Small unmanned aerial vehicle collision avoidance method, system, device, and medium with integrated terrain awareness capability

Abstract

The application discloses a small unmanned aerial vehicle anti-collision method and system with integrated terrain perception capability, equipment and a medium, relates to the small unmanned aerial vehicle anti-collision technical field, and first establishes a unified target database containing an aerial target, a terrain obstacle and a local state; then, potential conflict time and risk levels between each target and the local target are predicted respectively, and threat sorting is completed according to a unified risk assessment mechanism; finally, horizontal and vertical candidate collision avoidance actions are jointly evaluated in combination with unmanned aerial vehicle kinematics constraints, coordinated horizontal and vertical collision avoidance instructions are generated, and are output to a flight control system for execution. The application completes unified reception of heterogeneous information at a data input layer, realizes integrated modeling and cooperative processing of multi-source threats at a decision layer, fundamentally eliminates information barriers between different threat sources, and provides a unified data basis for joint collision avoidance decision.

Description

Technical Field

[0001] This invention relates to the field of collision avoidance technology for small unmanned aerial vehicles (UAVs), specifically to collision avoidance methods, systems, devices, and media for small UAVs with integrated terrain perception capabilities. Background Technology

[0002] With the continuous expansion of low-altitude operation scenarios for drones, their flight activities in complex low-altitude environments such as mountainous areas and urban canyons are becoming increasingly frequent, raising growing concerns about flight safety. During autonomous drone flight, collision avoidance systems are one of the key technologies for ensuring flight safety.

[0003] Current drone collision avoidance technologies primarily focus on detecting and avoiding aerial conflicts between the drone and surrounding aircraft. This involves using onboard sensors to detect nearby aerial targets and generating collision avoidance maneuvers based on relative motion relationships to achieve air-to-air collision avoidance. However, in complex low-altitude operating environments, considering only aerial target conflicts is insufficient to fully meet flight safety requirements. Terrain and static obstacles constrain the drone's flight maneuverability, becoming a significant factor affecting low-altitude flight safety. For example, in mountainous or canyon environments, even if the drone currently poses no risk of collision with other aircraft, if the canyon slope is too steep or the space too narrow, the drone may be unable to safely leave the current area due to insufficient climb or limited turning radius, thus entering a dangerous "entry but no exit" state. Similarly, in urban low-altitude environments, although there may be gaps between multiple tall buildings, if the distance between buildings is less than the minimum space required for the drone to complete a turning maneuver, the drone will also be unable to pass safely, posing a collision risk. Therefore, there is an urgent need for a low-altitude fusion collision avoidance method that can uniformly consider aerial targets, terrain and fixed obstacles, to achieve unified modeling, unified assessment and cooperative collision avoidance of multi-source threats, so as to improve the autonomous collision avoidance capability and flight safety of UAVs in complex low-altitude environments.

[0004] Existing UAV collision avoidance systems suffer from the following shortcomings in handling aerial targets and terrain obstacles: Firstly, they typically treat aerial targets and terrain obstacles separately, employing different target representation methods and risk assessment models. This lack of a unified target modeling mechanism and threat assessment mechanism prevents different types of threat targets from being represented, prioritized, and processed within a unified framework. Secondly, current technologies lack the ability to make joint collision avoidance decisions for aerial targets and terrain obstacles. When aerial and ground threats coexist, existing systems cannot synthesize the risk levels of different targets to generate coordinated collision avoidance actions. Airborne collision avoidance commands and terrain warning commands may be independent or even conflicting, reducing the effectiveness and reliability of collision avoidance decisions. Thirdly, traditional terrain perception systems typically only provide hazard warnings and cannot directly generate specific collision avoidance maneuvers. Terrain obstacles are not directly incorporated into the collision avoidance decision-making process, resulting in the inability to output effective horizontal and vertical collision avoidance commands, thus limiting the autonomous collision avoidance capabilities of UAVs in complex low-altitude environments.

