A method and system for synchronous positioning and mapping based on millimeter-wave radar data
By slicing and interpolating millimeter-wave radar data to convert it into laserscan data and performing SLAM, the problems of high cost and susceptibility to weather interference in lidar SLAM are solved, achieving low-cost, high-precision synchronous positioning and mapping.
Patent Information
- Application Number
- CN202311079723.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-08-25
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-08-25
AI Technical Summary
Existing lidar SLAM technology is costly and susceptible to weather interference. Millimeter-wave radar SLAM with multi-sensor fusion is costly and has complex algorithms, making it difficult to effectively perform synchronous localization and mapping.
By slicing and interpolating millimeter-wave radar data to suppress noise and increase the frame rate, the millimeter-wave radar data is converted into laserscan data, and SLAM is performed using the open-source cartographer algorithm.
It achieves low-cost simultaneous localization and mapping with accuracy close to that of LiDAR SLAM, reduces the impact of noise, and improves the data frame rate.
Smart Images

Figure CN117092645B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of millimeter-wave radar SLAM technology, and in particular to a method and system for synchronous localization and mapping based on millimeter-wave radar data. Background Technology
[0002] Simultaneous localization and mapping (SLAM) is generally divided into LiDAR SLAM and visual SLAM. LiDAR SLAM is more mature and universal, with various open-source algorithms available, such as Google's Cartographer. Typical LiDAR SLAM involves a robot equipped with a LiDAR sensor that moves through the environment while simultaneously using the LiDAR to detect objects, then using the LiDAR data to create a topographic map of the entire environment. There are also methods that utilize multi-sensor data fusion for millimeter-wave radar SLAM.
[0003] However, lidar equipment is expensive and susceptible to interference from weather conditions such as rain, snow, and fog, resulting in a limited detection range. Millimeter-wave radar SLAM, which integrates multiple sensors, is also costly and requires complex algorithm development.
[0004] Therefore, a synchronous positioning and mapping method based on millimeter-wave radar data is needed. Summary of the Invention
[0005] In view of this, the purpose of the present invention is to provide a synchronous positioning and mapping method based on millimeter-wave radar data. This method uses slicing and interpolation processing of millimeter-wave radar data to suppress noise, enabling synchronous positioning and mapping of millimeter-wave radar data.
[0006] To achieve the above objectives, the present invention provides the following technical solution:
[0007] The synchronous positioning and mapping method based on millimeter-wave radar data provided by this invention includes the following steps:
[0008] First, acquire some millimeter-wave data; then convert the millimeter-wave data into laserscan data.
[0009] Secondly, increase the frame rate between adjacent millimeter-wave radar data frames;
[0010] Finally, the processed millimeter-wave radar data is used for synchronous positioning and mapping.
[0011] Furthermore, the specific steps for converting millimeter-wave data into laserscan data are as follows:
[0012] Obtain millimeter-wave radar model information;
[0013] Extract a frame of raw millimeter-wave radar data;
[0014] By analyzing millimeter-wave radar data, the sampling frequency, sampling azimuth angle, range values of each azimuth angle, and intensity values are obtained.
[0015] Using the obtained information, a corresponding frame of millimeter-wave radar laserscan data is created.
[0016] Determine if all millimeter-wave data frames have been retrieved. If not, return to retrieve the original millimeter-wave radar data frames; if yes, the laserscan data conversion is complete.
[0017] Furthermore, the process of converting millimeter-wave data into laserscan data also includes the following steps:
[0018] Set parameters for laserscan data: These parameters include minimum angle, maximum angle, or scan range;
[0019] The millimeter-wave radar data interval is divided into several smaller intervals; the minimum time division, minimum range value, and maximum range value are set according to the sampling frequency, and the original millimeter-wave data is converted into laserscan data.
[0020] Furthermore, the specific steps for increasing the frame rate between adjacent millimeter-wave radar data frames are as follows: slice the difference between adjacent frames of the millimeter-wave radar data, and then interpolate the inter-frame data obtained from the slices into the original data to increase the frame rate.
