Aircraft autonomous navigation system of integrated semiconductor radar sensor

Abstract

The invention discloses an aircraft autonomous navigation system of an integrated semiconductor radar sensor, and relates to the technical field of aircrafts, and the system comprises a semiconductor radar integration module which is used for carrying out the multi-modal fusion of radar data obtained by the integrated semiconductor radar sensor and visual data obtained by a visual sensor, generating a real-time environment feature set; the real-time data processing module is used for constructing a three-dimensional map of the real-time environment according to the real-time environment feature set; the radar communication integrated module is used for realizing short-distance communication by multiplexing radar hardware and sharing the constructed three-dimensional map in real time; the dynamic path planning module is used for dynamically planning the flight path of the current aircraft according to the real-time map sharing data; and the semiconductor radar adaptive adjustment module is used for adaptively adjusting the scanning frequency of the semiconductor radar in the flight process. The positioning precision distance and speed of the aircraft are improved, and meanwhile the size and weight of a traditional navigation system are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of aircraft, and in particular to an aircraft autonomous navigation system with an integrated semiconductor radar sensor. Background Art

[0002] With the rapid development of unmanned aerial vehicles, autonomous driving aircraft (UAVs), and various types of aircraft, autonomous navigation technology has become one of the core elements for realizing intelligent flight. Traditional aircraft navigation systems mainly rely on technologies such as the Global Positioning System (GPS), Inertial Navigation System (INS), and vision sensors. However, the application of these technologies in complex environments has certain limitations. For example, GPS signals are easily blocked or interfered with indoors, in canyons, or in areas with dense urban high-rise buildings, resulting in a decrease in positioning accuracy or even failure; although the inertial navigation system does not rely on external signals, its errors will accumulate over time and it is difficult to maintain high accuracy for a long time; vision sensors are greatly affected by lighting and weather conditions, and their performance significantly decreases at night or in bad weather.

[0003] To solve the above problems, radar technology has gradually been introduced into aircraft navigation systems. Radar sensors have the ability to work all-weather, can penetrate harsh environments such as rain, fog, and dust, and provide stable distance, speed, and azimuth information. However, traditional mechanical radar sensors are large in size, heavy in weight, and high in power consumption, making it difficult to meet the requirements of small aircraft for lightweight and low power consumption. In addition, the data processing speed of traditional radar systems is relatively slow, making it difficult to meet the real-time requirements of high-speed aircraft.

[0004] Radar sensors based on semiconductor processes are not only small in size, light in weight, and low in power consumption, but also can achieve high-precision detection and fast data processing. Applying such an integrated semiconductor radar sensor to an aircraft autonomous navigation system can significantly improve the navigation ability of the aircraft in complex environments, while reducing the overall cost and power consumption of the system.

[0005] Therefore, the present invention proposes an aircraft autonomous navigation system based on an integrated semiconductor radar sensor, which can overcome the limitations of traditional navigation technologies, provide high-precision, high-reliability, all-weather autonomous navigation capabilities for aircraft, and promote the development of aircraft intelligent technologies. Summary of the Invention

[0006] The present invention provides an aircraft autonomous navigation system with an integrated semiconductor radar sensor, including:

[0007] A semiconductor radar integration module, a real-time data processing module, a radar communication integration module, a dynamic path planning module, and a semiconductor radar adaptive adjustment module;

[0008] A semiconductor radar integration module is used to perform multimodal fusion on the radar data obtained by an integrated semiconductor radar sensor and the visual data obtained by a visual sensor to generate a real-time environmental feature set;

[0009] A real-time data processing module is used to construct a three-dimensional map of the real-time environment based on the real-time environmental feature set;

[0010] A radar communication integration module is used to achieve short-distance communication by reusing radar hardware and share the constructed three-dimensional map in real time;

[0011] A dynamic path planning module is used to dynamically plan the flight path of the current aircraft according to the real-time map sharing data;

[0012] A semiconductor radar adaptive adjustment module is used to adaptively adjust the scanning frequency of the semiconductor radar during flight.

