieve precise localization and motion state estimation of obstacles within the perception blind spots.

[0096] During data processing, the first environmental perception data sent by the collaborative robot is used as the primary perception data because the collaborative robot is in an area with better environmental conditions, and its perception data has higher accuracy and reliability. The second environmental perception data sent by the robot in the perception blind spot is used as auxiliary perception data. Although its perception ability is limited, this data still contains valuable environmental information and can be used to verify and supplement the observation results of the primary perception data.

[0097] When determining the initial position and initial motion state of an obstacle based on the main sensing data, a Kalman filter algorithm is used for target tracking. This algorithm, through two stages—prediction and update—effectively handles noise in the sensing data and provides optimal estimates of the obstacle's position and velocity. Specifically, the motion state equation and observation equation of the obstacle are established, where the state vector contains the obstacle's two-dimensional coordinates and velocity components. Through recursive calculation, the initial position coordinates (x0, y0) and initial velocity vector (vx0, vy0) of the obstacle are obtained.

[0098] When using auxiliary sensing data for correction, a confidence-based data verification method is employed. First, the difference between the two sets of sensing data is calculated. When the difference exceeds a preset threshold, an anomaly detection mechanism is triggered. The reliability of the data is assessed by analyzing the temporal consistency, spatial continuity, and conformity with historical data of the two sets. If the primary sensing data is deemed reliable, its results are directly adopted; if anomalies are detected, further analysis using auxiliary sensing data is necessary, and the position and motion state of the obstacle are reassessed using a state estimator.

[0099] S106 constructs an obstacle avoidance map based on the location and movement of obstacles, and plans an obstacle avoidance path for the cargo robot to traverse the perception blind spot based on the obstacle avoidance map.

[0100] In factory environments, cargo robots need to safely and efficiently traverse blind spots. This requires constructing a dynamic obstacle avoidance map based on the obstacle information obtained in the preceding steps, and planning appropriate obstacle avoidance paths accordingly. Since obstacles within the blind spots may be in motion, traditional static map planning methods struggle to guarantee the safety and feasibility of the planned paths.

[0101] First, a dynamic obstacle avoidance map based on probabilistic grids is constructed. The factory environment is divided into uniform grids, each containing two attributes: an occupancy probability value and a dynamic risk value. The occupancy probability value reflects the likelihood that a grid will be occupied by an obstacle, ranging from [0,1]; the dynamic risk value represents the overall risk level after considering the obstacle's movement state. This calculation method ensures that obstacle areas with faster movement speeds or greater accelerations have higher risk values.

[0102] To address the uncertainty of obstacle motion, a spatiotemporal probabilistic prediction method is employed to update the obstacle avoidance map. Based on the current motion state of the obstacles, an improved Kalman filter is used to predict their possible future location distribution. The prediction results are mapped onto a grid map using a Gaussian probability model, forming a time-varying risk distribution. Considering that the uncertainty of the prediction increases over time, a time decay factor is introduced to attenuate the impact of long-term prediction results.

[0103] Based on the obstacle avoidance map, an improved spatiotemporal RRT* (Rapidly-exploring Random Tree*) algorithm is used for path planning. This algorithm adds a time dimension to the traditional RRT*, enabling the planned path to proactively avoid high-risk areas. A safe movement path is constructed through three main steps: random sampling, tree expansion, and path optimization. The cost function comprehensively considers path length, dynamic risk accumulation, and time penalty terms, and the optimal path is determined through a multi-objective optimization method.

[0104] To improve planning efficiency, an adaptive sampling strategy is adopted. A larger sampling step size is used in the initial planning phase to quickly find feasible solutions, and then a smaller sampling step size is used to refine the path in key areas (such as turning points and crossing points). Simultaneously, parallel computing technology is used to accelerate the path search process, ensuring real-time planning.

[0105] Based on the above embodiments, as an optional implementation, in S106, constructing the obstacle avoidance map according to the obstacle's position and movement state specifically includes S61-S64:

[0106] S61, based on the obstacle's position and movement state, predict the obstacle's trajectory within a preset time period.

[0107] S62 maps the motion trajectory to the factory's regional coordinate system to obtain the area affected by the obstacle.