[0005] In summary, to address the issues of separate processing of aerial targets and terrain obstacles, lack of unified threat modeling, and joint collision avoidance decision-making mechanisms in existing low-altitude UAV collision avoidance systems, it is necessary to propose a small UAV collision avoidance system integrating terrain perception capabilities. This system would enable unified target modeling and risk assessment of aerial targets, terrain, and obstacles; joint threat assessment and coordinated collision avoidance decisions for aerial targets and terrain obstacles; and direct integration of terrain obstacles into the collision avoidance decision-making process to output horizontal and vertical collision avoidance commands. This would improve the autonomous collision avoidance capability and flight safety of UAVs in complex low-altitude environments. Summary of the Invention

[0006] Based on the problems raised in the background technology above, the purpose of this invention is to provide a collision avoidance method, system, device and medium for small UAVs with integrated terrain perception capabilities, which solves the problems of separate processing of aerial targets and terrain obstacles, lack of unified threat modeling and joint collision avoidance decision-making mechanism in existing UAV low-altitude collision avoidance systems.

[0007] This invention is achieved through the following technical solution:

[0008] The first aspect of this invention provides a collision avoidance method for small unmanned aerial vehicles (UAVs) with integrated terrain perception capabilities, comprising the following steps:

[0009] The UAV receives multi-source environmental information within its surveillance range, including aerial target information, its own status information, and terrain and obstacle environmental information.

[0010] A unified target model is constructed for the targets corresponding to the aerial target information, the terrain targets corresponding to the terrain and obstacle environment information, and the fixed obstacle targets. Each target in the unified target set is represented by a unified threat description attribute.

[0011] A unified threat assessment is performed on each target in the unified target set to predict the potential conflict time between each target and the local machine and determine the threat level. The targets are then sorted according to their threat levels to form a threat target set.

[0012] Based on the set of threat targets, a joint collision avoidance decision is made, and the collision avoidance effects of horizontal and vertical maneuvers are evaluated simultaneously to generate coordinated horizontal and vertical collision avoidance commands.

[0013] In the above technical solution, a unified target database is first established, including aerial targets, terrain obstacles, and the drone's status. Then, the potential conflict time and risk level between each target and the drone are predicted, and threats are ranked according to a unified risk assessment mechanism. Finally, combined with the drone's kinematic constraints, horizontal and vertical candidate collision avoidance maneuvers are jointly evaluated to generate coordinated horizontal and vertical collision avoidance commands, which are then output to the flight control system for execution. Compared to the existing decentralized architecture that establishes separate databases for aerial targets and terrain obstacles and conducts separate evaluations and decisions, this solution achieves unified reception of heterogeneous information at the data input layer and realizes integrated modeling and collaborative processing of multi-source threats at the decision layer. This fundamentally eliminates information barriers between different threat sources and provides a unified data foundation for joint collision avoidance decision-making.

[0014] In an optional embodiment, prior to the unified target modeling, target monitoring and environmental information processing are further included, the target monitoring and environmental information processing comprising:

[0015] Receive multi-source environmental information, including ADS-B target information, V2V target information, non-cooperative target information, local navigation status information, and terrain and obstacle environment information;

[0016] Perform data validity checks on the input data;

[0017] Update the target's status, establish the target's trajectory, and continuously maintain the target's status;

[0018] Multi-source targets are fused and associated based on their address, spatial location, velocity, and temporal consistency.

[0019] In one alternative embodiment, the unified target modeling includes:

[0020] Based on the results of environmental information processing, targets within the monitoring range are classified into aerial targets, terrain targets, fixed obstacle targets, and local targets.

[0021] Different types of targets are represented in a unified manner. Among them, aerial targets retain dynamic information, terrain targets and fixed obstacle targets are described using static target models, and local targets are represented using real-time updated status information.

[0022] Establish a unified threat description attribute for all targets, which includes target number, target type, spatial status, security protection boundary, prediction model, and target effective status;

[0023] Organize all objectives into a unified set of objectives.

[0024] In one optional embodiment, the unified threat assessment includes:

[0025] The state prediction is performed on each target in the unified target set. For aerial targets, the future trajectory is predicted using a motion prediction model, and for terrain and fixed obstacle targets, the relative motion relationship between the machine and the target is predicted.