[0013] An aircraft autonomous navigation system with an integrated semiconductor radar sensor as described above, wherein the semiconductor radar integration module specifically includes the following sub-modules:

[0014] A sensing data receiving sub-module is used to receive in real time the environmental data detected by the radar sensor and the visual sensor;

[0015] A target feature extraction sub-module is used to quickly extract the radar features and visual features of each target in the environmental data;

[0016] A multimodal fusion sub-module is used to fuse the radar features and visual features of each target to generate a real-time environmental feature set.

[0017] An aircraft autonomous navigation system with an integrated semiconductor radar sensor as described above, wherein the fusion of radar features and visual features specifically includes the following sub-steps:

[0018] Align the radar features and visual features in space and time;

[0019] Fuse the radar features and visual features after space-time alignment into a joint feature vector;

[0020] Sort out the fused joint feature vector to generate a real-time environmental feature set.

[0021] An aircraft autonomous navigation system with an integrated semiconductor radar sensor as described above, wherein the real-time data processing module specifically includes the following sub-modules:

[0022] A target classification sub-module is used to sequentially input the real-time environmental features into a pre-trained target classification model and output the classification label of each target;

[0023] A three-dimensional map construction sub-module, which is used to map the classification labels of each target into the world coordinate system to generate a three-dimensional map of the real-time environment;

[0024] A target trajectory prediction sub-module, which is used to predict the motion trajectory of a dynamic target within a future time period according to the coordinate changes of the dynamic target, and mark the predicted motion trajectory on the three-dimensional map.

[0025] An aircraft autonomous navigation system with an integrated semiconductor radar sensor as described above, wherein the radar communication integration module specifically includes the following sub-modules:

[0026] A map compression sub-module, which is used to compress the size of the three-dimensional map and reduce the transmission bandwidth requirements;

[0027] A communication time slot allocation sub-module, which is used to adaptively adjust the communication time slot ratio according to the target density;

[0028] A map transmission sub-module, which is used to embed the compressed three-dimensional map into the radar signal and transmit it to the aircraft within the communication range.

[0029] An aircraft autonomous navigation system with an integrated semiconductor radar sensor as described above, wherein the dynamic path planning module specifically includes the following sub-modules:

[0030] A map fusion sub-module, which is used to fuse the received map data with its own map data to obtain more comprehensive environmental information;

[0031] A global path planning sub-module, which is used to search for the optimal flight path in the fused three-dimensional map;

[0032] A local path optimization sub-module, which is used to optimize the flight path of the aircraft when facing dynamic targets.

[0033] The present invention also provides an aircraft autonomous navigation method with an integrated semiconductor radar sensor, including:

[0034] Step S10: Perform multi-modal fusion on the radar data obtained by the integrated semiconductor radar sensor and the visual data obtained by the visual sensor to generate a real-time environmental feature set;

[0035] Step S20: Construct a three-dimensional map of the real-time environment according to the real-time environmental feature set;

[0036] Step S30: Implement short-distance communication by reusing the radar hardware to share the constructed three-dimensional map in real time;

[0037] Step S40: Dynamically plan the flight path of the current aircraft according to the real-time map sharing data;

[0038] Step S50: Adaptively adjust the scanning frequency of the semiconductor radar during flight.

[0039] The beneficial effects achieved by the present invention are as follows: It improves the positioning accuracy of the aircraft in terms of distance and speed, while reducing the volume and weight of the traditional navigation system, making it suitable for small aircraft; it can still work stably in bad weather, has strong adaptability, and is applicable to a variety of complex scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0041] Figure 1 It is a schematic diagram of an aircraft autonomous navigation system with an integrated semiconductor radar sensor provided in Embodiment 1 of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0042] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

[0043] Embodiment 1

[0044] As Figure 1 shown, Embodiment 1 of the present application provides an aircraft autonomous navigation system with an integrated semiconductor radar sensor, including: a semiconductor radar integration module 11, a real-time data processing module 12, a radar communication integration module 13, a dynamic path planning module 14, and a semiconductor radar adaptive adjustment module 15;

[0045] The semiconductor radar integration module 11 is used to perform multi-modal fusion on the radar data obtained by the integrated semiconductor radar sensor and the visual data obtained by the visual sensor to generate a real-time environmental feature set; specifically, it includes the following sub-modules:

[0046] 1. A sensing data receiving sub-module, which is used to receive the environmental data detected by the radar sensor and the visual sensor in real time;

[0047] The integrated semiconductor radar sensor emits and receives radar waves in the millimeter wave / terahertz frequency band, and supports multi-band collaborative detection, which can achieve higher-precision environmental detection. Coupled with the visual sensor, more comprehensive and complete environmental perception data of the current environment can be obtained.