[0021] Furthermore, the frame rate between adjacent millimeter-wave radar data frames is determined through the following specific steps:
[0022] Extract two adjacent frames of millimeter-wave laserscan data, and designate the frame with the earlier timestamp as frame A and the frame with the later timestamp as frame B;
[0023] The inter-frame difference is obtained by subtracting the range and intensity values of all azimuth angles in frame B from the range and intensity values of the corresponding azimuth angles in frame A.
[0024] The number of segments is determined based on the sampling frequency in the millimeter-wave radar laserscan data, and the inter-frame differences are averaged and sliced.
[0025] The difference after segmentation is accumulated into frame A to obtain multiple new data frames. The new data frames are then interpolated between frames A and B; this process continues until all frames have been processed.
[0026] Furthermore, the process of performing SLAM on the processed millimeter-wave radar data is achieved by inputting the processed millimeter-wave radar data into the open-source Cartographer algorithm.
[0027] The SLAM system based on millimeter-wave radar data provided by the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the above-described method.
[0028] The beneficial effects of this invention are as follows:
[0029] This invention provides a synchronous localization and mapping method based on millimeter-wave radar data. This method uses only millimeter-wave radar data for SLAM. Compared with lidar data, millimeter-wave data has a lower frame rate and is more susceptible to noise. To address the noise characteristics of millimeter-wave radar data, the method slices and interpolates the data to suppress its impact before performing SLAM. Compared with traditional lidar SLAM, this method uses only millimeter-wave radar for SLAM, which can effectively reduce costs while maintaining accuracy.
[0030] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description
[0031] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the following figures are provided for illustration:
[0032] Figure 1 Flowchart for converting millimeter-wave data into LaserScan data.
[0033] Figure 2 This is a schematic diagram of millimeter-wave radar data before and after conversion.
[0034] Figure 3 This is a schematic diagram of the slicing and interpolation process.
[0035] Figure 4 This is a diagram comparing millimeter-wave radar data and lidar data.
[0036] Figure 5 This is a schematic diagram of inter-frame interpolation for millimeter-wave data.
[0037] Figure 6 This is a schematic diagram of the mapping results of processed millimeter-wave data using Google's open-source SLAM algorithm, Cartographer.
[0038] Figure 7 This provides a quantitative assessment of the similarity between millimeter-wave radar SLAM results and lidar SLAM results. Detailed Implementation
[0039] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.
[0040] This embodiment provides a synchronous localization and mapping method based on millimeter-wave radar data. SLAM primarily uses lidar data. However, millimeter-wave radar data, due to its significantly lower sampling frequency and fewer sampling points compared to lidar, cannot meet the requirements of SLAM algorithms. Limited research on millimeter-wave radar SLAM focuses on fusing millimeter-wave radar data with other sensor data, which increases both cost and algorithm development difficulty. The method provided in this embodiment, by processing millimeter-wave radar data, successfully utilizes the existing lidar SLAM framework to achieve synchronous localization and mapping based on millimeter-wave radar data, achieving accuracy close to that of traditional lidar SLAM. The main steps are as follows:
[0041] First, acquire some millimeter-wave data; then convert the millimeter-wave data into laserscan data.
[0042] The raw millimeter-wave data consists of several data frames, each composed of three values: timestamp, distance at each azimuth angle, and intensity at each azimuth angle. It cannot be directly used for SLAM and needs to be converted into laserscan data before further processing. The specific data conversion steps are as follows: Figure 1 As shown. Figure 2 This shows a comparison between a single frame of raw millimeter-wave data and a point cloud image formed by converting it into laserscan data and then visualizing it.
[0043] like Figure 1 As shown, the specific steps for converting millimeter-wave data into LaserScan data are as follows:
[0044] Obtain millimeter-wave radar model information;
[0045] Extract a frame of raw millimeter-wave radar data;
[0046] By analyzing millimeter-wave radar data, the sampling frequency, sampling azimuth angle, range values of each azimuth angle, and intensity values are obtained.
[0047] Using the obtained information, a corresponding frame of millimeter-wave radar laserscan data is created.