[0026] The target conflict time is calculated based on the predicted target status. For aerial targets, the estimated time to enter the safety protection zone is calculated based on relative position, relative speed and motion trend. For terrain and fixed obstacles, the estimated collision time is calculated based on the flight trajectory of the aircraft and the spatial position of the obstacle.

[0027] The threat level is calculated by considering the time of target conflict, security protection boundary, target movement trend, and target effective status.

[0028] All targets are sorted according to their threat level, and the N targets with the highest threat level are selected to form a threat target set.

[0029] In one optional embodiment, the unified target set includes ground obstacle targets; wherein, the target conflict time calculation for the ground obstacle targets includes:

[0030] When the drone's current altitude is lower than the height of the obstacle plus the safety separation boundary, it is determined that it has entered the danger zone, and the conflict time is set to the minimum value.

[0031] When the drone's vertical speed approaches zero, it is determined that it has no maneuverability in the vertical direction, and the conflict time is set to the maximum value.

[0032] In other cases, a one-dimensional uniform motion model is used to estimate the time to reach the safe separation boundary, and boundary constraint processing is performed based on the estimation results;

[0033] Based on the time of the conflict and the current vertical speed of the drone, the equivalent height of the obstacle is forward-corrected.

[0034] In one optional embodiment, the unified target set includes point-like obstacle targets; wherein, the target conflict time calculation for the point-like obstacle targets includes:

[0035] Based on the current direction of movement of the drone, the distance to the obstacle is projected forward, and the projected distance of the obstacle in the current direction of movement is calculated. When the projected distance is greater than the current distance, it is determined that the drone is moving away from the obstacle.

[0036] When the drone has entered the target's safe interval, the conflict time is set to the minimum value;

[0037] Set the conflict time to the maximum value when the drone is moving away from the obstacle;

[0038] In other cases, the time to reach the target safety interval is estimated based on the current horizontal approach velocity, and boundary constraints are applied;

[0039] Based on the time of the conflict and the current vertical speed of the drone, the equivalent height of the obstacle is forward-corrected.

[0040] In one optional embodiment, the joint collision avoidance decision includes:

[0041] A set of joint collision avoidance candidate actions is generated based on the set of threat targets and the current flight status of the aircraft. The candidate actions include a combination of horizontal and vertical actions. The horizontal actions include maintaining the heading, turning left and turning right, and the vertical actions include maintaining altitude, climbing and descending.

[0042] For each joint candidate action, the future flight trajectory of the machine is predicted with a fixed time step, while all threat targets update their future states and the relative positional relationship between the machine and each threat target in the future prediction time domain is calculated.

[0043] For each joint candidate action, a joint cost assessment is performed in the prediction time domain, and a comprehensive cost value is formed by combining the changes in the safety intervals of each threat target, the maneuver range of the aircraft, the continuity of the action, and the satisfaction of flight constraints.

[0044] The joint candidate action that minimizes the overall cost and satisfies the flight constraints is selected as the final collision avoidance strategy, and horizontal and vertical collision avoidance commands are generated.

[0045] A second aspect of the present invention provides a collision avoidance system for small unmanned aerial vehicles (UAVs) with integrated terrain perception capabilities, comprising:

[0046] The target monitoring and environmental information processing module is used to receive multi-source environmental information and perform data validity detection, target status update and multi-source target fusion.

[0047] The unified target modeling module is used to convert aerial targets, terrain targets, and fixed obstacle targets into threat targets in a unified manner, and to establish a unified target set. Each target in the unified target set is represented by a unified threat description attribute.

[0048] The unified threat assessment module is used to perform status prediction, conflict time calculation, threat quantification, and threat ranking for each target in the unified target set, forming a threat target set.

[0049] The joint collision avoidance decision module is used to make joint collision avoidance decisions based on the set of threat targets, simultaneously evaluate the collision avoidance effects of horizontal and vertical maneuvers, and generate coordinated horizontal and vertical collision avoidance commands.

[0050] A third aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a collision avoidance method for a small unmanned aerial vehicle with integrated terrain perception capabilities.

[0051] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a collision avoidance method for small unmanned aerial vehicles with integrated terrain perception capabilities.