[0048] 2. The target feature extraction sub-module is used to quickly extract the radar features and visual features of each target in the environmental data;

[0049] Set a pre-judgment. If no target is detected currently, that is, the number of targets included in the environmental data is 0, then return the status code "S00"; otherwise, return the status code "S01";

[0050] If the status code returned by the pre-judgment is "S01", then perform the feature extraction operation. If the status code returned is "S00", then continue to detect; among them, the extracted radar features include but are not limited to the distance feature, speed feature (radial speed, tangential speed, acceleration), geometric feature, azimuth feature (azimuth angle, pitch angle) and physical feature of the target; the extracted visual features include but are not limited to the geometric feature (size, shape), color feature and texture feature of the target.

[0051] 3. The multi-modal fusion sub-module is used to fuse the radar features and visual features of each target to generate a real-time environmental feature set;

[0052] The fusion of radar features and visual features specifically includes the following sub-steps:

[0053] ① Align the radar features and visual features in space and time;

[0054] Convert both the radar coordinate system and the visual coordinate system to the world coordinate system. Align the radar features and visual features in time according to the time stamp of the received sensing data, and align the radar features and visual features in space according to the world coordinates of the target, so that a set of radar-visual features only describes one target at one time stamp. It should be noted that in a complex environment, the visual sensor may lose the target. At this time, the radar features and visual features cannot be spatially aligned for this target, but the radar features at this time should still be retained to ensure the integrity of the detection result.

[0055] ② Fuse the radar features and visual features after space-time alignment into a joint feature vector;

[0056] Substitute the radar features and visual features after space-time alignment into the formula in turn:

[0057] to obtain the joint feature vector f of the j-th target at the current time stamp t tj , where α represents the joint weight of the visual features, A p represents the p-th visual feature of the j-th target at the current time stamp t, W p represents the interference matrix of the environmental factors on the p-th visual feature, R TAn environmental factor matrix representing the current working scenario, where p ranges from 1 to P, and P is the total number of time features, B q Represents the q-th radar feature of the j-th target at the current timestamp t, W q Represents the interference matrix of environmental factors on the q-th radar feature, where q ranges from 1 to Q, and Q is the total number of radar features.

[0058] ③ Organize the combined feature vectors after fusion to generate a real-time environmental feature set;

[0059] Organize the combined feature vectors of all targets at the current timestamp t to generate a real-time environmental feature set F t 。

[0060] The real-time data processing module 12 is used to construct a three-dimensional map of the real-time environment based on the real-time environmental feature set; specifically, it includes the following sub-modules:

[0061] 1. The target classification sub-module is used to sequentially input the real-time environmental features into a pre-trained target classification model and output the classification label of each target;

[0062] The target classification model is trained with the combined feature vectors of historical detected targets as input and the actual target classification labels as output. It can identify the type of target (trees, vehicles, buildings, etc.) based on the input combined feature vectors. The mathematical expression of this model is:

[0063] Where Y is the model output, X is the model input, c is any class label in the class set C, W c Is the weight vector corresponding to the class label c, b c Is the bias vector corresponding to class c.

[0064] 2. The three-dimensional map construction sub-module is used to map the classification label of each target into the world coordinate system to generate a three-dimensional map of the real-time environment;

[0065] Previously, the world coordinate system was used to describe the position of each target during feature fusion. Then, the classification labels generated here can directly inherit the coordinate information of their corresponding targets. Combining these classification labels, as well as the original radar features and visual features, the operation of mapping while detecting the real-time environment can be completed.

[0066] 3. The target trajectory prediction sub-module is used to predict the movement trajectory of a dynamic target within the next time period based on the coordinate changes of the dynamic target and mark the predicted movement trajectory on the three-dimensional map;

[0067] The prediction of the movement trajectory can be implemented using a pre-trained LSTM model or a Kalman filter framework, which is not restricted here. The predicted movement trajectory is represented as: {[(xt+τ ,y t+τ ,z t+τ ),τ]}, where (x t+τ ,y t+τ ,z t+τ ) represents the landing coordinates of the target at timestamp t+τ, that is, the trajectory point in the motion trajectory, t represents the current timestamp, and τ is an increment used to control the interval between trajectory points; adjacent trajectory points are connected to form a trajectory line, and the trajectory line is wrapped with a translucent elliptical sphere, whose transparency decreases as τ increases, to facilitate subsequent dynamic path planning.