[0108] S63, combining the area affected by the obstacle and the distribution of multiple pieces of equipment in the factory, determines the risk level of each area within the factory.

[0109] S64 generates an obstacle avoidance map based on the risk level.

[0110] First, based on the acquired obstacle positions (x, y) and motion states (vx, vy), a kinematic model is used to predict the obstacle's trajectory over a future time period. Considering the motion characteristics of obstacles (such as forklifts and workers) in a factory environment, a combination of a uniform motion model and a steering constraint model is used for trajectory prediction. The prediction time period is set to 10 seconds, and discretization is performed using 100 milliseconds as the time step. For each time step, the predicted position of the obstacle is calculated as: x(t) = x + vx × t + 0.5 × ax × t², y(t) = y + vy × t + 0.5 × ay × t², where ax and ay are the acceleration components of the obstacle, estimated from historical data.

[0111] When mapping the predicted motion trajectory to the factory's regional coordinate system, the actual size of the obstacle and its safety margin need to be considered. Using the outer contour of the obstacle as a reference, a safety distance (usually set to 0.5 meters) is extended outwards to form the obstacle's influence area. This area is represented by a polygon, and its vertex coordinates are calculated by combining the obstacle trajectory points with the safety margin. To improve computational efficiency, a convex hull algorithm is used to simplify the geometric representation of the influence area.

[0112] When determining the risk level of each area within the factory, three key factors are comprehensively considered: the area affected by obstacles, the location of factory equipment, and the functional attributes of the area. Risk levels are divided into four categories: hazardous areas, high-risk areas, medium-risk areas, and safe areas. Areas located within the area affected by obstacles are automatically classified as hazardous areas; areas adjacent to high-value equipment (such as precision machining equipment) are classified as high-risk areas; areas with frequent personnel flow are classified as medium-risk areas; and the remaining areas are safe areas. Simultaneously, considering the dynamic changes in equipment operating status, the system updates the risk assessment results for each area in real time.

[0113] Based on risk level information, the system generates a gridded obstacle avoidance map. The factory floor is divided into 0.1m x 0.1m grid cells, each assigned a corresponding risk value: infinity for hazardous areas (representing no passage), 80 for high-risk areas, 40 for medium-risk areas, and 10 for safe areas. To make the obstacle avoidance map smoother, a Gaussian blur algorithm is used to locally smooth the risk values, preventing drastic directional changes during robot path planning.

[0114] After planning obstacle avoidance paths for cargo robots to cross the perception blind zone based on the obstacle avoidance map, the process also includes: establishing a communication network between multiple cargo robots within the perception blind zone; and allocating time windows for each cargo robot to cross the perception blind zone based on the task priority of each cargo robot.

[0115] When establishing a communication network within the perception blind spot, a distributed self-organizing network architecture is adopted. Each cargo robot is equipped with a wireless communication module, using an industrial-grade WiFi protocol for data transmission. Communication content includes key information such as the robot's real-time position, speed, task status, and planned path. To ensure real-time performance and reliability, Time Division Multiple Access (TDMA) is used for channel allocation, assigning a fixed time segment to each robot for data broadcasting. Simultaneously, a communication redundancy mechanism is implemented; if a robot's communication module fails, adjacent robots can act as relay nodes to maintain network connectivity.

[0116] When determining the task priority of cargo robots, multiple factors are considered: task urgency, cargo importance, energy consumption, and path length. Robots carrying urgent tasks (such as perishable goods or critical production materials) receive higher priority; robots carrying important cargo (such as valuable or fragile items) receive lower priority; robots with low battery levels have relatively higher priority to prevent them from running out of energy in blind spots; robots with shorter planned paths also receive higher priority to improve traffic efficiency. The system calculates a comprehensive priority score based on these factors, ranging from 0 to 100, with higher scores indicating higher priority.

[0117] Based on the calculated priorities, the system employs a dynamic time window allocation strategy. First, the process of traversing the perception blind spot is divided into multiple time windows, the duration of which is determined by the length of the area and the robot's standard travel speed. To avoid interference between robots, a safety interval (typically 2 seconds) is set between adjacent time windows. High-priority robots preferentially select the earliest available time window, while the system reserves some emergency time windows to handle unforeseen circumstances. To improve passage efficiency, robots with similar priorities and non-intersecting paths are allowed to pass simultaneously.