[0052] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0053] 1. Aerial targets, terrain targets, and fixed obstacles are uniformly constructed as threat targets, and described using a unified data structure and security protection boundary, thereby achieving unified management and processing of targets from different sources;

[0054] 2. Establish corresponding conflict time prediction models for dynamic aerial targets and static terrain obstacles respectively, and quantify and rank the risks of different types of targets through a unified threat index, so that various targets can be compared under a unified evaluation scale, thereby improving the accuracy and real-time performance of multi-source threat identification in complex low-altitude environments.

[0055] 3. In the candidate avoidance action set, both vertical maneuvering and horizontal steering strategies are considered simultaneously. By calculating and comparing the comprehensive risk costs of different actions, the optimal collision avoidance decision is achieved through multi-dimensional collaborative optimization. Attached Figure Description

[0056] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0057] Figure 1 A flowchart illustrating a collision avoidance method for small unmanned aerial vehicles (UAVs) with integrated terrain perception capabilities, provided in an embodiment of the present invention.

[0058] Figure 2 A schematic diagram illustrating the process of unified target modeling provided in an embodiment of the present invention;

[0059] Figure 3 A schematic diagram of the unified threat assessment process provided in this embodiment of the invention;

[0060] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0062] Example 1

[0063] Figure 1 This is a flowchart illustrating the collision avoidance method for small unmanned aerial vehicles (UAVs) with integrated terrain perception capabilities provided in Embodiment 1 of the present invention. Figure 1 As shown, the collision avoidance method for small drones with integrated terrain perception capabilities includes the following steps:

[0064] The UAV receives multi-source environmental information within its surveillance range, including aerial target information, its own status information, and terrain and obstacle environmental information.

[0065] A unified target model is constructed for the targets corresponding to the aerial target information, the terrain targets corresponding to the terrain and obstacle environment information, and the fixed obstacle targets. Each target in the unified target set is represented by a unified threat description attribute.

[0066] A unified threat assessment is performed on each target in the unified target set to predict the potential conflict time between each target and the local machine and determine the threat level. The targets are then sorted according to their threat levels to form a threat target set.

[0067] Based on the set of threat targets, a joint collision avoidance decision is made, and the collision avoidance effects of horizontal and vertical maneuvers are evaluated simultaneously to generate coordinated horizontal and vertical collision avoidance commands.

[0068] It should be noted that, firstly, a unified target database is established, including aerial targets, terrain obstacles, and the drone's status. Then, the potential conflict time and risk level between each target and the drone are predicted, and threats are ranked according to a unified risk assessment mechanism. Finally, combined with the drone's kinematic constraints, horizontal and vertical candidate collision avoidance maneuvers are jointly evaluated to generate coordinated horizontal and vertical collision avoidance commands, which are then output to the flight control system for execution. Compared to the existing decentralized architecture that establishes separate databases for aerial targets and terrain obstacles and conducts separate evaluations and decisions, this embodiment achieves unified reception of heterogeneous information at the data input layer and realizes integrated modeling and collaborative processing of multi-source threats at the decision layer. This fundamentally eliminates information barriers between different threat sources and provides a unified data foundation for joint collision avoidance decision-making.

[0069] In this embodiment, the multi-source environmental information includes: airborne target information (e.g., ADS-B target information, V2V target information, non-cooperative target information), local status information (e.g., local navigation status information), and terrain and obstacle environment information.

[0070] The aircraft status information includes flight status parameters such as position, speed, heading, radio altitude, barometric altitude, and WGS84 coordinates; terrain and obstacle environment information is provided by the airborne terrain database, including digital elevation model (DEM) and data on fixed obstacles such as buildings and towers.

[0071] In an optional embodiment, prior to the unified target modeling, target monitoring and environmental information processing are further included, the target monitoring and environmental information processing comprising:

[0072] Receive multi-source environmental information, including ADS-B target information, V2V target information, non-cooperative target information, local navigation status information, and terrain and obstacle environment information;

[0073] Perform data validity checks on the input data;

[0074] Update the target's status, establish the target's trajectory, and continuously maintain the target's status;

[0075] Multi-source targets are fused and associated based on their address, spatial location, velocity, and temporal consistency.