[0068] The radar communication integration module 13 is used to realize short-distance communication by reusing radar hardware and share the constructed three-dimensional map in real time; it specifically includes the following submodules:

[0069] 1. Map compression submodule, used to compress the size of 3D maps and reduce transmission bandwidth requirements;

[0070] Octree coding is used for lossless compression in static areas of the map, while differential coding is used for lossy compression in dynamic areas, ensuring that the compressed map size is within 100MB to adapt to the transmission environment of radar hardware.

[0071] 2. Communication time slot allocation submodule, used to adaptively adjust the communication time slot ratio according to the target density;

[0072] Initially, the detection time slot accounts for 80%, the communication time slot accounts for 15%, and a 5% protection interval is reserved. When the target density is large, the proportion of the communication time slot is reduced, and the proportion of the detection time slot is increased accordingly. Conversely, the proportion of the communication time slot is increased, and the proportion of the detection time slot is reduced accordingly. This adaptive adjustment process is based on the formula: where γ comm is the proportion of communication time slots after adaptive adjustment, N current is the number of targets currently detected, N max is the maximum number of target detections by the radar, γ inco is the initial communication time slot ratio.

[0073] 3. Map transmission submodule, used to embed the compressed 3D map into the radar signal and transmit it to the aircraft within the communication distance;

[0074] According to the time slot ratio adjusted by the communication time slot allocation submodule, the current OFDM waveform structure is designed: Assuming the total bandwidth is L and the number of subcarriers is M, then M*γ commOne subcarrier is used for communication data transmission, and the remaining subcarriers are used for radar detection (round down when the calculation result is a decimal); the compressed 3D map data is a binary stream, which is modulated into complex symbols by 16QAM. Every 4 bits are mapped to a 16QAM symbol, and then the symbols are filled into the communication subcarriers in sequence. Aircraft within the communication range can share the constructed 3D map data with each other through radar signals.

[0075] The dynamic path planning module 14 is used to dynamically plan the flight path of the current aircraft according to real-time map sharing data; it specifically includes the following sub-modules:

[0076] 1. The map fusion sub-module is used to fuse the received map data with its own map data to obtain more comprehensive environmental information;

[0077] First, decompress the received map data to obtain the original 3D map, then align the received map and its own map in terms of timestamp and coordinates, and then compare the target differences between the two maps, including the differences in classification labels and target coordinates. Then start to judge: if a target exists in the received map but does not exist in its own map, update the target to its own map; if the target does not exist in the received map but exists in its own map, keep it unchanged; if both maps exist but there are differences between the targets, use the weighted summation method for fusion; if both maps exist and there are no differences between the targets, keep it unchanged.

[0078] 2. The global path planning sub-module is used to search for the optimal flight path in the fused 3D map;

[0079] The path planning algorithm can use the A* algorithm, or the RRT* algorithm or other query algorithms, which are not restricted here. The planned path is represented by a set of continuous navigation point coordinates + timestamp.

[0080] 3. The local path optimization sub-module is used to optimize the flight path of the aircraft when facing dynamic targets;

[0081] If there are dynamic targets on the planned path, it is necessary to locally optimize this part of the path to ensure that the aircraft can pass through this section safely and quickly. Design a local optimization function and introduce it into the path planning algorithm, loop to plan the local path, calculate the return value of the local optimization function, and finally select the path when the return value of the local optimization function is the largest as the optimal passing path for this section. The local optimization function is expressed as:

[0082] where E is the return value of the local optimization function, and are the optimization weights for target alignment and separation distance respectively, v traj(i)is the direction vector of the i-th navigation point in the local path, v goal is the direction vector of the dynamic target at the timestamp of the i-th navigation point, (x i ,y i , z i ) is the coordinate of the i-th navigation point in the local path, (x g ,y g , z g ) is the coordinate of the dynamic target at the timestamp of the i-th navigation point, i ranges from 1 to n, n is the total number of navigation points in the local path, and it should be noted that ||v traj(i) || Navigation points with a value of 0 do not participate in the calculation, ||v goal When || is 0, the function Remove.