[0048] Determine if all millimeter-wave data frames have been retrieved. If not, return to retrieve the original millimeter-wave radar data frames; if yes, the laserscan data conversion is complete.
[0049] like Figure 2 As shown, Figure 2 This is a schematic diagram of millimeter-wave radar data before and after conversion, where... Figure 2 (a) in the image is a single frame of raw millimeter-wave radar data; (b) is a visualized single frame of millimeter-wave radar laserscan data.
[0050] This embodiment sets specific parameters for the laserscan data based on the characteristics of the millimeter-wave radar data itself: the minimum angle is set to 0, the maximum angle is set to 6.28, corresponding to a scanning range of 0-2π for the millimeter-wave radar. Simultaneously, since the millimeter-wave radar data divides the total 0-6.28 interval into 400 smaller intervals as a total of 400 azimuth angles, the minimum angle division is 0.0157. Based on the sampling frequency, the minimum time division is set to 0.00000625. Furthermore, to reduce the influence of some impulse noise, the minimum range value is set to 0, and the maximum range value is set to 200. Thus, the original millimeter-wave data is converted into laserscan data.
[0051] Secondly, the frame rate between adjacent millimeter-wave radar data frames is increased. This is achieved by slicing the difference between adjacent millimeter-wave radar data frames and then interpolating the sliced inter-frame data back into the original data. The specific steps are as follows: Take a pair of adjacent millimeter-wave data frames. First, calculate the difference between the two frames and then average the inter-frame difference. The number of slices depends on the sampling frequency of the source data. Then, accumulate the sliced differences from smallest to largest onto the frame with the earlier timestamp, forming multiple new data frames. Interpolate these new data frames between the original two adjacent data frames to simulate lidar data with a higher sampling frequency. Each pair of adjacent millimeter-wave data frames, after being processed in this way, can be used for SLAM.
[0052] like Figure 3 As shown, Figure 3 This diagram illustrates the slicing and interpolation process for two adjacent data frames. The specific steps of the slicing and interpolation process are as follows:
[0053] Extract two adjacent frames of millimeter-wave laserscan data, and designate the frame with the earlier timestamp as frame A and the frame with the later timestamp as frame B;
[0054] The inter-frame difference is obtained by subtracting the range and intensity values of all azimuth angles in frame B from the range and intensity values of the corresponding azimuth angles in frame A.
[0055] The number of segments is determined based on the sampling frequency in the millimeter-wave radar laserscan data, and the inter-frame differences are averaged and sliced.
[0056] The difference after segmentation is accumulated into frame A to obtain multiple new data frames. The new data frames are then interpolated between frames A and B; this process continues until all frames have been processed.
[0057] like Figure 4 As shown, compared to lidar data, millimeter-wave radar data is more susceptible to noise interference due to the longer wavelength of millimeter waves. Figure 4 (a) is a millimeter-wave radar point cloud image, and (b) is a lidar point cloud image. Figure 4 Point cloud data from both types of radar were visualized. The millimeter-wave radar data showed significantly more noise. Analysis of the millimeter-wave radar data revealed that non-noise points were relatively small in adjacent frames, roughly corresponding to the distance the radar itself had moved; while noise points exhibited significant and irregular positional changes between adjacent frames.
[0058] like Figure 5 As shown, this embodiment increases the frame rate by slicing the differences between adjacent frames of data and then interpolating the sliced inter-frame data back into the original data. Given the characteristics of millimeter-wave radar data mentioned above, increasing the frame rate enhances the influence of correct points and reduces the influence of noise.
[0059] Finally, the processed millimeter-wave radar data is subjected to SLAM, which involves inputting the processed millimeter-wave radar data into the open-source Cartographer algorithm, performing SLAM, and then outputting the final result.
[0060] like Figure 6 As shown, Figure 6 To demonstrate the results of SLAM on the open-source Cartographer algorithm for millimeter-wave radar data processed using the method in this embodiment.
[0061] As can be seen, SLAM using millimeter-wave radar reconstructed the outlines of buildings on both sides of the road, as well as the radar's travel path and trajectory, thus accurately reproducing the street conditions captured in the data.