[0076] It should be noted that, in order to achieve unified collision avoidance of aerial targets and terrain obstacles, the airborne surveillance information, the machine's status information, and the environmental information are first processed in a unified manner to form the environmental data required for subsequent threat assessment.

[0077] Specifically, the data integrity and validity of multi-source environmental information are first checked, including consistency checks of information such as orientation, altitude, speed, heading, and timestamps; data quality is checked for ADS-B target information and V2V target information, and existing methods can be used for data quality checks, which are not further limited in this embodiment; data that has timed out, failed, or does not meet the quality requirements is not included in subsequent processing, and the corresponding target is set to a degraded state.

[0078] By conducting unified validity testing and quality control on multi-source heterogeneous data, the system ensures that the data entering subsequent processing stages has reliable integrity and consistency, avoids erroneous threat assessments caused by anomalies in data from a single monitoring source, and improves the system's robustness in complex electromagnetic environments.

[0079] Secondly, if the target data is the first time the target has been received, a target track is established and the target state of the target is initialized; if it is not the first time the target has been received, the target state, including position, speed, heading and altitude, is updated according to the latest received data.

[0080] Finally, multi-source fusion is performed on target information from different monitoring sources. In this embodiment, multi-source fusion refers to completing target association based on target address, spatial location, speed and time consistency to avoid establishing duplicate tracks for the same target.

[0081] Furthermore, when multiple surveillance sources simultaneously provide information about the same target, the optimal surveillance source is selected based on the target's credibility and data quality, and the M-of-N mechanism is used to maintain track continuity and reduce frequent track switching.

[0082] After completing the unified processing of the above data, the environmental information of aerial targets, the status of the aircraft, and terrain obstacles is organized into an environmental information set, which is then used for subsequent unified target modeling.

[0083] In one alternative embodiment, the unified target modeling includes:

[0084] Based on the results of environmental information processing, targets within the monitoring range are classified into aerial targets, terrain targets, fixed obstacle targets, and local targets.

[0085] Different types of targets are represented in a unified manner. Among them, aerial targets retain dynamic information, terrain targets and fixed obstacle targets are described using static target models, and local targets are represented using real-time updated status information.

[0086] Establish a unified threat description attribute for all targets, which includes target number, target type, spatial status, security protection boundary, prediction model, and target effective status;

[0087] Organize all objectives into a unified set of objectives.

[0088] like Figure 2 As shown, based on the aforementioned uniformly processed data, targets within the monitoring range are classified into aerial targets, terrain targets, fixed obstacle targets, and local targets.

[0089] Then, different types of targets are represented in a unified way. For aerial targets, dynamic information such as position, speed, heading, altitude and motion status are retained. For terrain targets and fixed obstacle targets, since their positions are fixed, a static target model is used to describe them, retaining only spatial position, elevation and geometric attributes, and setting speed and heading to zero. For local targets, the position, speed, heading and altitude information updated in real time by the navigation system is used for representation.

[0090] To achieve unified processing of different types of targets, this embodiment assigns a unified threat description attribute to each target. The threat description attribute includes target number, target type, spatial status, security protection boundary, prediction model, and target effective status. The security protection boundary describes the minimum safe distance that must be maintained around the target; the prediction model describes the target's future motion state. A motion prediction model is used for aerial targets, while a static prediction model is used for terrain and fixed obstacles, enabling risk assessment to be completed for all types of targets.

[0091] After completing the unified target representation, all targets are organized into a unified target set. This unified target set is managed according to a unified interface. Subsequent unified threat assessments obtain the target status through the unified target set and automatically call the corresponding prediction model and risk calculation method according to the target type, so as to achieve unified processing of different types of targets.

[0092] In one optional embodiment, the unified threat assessment includes:

[0093] The state prediction is performed on each target in the unified target set. For aerial targets, the future trajectory is predicted using a motion prediction model, and for terrain and fixed obstacle targets, the relative motion relationship between the machine and the target is predicted.

[0094] The target conflict time is calculated based on the predicted target status. For aerial targets, the estimated time to enter the safety protection zone is calculated based on relative position, relative speed and motion trend. For terrain and fixed obstacles, the estimated collision time is calculated based on the flight trajectory of the aircraft and the spatial position of the obstacle.