[0083] In addition, when doing path planning, for dynamic targets, the entire ellipsoid surrounding the trajectory line needs to be regarded as an obstacle to reduce the complexity of the algorithm and speed up the response.

[0084] A semiconductor radar adaptive adjustment module 15, used for adaptively adjusting the scanning frequency of the semiconductor radar during flight;

[0085] The adaptive control of the semiconductor radar scanning frequency is realized by a frequency modulation function, which can increase the scanning frequency when the target is dense in the environment to obtain more precise detection data, and reduce the scanning frequency when the environment is empty to reduce power consumption and increase the endurance of the aircraft. Its mathematical expression is:

[0086] Where H is the adjusted scanning frequency, h max represents the maximum scanning frequency supported by the semiconductor radar, k is the response coefficient, N current Indicates the number of targets detected in the current environment, N threshold is the preset maximum target density threshold.

[0087] Embodiment 2

[0088] Embodiment 2 of the present application provides an autonomous navigation method for an aircraft using an integrated semiconductor radar sensor, comprising:

[0089] Step S10: performing multimodal fusion of radar data acquired by the integrated semiconductor radar sensor and visual data acquired by the visual sensor to generate a real-time environmental feature set; specifically comprising the following sub-steps:

[0090] Step S11: receiving environmental data detected by the radar sensor and the visual sensor in real time;

[0091] The integrated semiconductor radar sensor transmits and receives radar waves in the millimeter-wave / terahertz frequency band, supports multi-band collaborative detection simultaneously, can achieve higher-precision environmental detection, and can obtain more comprehensive and complete environmental perception data of the current environment when combined with a vision sensor.

[0092] Step S12: Quickly extract the radar features and visual features of each target in the environmental data;

[0093] Set a pre-judgment. If no target is detected currently, that is, the number of targets included in the environmental data is 0, then return the status code "S00"; otherwise, return the status code "S01";

[0094] If the status code returned by the pre-judgment is "S01", then perform the feature extraction operation. If the returned status code is "S00", then continue the detection; among them, the extracted radar features include but are not limited to the distance feature, speed feature (radial speed, tangential speed, acceleration), geometric feature, azimuth feature (azimuth angle, elevation angle), and physical feature of the target; the extracted visual features include but are not limited to the geometric feature (size, shape), color feature, and texture feature of the target.

[0095] Step S13: Fuse the radar features and visual features of each target to generate a real-time environmental feature set;

[0096] The fusion of radar features and visual features specifically includes the following sub-steps:

[0097] ① Align the radar features and visual features in space and time;

[0098] Convert both the radar coordinate system and the visual coordinate system to the world coordinate system. Align the radar features and visual features in time according to the time stamp of the received sensing data, and align the radar features and visual features in space according to the world coordinates of the target, so that a set of radar-visual features only describes one target at one time stamp. It should be noted that in a complex environment, the vision sensor may lose the target. At this time, the radar features and visual features cannot be spatially aligned for this target, but the radar features at this time still need to be retained to ensure the integrity of the detection result.

[0099] ② Fuse the radar features and visual features after space-time alignment into a joint feature vector;

[0100] Substitute the radar features and visual features after space-time alignment into the formula:

[0101] to obtain the joint feature vector f of the j-th target at the current time stamp t tj , where α represents the joint weight of the visual features, A p represents the p-th visual feature of the j-th target at the current time stamp t, Wp The interference matrix of environmental factors on the p-th visual feature, R T The environmental factor matrix of the current working scenario, where p takes values from 1 to P, and P is the total number of temporal features, B q The q-th radar feature of the j-th target at the current timestamp t, W q The interference matrix of environmental factors on the q-th radar feature, where q takes values from 1 to Q, and Q is the total number of radar features.