[0118] Based on the above method, this application also discloses a robot obstacle detection system in a smart factory environment, such as... Figure 2 As shown, Figure 2 This is a schematic diagram of the structure of a robot obstacle detection system in a smart factory environment provided in an embodiment of this application. The system includes: an acquisition module, a determination module, an output module, an evaluation module, a combination module, and a generation module; wherein,

[0119] The system comprises the following modules: an acquisition module for acquiring temperature distribution data, metal dust concentration data, and light intensity data from multiple robots within the factory; a determination module for identifying areas of intense light and high temperature within the factory based on light intensity data; an output module for designating the overlapping areas of intense light and high temperature areas as blind spots and the remaining areas as non-blind spots; an evaluation module for assessing the perception capability level of each robot based on metal dust concentration data, designating robots with perception capability levels greater than a preset level and located within non-blind spots as collaborative robots; a combination module for combining the first environmental perception data from collaborative robots with the second environmental perception data from robots in the blind spots to determine the location and motion state of obstacles within the blind spots; and a generation module for constructing an obstacle avoidance map based on obstacle location and motion state, and planning obstacle avoidance paths for cargo robots attempting to traverse the blind spots based on the obstacle avoidance map.

[0120] It should be noted that the above embodiments of the apparatus are only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0121] Please see Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 3 As shown, the electronic device 1000 may include: at least one processor 1001, at least one network interface 1004, a user interface 1003, a memory 1005, and at least one communication bus 1002.

[0122] The communication bus 1002 is used to realize the connection and communication between these components.

[0123] The user interface 1003 may include a display screen and a camera. Optionally, the user interface 1003 may also include a standard wired interface and a wireless interface.

[0124] The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0125] The processor 1001 may include one or more processing cores. The processor 1001 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 1005, and by calling data stored in the memory 1005. Optionally, the processor 1001 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 1001 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content to be displayed on the screen; and the modem handles wireless communication. It is understood that the modem may also not be integrated into the processor 1001 and may be implemented as a separate chip.

[0126] The memory 1005 may include random access memory (RAM) or read-only memory. Optionally, the memory 1005 may include a non-transitory computer-readable storage medium. The memory 1005 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 1005 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 1005 may also be at least one storage device located remotely from the aforementioned processor 1001. Figure 3 As shown, the memory 1005, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for a robot obstacle detection method in a smart factory environment.

[0127] exist Figure 3In the electronic device 1000 shown, the user interface 1003 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 1001 can be used to call the application program stored in the memory 1005 for a robot obstacle detection method in a smart factory environment. When executed by one or more processors, the electronic device performs one or more of the methods described in the above embodiments.

[0128] An electronic device readable storage medium stores instructions that, when executed by one or more processors, cause the electronic device to perform one or more of the methods described in the above embodiments.

[0129] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0130] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0131] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some service interfaces; indirect couplings or communication connections between devices or units may be electrical or other forms.

[0132] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0133] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0134] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0135] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Other embodiments of this disclosure will be readily apparent to those skilled in the art upon consideration of the specification and practice of the disclosure herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described herein. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A method for robot obstacle detection in a smart factory environment, the method comprising: The method comprises: acquiring temperature distribution data, metal dust concentration data and illumination intensity data sent by multiple robots in a factory; determining a strong light area of the factory according to the illumination intensity data, and determining a high temperature area of the factory according to the temperature distribution data; regarding an overlapping area of the strong light area and the high temperature area as a perception blind area, and regarding a remaining area as a non-perception blind area; evaluating a perception ability level of each robot according to the metal dust concentration data, wherein the evaluation comprises: determining a dust concentration of a position where each robot is located according to the metal dust concentration data; determining a perception ability level of each robot according to the dust concentration, wherein the perception ability level is inversely proportional to the dust concentration; regarding a robot whose perception ability level is greater than a preset level and located in the non-perception blind area as a collaborative robot; determining a position and a motion state of an obstacle in the perception blind area by combining first environment perception data sent by the collaborative robot and second environment perception data sent by a robot in the perception blind area; constructing an obstacle avoidance map according to the position and the motion state of the obstacle, and planning an obstacle avoidance path for a cargo robot to be crossed through the perception blind area according to the obstacle avoidance map.