[0095] The threat level is calculated by considering the time of target conflict, security protection boundary, target movement trend, and target effective status.

[0096] All targets are sorted according to their threat level, and the N targets with the highest threat level are selected to form a threat target set.

[0097] like Figure 3 As shown, the state prediction of each target in the unified target set is first performed. Specifically, for aerial targets, the future trajectory of the target is predicted by a motion prediction model based on the current position, speed, heading and altitude. For terrain and fixed obstacle targets, since their positions are fixed, only the relative motion relationship between the machine and the target is predicted.

[0098] Subsequently, the target collision time is calculated based on the predicted target status. For aerial targets, the estimated time to enter the safety protection zone is calculated based on relative position, relative speed, and motion trend; for terrain and fixed obstacles, the estimated collision time is calculated based on the aircraft's flight trajectory and the spatial position of the obstacles.

[0099] After calculating the conflict time, a unified threat quantification is performed on each target. The threat level is calculated by comprehensively considering the target conflict time, security protection boundary, target movement trend, and target effective status. The shorter the target conflict time and the greater the degree of security protection boundary intrusion, the higher the corresponding threat level.

[0100] Finally, all targets are uniformly sorted according to their threat level, and the N targets with the highest threat level are selected to form a threat target set. The threat target set includes at least the target number, target type, conflict time, threat level, and target status. The threat target set obtained at this time is used for subsequent joint collision avoidance strategy analysis.

[0101] In one optional embodiment, the unified target set includes ground obstacle targets; wherein, the target conflict time calculation for the ground obstacle targets includes:

[0102] When the drone's current altitude is lower than the height of the obstacle plus the safety separation boundary, it is determined that it has entered the danger zone, and the conflict time is set to the minimum value.

[0103] When the drone's vertical speed approaches zero, it is determined that it has no maneuverability in the vertical direction, and the conflict time is set to the maximum value.

[0104] In other cases, a one-dimensional uniform motion model is used to estimate the time to reach the safe separation boundary, and boundary constraint processing is performed based on the estimation results;

[0105] Based on the time of the conflict and the current vertical speed of the drone, the equivalent height of the obstacle is forward-corrected.

[0106] It should be noted that, to achieve vertical risk assessment of ground obstacles by UAVs, a time parameter estimation model is constructed based on the height difference and vertical motion state. The obstacle is abstracted as a static vertical constraint target, and risk quantification is achieved through safety separation constraints. The following cases are handled based on the UAV's current altitude, obstacle height, and vertical speed: Specifically, when the UAV's altitude is lower than the obstacle height plus the safety separation boundary, it is considered to have entered a danger zone, and the conflict time is directly set to the minimum value.

[0107] When the drone's altitude is below the height of the obstacle plus the safety separation boundary, it is represented as follows:

[0108] ;

[0109] In the above formula, This is the current altitude of the drone. The height of the obstacle. To ensure safe separation of boundaries.

[0110] Setting the conflict time to the minimum value is as follows:

[0111] ;

[0112] In the above formula, For conflict time, To minimize the conflict time, in this embodiment, the minimum conflict time is set to 6.0s.

[0113] When the drone's vertical velocity approaches zero, it is determined that it has no maneuverability in the vertical direction, and the conflict time is set to the maximum value. In this implementation, a vertical velocity with an absolute value less than 0.001 is considered to be close to zero (i.e., ):

[0114] ;

[0115] In the above formula, The maximum conflict time is set to 60.0s in this embodiment.

[0116] In other cases, a one-dimensional uniform motion model is used to estimate the time to reach the safe separation boundary, i.e.:

[0117] ;

[0118] Boundary constraints are applied to conflict times. If the conflict time is less than zero, it indicates that the obstacle is far away, and the value is reset to the maximum value; if the conflict time is less than the minimum threshold, the value is reset to the minimum value; if the conflict time is greater than the maximum threshold, the value is reset to the maximum value.

[0119] Based on the time of the conflict and the current vertical velocity of the drone, the equivalent height of the obstacle is forward-corrected, and the corrected height is expressed as follows:

[0120] ;

[0121] In the above formula, This is the corrected height.