[0102] ③ Organize the fused joint feature vectors to generate a real-time environmental feature set;

[0103] Organize the joint feature vectors of all targets at the current timestamp t to generate a real-time environmental feature set F t 。

[0104] Step S20: Construct a three-dimensional map of the real-time environment based on the real-time environmental feature set; specifically including the following sub-steps:

[0105] Step S21: Input the real-time environmental features into the pre-trained target classification model in sequence, and output the classification label of each target;

[0106] The target classification model is trained with the joint feature vectors of historical detected targets as input and the actual target classification labels as output. It can identify the type of target (trees, vehicles, buildings, etc.) according to the input joint feature vectors. The mathematical expression of this model is:

[0107] where Y is the model output, X is the model input, c is any class label in the class set C, W c is the weight vector corresponding to the class label c, b c is the bias vector corresponding to the class c.

[0108] Step S22: Map the classification label of each target to the world coordinate system to generate a three-dimensional map of the real-time environment;

[0109] Previously, the world coordinate system has been used to describe the position of each target during feature fusion. Then the classification labels generated here can directly inherit the coordinate information of their corresponding targets. Combining these classification labels, as well as the original radar features and visual features, the operation of mapping while detecting the real-time environment can be completed.

[0110] Step S23: According to the coordinate changes of dynamic targets, predict their movement trajectories within the next time period, and mark the predicted movement trajectories on the three-dimensional map;

[0111] The prediction of motion trajectory can be achieved using a pre-trained LSTM model or Kalman filter framework. There is no restriction here. The predicted motion trajectory is expressed as: {[(x t+τ ,y t+τ ,z t+τ ),τ]}, where (x t+τ ,y t+τ ,z t+τ ) represents the landing coordinates of the target at timestamp t+τ, that is, the trajectory point in the motion trajectory, t represents the current timestamp, and τ is an increment used to control the interval between trajectory points; adjacent trajectory points are connected to form a trajectory line, and the trajectory line is wrapped with a translucent elliptical sphere, whose transparency decreases as τ increases, to facilitate subsequent dynamic path planning.

[0112] Step S30: short-distance communication is achieved by multiplexing radar hardware, and the constructed three-dimensional map is shared in real time; specifically, the following sub-steps are included:

[0113] Step S31: compressing the size of the three-dimensional map to reduce the transmission bandwidth requirement;

[0114] Octree coding is used for lossless compression in static areas of the map, while differential coding is used for lossy compression in dynamic areas, ensuring that the compressed map size is within 100MB to adapt to the transmission environment of radar hardware.

[0115] Step S32: adaptively adjusting the communication time slot ratio according to the target density;

[0116] Initially, the detection time slot accounts for 80%, the communication time slot accounts for 15%, and a 5% protection interval is reserved. When the target density is large, the proportion of the communication time slot is reduced, and the proportion of the detection time slot is increased accordingly. Conversely, the proportion of the communication time slot is increased, and the proportion of the detection time slot is reduced accordingly. This adaptive adjustment process is based on the formula: where γ comm is the proportion of communication time slots after adaptive adjustment, N current is the number of targets currently detected, N max is the maximum number of target detections by the radar, γ inco is the initial communication time slot ratio.

[0117] Step S33: embedding the compressed three-dimensional map into the radar signal and transmitting it to the aircraft within the communication distance;

[0118] According to the time slot ratio adjusted by the communication time slot allocation submodule, the current OFDM waveform structure is designed: Assuming the total bandwidth is L and the number of subcarriers is M, then M*γ commOne sub - carrier is used for communication data transmission, and the remaining sub - carriers are used for radar detection (round down when the calculation result is a decimal); the compressed 3D map data is a binary stream, which is modulated into complex symbols by 16QAM. Every 4 bits are mapped to a 16QAM symbol, and then the symbols are filled into the communication sub - carriers in sequence. Aircraft within the communication range can share the constructed 3D map data with each other through radar signals.

[0119] Step S40: Dynamically plan the flight path of the current aircraft according to the real - time map sharing data; specifically including the following sub - steps:

[0120] Step S41: Integrate the received map data with its own map data to obtain more comprehensive environmental information;

[0121] First, decompress the received map data to obtain the original 3D map, then align the received map and its own map in terms of time stamp and coordinates, and then compare the target differences between the two maps, including the differences in classification labels and target coordinates. Then start to judge: if a target exists in the received map but does not exist in its own map, update the target to its own map; if the target does not exist in the received map but exists in its own map, keep it unchanged; if both maps exist but there are differences between the targets, use the weighted sum method for integration; if both maps exist and there are no differences between the targets, keep it unchanged.