2. The method of claim 1, wherein the method further comprises: The determination of the strong light area of the factory according to the illumination intensity data comprises: performing time domain filtering on the illumination intensity data to obtain target illumination intensity data; establishing an illumination intensity distribution matrix according to the target illumination intensity data; dividing each area in the factory into different illumination levels according to the illumination intensity distribution matrix; regarding an area whose illumination level is greater than a preset level as a strong light area. 3.The robot obstacle detection method in a smart factory environment of claim 1, wherein, The determination of the high temperature area of the factory according to the temperature distribution data comprises: performing spatial interpolation on the temperature distribution data to obtain a temperature distribution thermal map of the area of the factory; regarding an area whose temperature is higher than a preset temperature threshold and whose temperature gradient is greater than a preset gradient threshold as a high temperature area according to the temperature distribution thermal map.

4. The method for robot obstacle detection in a smart factory environment as claimed in claim 1, wherein, The determination of the position and the motion state of the obstacle in the perception blind area by combining the first environment perception data sent by the collaborative robot and the second environment perception data sent by the robot in the perception blind area comprises: regarding the first environment perception data as main perception data and regarding the second environment perception data as auxiliary perception data; determining an initial position and an initial motion state of the obstacle in the perception blind area according to the main perception data; correcting the initial position and the initial motion state according to the auxiliary perception data to obtain a corrected target obstacle position and a target motion state; regarding the target obstacle position and the target motion state as the position and the motion state of the obstacle in the perception blind area.

5. The method for robot obstacle detection in a smart factory environment as claimed in claim 1, wherein, The construction of the obstacle avoidance map according to the position and the motion state of the obstacle comprises: predicting a motion trajectory of the obstacle in a preset time period according to the position and the motion state of the obstacle. mapping the motion trajectory into a region coordinate system of the factory to obtain an obstacle influence region; determining a risk level of each region in the factory in combination with the obstacle influence region and distribution positions of a plurality of devices in the factory; generating an obstacle avoidance map according to the risk level.

6. The method for robot obstacle detection in a smart factory environment as claimed in claim 1, wherein, After the obstacle avoidance path is planned for the goods robot to be crossed through the perception blind area according to the obstacle avoidance map, the method further includes: establishing a communication network among a plurality of goods robots in the perception blind area; allocating a time window for each goods robot to cross through the perception blind area according to a task priority of each goods robot.

7. A robot obstacle detection system in a smart factory environment, characterized by, The system includes an acquisition module, a determination module, an output module, an evaluation module, a combination module and a generation module; wherein, the acquisition module is configured to acquire temperature distribution data, metal dust concentration data and illumination intensity data sent by a plurality of robots in a factory; the determination module is configured to determine a strong light region of the factory according to the illumination intensity data, and determine a high temperature region of the factory according to the temperature distribution data; the output module is configured to take an overlapping region of the strong light region and the high temperature region as a perception blind area, and take a remaining region as a non-perception blind area; the evaluation module is configured to evaluate a perception ability level of each robot according to the metal dust concentration data, and the evaluation of the perception ability level of each robot according to the metal dust concentration data includes: determining a dust concentration of a position where each robot is located according to the metal dust concentration data; determining a perception ability level of each robot according to the dust concentration, wherein the perception ability level is inversely proportional to the dust concentration; and taking a robot whose perception ability level is greater than a preset level and located in the non-perception blind area as a collaborative robot; the combination module is configured to combine first environment perception data sent by the collaborative robot and second environment perception data sent by a robot in the perception blind area to determine a position and a motion state of an obstacle in the perception blind area; the generation module is configured to construct an obstacle avoidance map according to the position and the motion state of the obstacle, and plan an obstacle avoidance path for a goods robot to be crossed through the perception blind area according to the obstacle avoidance map.

8. An electronic device, comprising: The electronic device includes a processor, a memory, a user interface and a network interface, the memory is configured to store instructions, the user interface and the network interface are configured to communicate with other devices, and the processor is configured to execute the instructions stored in the memory to enable the electronic device to perform the method of any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer program stored in the memory can be loaded and executed by the processor to perform the method of any one of claims 1-6.

Images