[0062] like Figure 7 As shown, Figure 7 The figure shows the results of a quantitative evaluation of the similarity between the trajectories obtained by millimeter-wave radar SLAM and those obtained by lidar SLAM.
[0063] APE stands for Absolute Pose Error;
[0064] RPE stands for Relative Pose Error;
[0065] In both graphs, rmse, median, mean, and std represent global statistical data on the error, where:
[0066] rmse represents the root mean square error;
[0067] median represents the overall median of the error;
[0068] mean represents the overall average value of the error;
[0069] std represents the global standard deviation of the error;
[0070] Figure 7 In the figures, (a) represents the absolute pose error (APE), and (b) represents the relative pose error (RPE). Specifically, (a) represents the absolute pose error of both poses, with an average value of 0.2294, and (b) represents the relative pose error of both poses, with an average value of 0.3013. Quantitative evaluation results show that millimeter-wave radar SLAM has achieved accuracy very close to that of lidar SLAM, demonstrating the effectiveness of the method proposed in this embodiment.
[0071] This embodiment also provides a system that uses the above method to achieve its function, namely a SLAM system based on millimeter-wave radar data, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the above method.
[0072] The above-described embodiments are merely preferred embodiments provided to fully illustrate the present invention, and the scope of protection of the present invention is not limited thereto. Equivalent substitutions or modifications made by those skilled in the art based on the present invention are all within the scope of protection of the present invention. The scope of protection of the present invention is defined by the claims.
Claims
1. A synchronous positioning and mapping method based on millimeter-wave radar data, characterized in that: Includes the following steps: First, acquire some millimeter-wave data; then convert the millimeter-wave data into laserscan data, the specific steps are as follows: Obtain millimeter-wave radar model information; Extract a frame of raw millimeter-wave radar data; By analyzing millimeter-wave radar data, the sampling frequency, sampling azimuth angle, range values of each azimuth angle, and intensity values are obtained. Using the obtained information, a corresponding frame of millimeter-wave radar laserscan data is created. Determine if all millimeter-wave data frames have been retrieved. If not, return to retrieve the original millimeter-wave radar data frames; if yes, the laserscan data conversion is complete. Secondly, the frame rate between adjacent millimeter-wave radar data frames is increased. The steps are as follows: the difference between adjacent frames of the millimeter-wave radar data is sliced, and then the inter-frame data obtained from the slices is interpolated back into the original data to increase the frame rate. The specific steps are as follows: Extract two adjacent frames of millimeter-wave laserscan data, and designate the frame with the earlier timestamp as frame A and the frame with the later timestamp as frame B; The inter-frame difference is obtained by subtracting the range and intensity values of all azimuth angles in frame B from the range and intensity values of the corresponding azimuth angles in frame A. The number of segments is determined based on the sampling frequency in the millimeter-wave radar laserscan data, and the inter-frame differences are averaged and sliced. The difference after segmentation is accumulated into frame A to obtain multiple new data frames. The new data frames are then interpolated between frames A and B. This process continues until all frames have been processed. Finally, the processed millimeter-wave radar data is used for synchronous positioning and mapping.
2. The synchronous positioning and mapping method based on millimeter-wave radar data as described in claim 1, characterized in that: The process of converting millimeter-wave data into laserscan data also includes the following steps: Set parameters for laserscan data: These parameters include minimum angle, maximum angle, or scan range; The millimeter-wave radar data interval is divided into several smaller intervals; the minimum time division, minimum range value, and maximum range value are set according to the sampling frequency, and the original millimeter-wave data is converted into laserscan data.
3. The synchronous positioning and mapping method based on millimeter-wave radar data as described in claim 1, characterized in that: The process of performing SLAM on the processed millimeter-wave radar data involves inputting the processed millimeter-wave radar data into the open-source Cartographer algorithm.
4. A SLAM system based on millimeter-wave radar data, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method described in any one of claims 1 to 3.
Citation Information
Patent Citations
Adjusting frame rates for video applications
US8619198B1