[0122] Step S42: Search for the optimal flight path in the integrated 3D map;

[0123] The path planning algorithm can use the A* algorithm, or the RRT* algorithm or other query algorithms, which is not restricted here. The planned path is represented by a set of continuous navigation point coordinates + time stamp.

[0124] Step S43: Optimize the flight path of the aircraft when facing dynamic targets;

[0125] If there are dynamic targets on the planned path, it is necessary to locally optimize this part of the path to ensure that the aircraft can pass through this section safely and quickly. Design a local optimization function and introduce it into the path planning algorithm. Continuously plan the local path, calculate the return value of the local optimization function, and finally select the path when the return value of the local optimization function is the largest as the optimal passing path for this section. The local optimization function is expressed as:

[0126] where E is the return value of the local optimization function, and are the optimization weights for target alignment and separation distance respectively, v traj(i) is the direction vector of the i - th navigation point in the local path, v goalis the direction vector of the dynamic target at the timestamp of the i-th navigation point, (x i ,y i , z i ) is the coordinate of the i-th navigation point in the local path, (x g ,y g , z g ) is the coordinate of the dynamic target at the timestamp of the i-th navigation point, i ranges from 1 to n, n is the total number of navigation points in the local path, and it should be noted that ||v traj(i) || Navigation points with a value of 0 do not participate in the calculation, ||v goal When || is 0, the function Remove.

[0127] In addition, when doing path planning, for dynamic targets, the entire ellipsoid surrounding the trajectory line needs to be regarded as an obstacle to reduce the complexity of the algorithm and speed up the response.

[0128] Step S50: adaptively adjusting the scanning frequency of the semiconductor radar during flight;

[0129] The adaptive control of the semiconductor radar scanning frequency is realized by a frequency modulation function, which can increase the scanning frequency when the target is dense in the environment to obtain more precise detection data, and reduce the scanning frequency when the environment is empty to reduce power consumption and increase the endurance of the aircraft. Its mathematical expression is:

[0130] Where H is the adjusted scanning frequency, h max represents the maximum scanning frequency supported by the semiconductor radar, k is the response coefficient, N current Indicates the number of targets detected in the current environment, N threshold is the preset maximum target density threshold.

[0131] Corresponding to the above embodiment, an embodiment of the present invention provides a computer storage medium, including: at least one memory and at least one processor;

[0132] The memory is used to store one or more program instructions;

[0133] A processor is used to run one or more program instructions to execute an autonomous navigation method for an aircraft with an integrated semiconductor radar sensor.

[0134] Corresponding to the above-mentioned embodiment, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer storage medium contains one or more program instructions, and the one or more program instructions are used by a processor to execute an autonomous navigation method for an aircraft with an integrated semiconductor radar sensor.

[0135] An embodiment disclosed by the present invention provides a computer-readable storage medium, in which computer program instructions are stored. When the computer program instructions run on a computer, the computer is caused to execute the above-mentioned autonomous navigation method of an integrated semiconductor radar sensor for an aircraft.

[0136] In an embodiment of the present invention, the processor may be an integrated circuit chip with the ability to process signals. The processor may be a general-purpose processor, a digital signal processor (DSP for short), an application specific integrated circuit (ASIC for short), a field programmable gate array (FPGA for short), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0137] It can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in combination with the embodiments of the present invention can be directly embodied as being executed by a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software module may be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The processor reads the information in the storage medium and combines its hardware to complete the steps of the above method.

[0138] The storage medium may be a memory, for example, it may be a volatile memory or a non-volatile memory, or may include both a volatile memory and a non-volatile memory.

[0139] Among them, the non-volatile memory may be a read-only memory (ROM for short), a programmable read-only memory (PROM for short), an erasable programmable read-only memory (EPROM for short), an electrically erasable programmable read-only memory (EEPROM for short), or a flash memory.

[0140] The volatile memory may be a Random Access Memory (RAM) which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM).

[0141] The storage media described in the embodiments of the present invention are intended to include but not limited to these and any other suitable types of memories.

[0142] Those skilled in the art should be aware that in one or more of the above examples, the functions described in the present invention can be implemented by a combination of hardware and software. When applying software, the corresponding functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. The computer-readable medium includes computer storage media and communication media, where the communication media includes any medium that facilitates the transfer of a computer program from one place to another. The storage media can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0143] The specific embodiments described above further elaborate on the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above description is only the specific embodiments of the present invention and is not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made on the basis of the technical solutions of the present invention should be included in the protection scope of the present invention.

Claims

1. An aircraft autonomous navigation system with an integrated semiconductor radar sensor, characterized in that: include: Semiconductor radar integration module, real-time data processing module, radar communication integration module, dynamic path planning module, semiconductor radar adaptive adjustment module; A semiconductor radar integration module is used to perform multimodal fusion of radar data acquired by an integrated semiconductor radar sensor and visual data acquired by a visual sensor to generate a real-time environmental feature set; A real-time data processing module is used to construct a three-dimensional map of the real-time environment based on the real-time environmental feature set; Radar communication integration module, used to achieve short-distance communication by reusing radar hardware and share the constructed 3D map in real time; Dynamic path planning module, used to dynamically plan the flight path of the current aircraft based on real-time map sharing data; The semiconductor radar adaptive adjustment module is used to adaptively adjust the scanning frequency of the semiconductor radar during flight.

2. The aircraft autonomous navigation system of an integrated semiconductor radar sensor according to claim 1, characterized in that: Semiconductor radar integrated module Includes the following submodules: The sensor data receiving submodule is used to receive the environmental data detected by the radar sensor and the visual sensor in real time; The target feature extraction submodule is used to quickly extract the radar features and visual features of each target in the environmental data; The multimodal fusion submodule is used to fuse the radar features and visual features of each target to generate a real-time environmental feature set.

3. The aircraft autonomous navigation system of an integrated semiconductor radar sensor according to claim 2, characterized in that: The fusion of radar features and visual features specifically includes the following sub-steps: Spatiotemporal alignment of radar signatures with visual signatures; The spatiotemporally aligned radar features and visual features are fused into a joint feature vector; The fused joint feature vectors are sorted and merged to generate a real-time environmental feature set.

4. The aircraft autonomous navigation system of an integrated semiconductor radar sensor according to claim 1, characterized in that: The real-time data processing module specifically includes the following sub-modules: The target classification submodule is used to input the real-time environmental features into the pre-trained target classification model in sequence and output the classification label of each target; The 3D map construction submodule is used to map the classification label of each target into the world coordinate system to generate a 3D map of the real-time environment; The target trajectory prediction submodule is used to predict the motion trajectory of the dynamic target within a future time period according to the coordinate changes of the dynamic target, and mark the predicted motion trajectory in the three-dimensional map.

5. The aircraft autonomous navigation system of an integrated semiconductor radar sensor according to claim 1, characterized in that: Radar communication integrated module Includes the following submodules: The map compression submodule is used to compress the size of the 3D map and reduce the transmission bandwidth requirement; The communication time slot allocation submodule is used to adaptively adjust the communication time slot ratio according to the target density; The map transmission submodule is used to embed the compressed three-dimensional map into the radar signal and transmit it to the aircraft within the communication distance.

6. The aircraft autonomous navigation system of an integrated semiconductor radar sensor according to claim 1, characterized in that: The dynamic path planning module specifically includes the following sub-modules: The map fusion submodule is used to fuse the received map data with its own map data to obtain more comprehensive environmental information; The global path planning submodule is used to search for the optimal flight path in the fused 3D map; The local path optimization submodule is used to optimize the flight path of the aircraft when facing a dynamic target.

7. An autonomous navigation method for an aircraft using an integrated semiconductor radar sensor, characterized in that: include: Step S10: performing multimodal fusion of radar data acquired by the integrated semiconductor radar sensor and visual data acquired by the visual sensor to generate a real-time environmental feature set; Step S20: constructing a three-dimensional map of the real-time environment according to the real-time environment feature set; Step S30: short-distance communication is achieved by multiplexing radar hardware, and the constructed three-dimensional map is shared in real time; Step S40: dynamically planning the flight path of the current aircraft according to the real-time map sharing data; Step S50: adaptively adjusting the scanning frequency of the semiconductor radar during flight.

Images