[0122] In one optional embodiment, the unified target set includes point-like obstacle targets; wherein, the target conflict time calculation for the point-like obstacle targets includes:

[0123] Based on the current direction of movement of the drone, the distance to the obstacle is projected forward, and the projected distance of the obstacle in the current direction of movement is calculated. When the projected distance is greater than the current distance, it is determined that the drone is moving away from the obstacle.

[0124] When the drone has entered the target's safe interval, the conflict time is set to the minimum value;

[0125] Set the conflict time to the maximum value when the drone is moving away from the obstacle;

[0126] In other cases, the time to reach the target safety interval is estimated based on the current horizontal approach velocity, and boundary constraints are applied;

[0127] Based on the time of the conflict and the current vertical speed of the drone, the equivalent height of the obstacle is forward-corrected.

[0128] It should be noted that the point obstacle collision time estimation method is mainly used to assess the potential collision time between the UAV and the fixed obstacle. This method quantifies the threat level to the obstacle by constructing a modified collision time estimate based on the UAV's current horizontal distance, ground speed, relative azimuth angle and vertical motion state.

[0129] First, project the distance to the obstacle forward based on the drone's current direction of motion. Assuming the current horizontal distance, ground speed, and relative azimuth angle of the obstacle are known, calculate the projected distance of the obstacle in the current direction of motion. The velocity component of the drone towards the obstacle is used to characterize the approach trend. When the projected distance is greater than the current distance, it indicates that the drone is moving away from the obstacle, and in this case, it is assumed that there is no risk of collision in the short term. The calculation process for the projected distance is as follows:

[0130] ;

[0131] In the above formula, The projected distance of the obstacle in the current direction of motion. The current horizontal distance. For ground speed, The relative azimuth angle of the obstacle.

[0132] In the above formula, Used to represent the velocity component of the drone heading towards the obstacle.

[0133] When the drone has entered the target's safe interval, the conflict time is set to the minimum value, indicating an immediate risk of collision. The judgment process is as follows:

[0134] ;

[0135] In the above formula, The target safety interval.

[0136] When the drone is moving away from the obstacle, the collision time is set to the maximum value, indicating that collision avoidance is not currently required. The determination process is as follows:

[0137] .

[0138] In other cases, the time to reach the target safety interval is estimated based on the current horizontal approach velocity, and the estimation process is as follows:

[0139] ;

[0140] In the above formula, This is the time scaling factor for the conflict time.

[0141] To avoid abnormal fluctuations in conflict time values, further boundary restrictions need to be imposed on conflict time. When the conflict time is less than the minimum threshold, it is restricted to the minimum value; when the conflict time exceeds the maximum threshold, it is restricted to the maximum value. In this embodiment, the settings of the maximum and minimum values ​​are consistent with those in the ground obstacle constraint.

[0142] Based on the time of the conflict and the current vertical velocity of the drone, the equivalent height of the obstacle is forward-corrected, and the corrected height is expressed as follows:

[0143] .

[0144] This embodiment designs appropriate conflict time estimation models for the physical characteristics of dynamic aerial targets and static terrain obstacles. However, all models ultimately output a unified conflict time index and threat level, enabling the risk of terrain obstacles to be quantitatively compared and ranked with the risk of aerial targets on the same scale. This solves the technical defect of traditional methods that use different risk indices for terrain warning and aerial collision avoidance and cannot be directly compared.

[0145] In one optional embodiment, the joint collision avoidance decision includes:

[0146] A set of joint collision avoidance candidate actions is generated based on the set of threat targets and the current flight status of the aircraft. The candidate actions include a combination of horizontal and vertical actions. The horizontal actions include maintaining the heading, turning left and turning right, and the vertical actions include maintaining altitude, climbing and descending.

[0147] For each joint candidate action, the future flight trajectory of the machine is predicted with a fixed time step, while all threat targets update their future states and the relative positional relationship between the machine and each threat target in the future prediction time domain is calculated.

[0148] For each joint candidate action, a joint cost assessment is performed in the prediction time domain, and a comprehensive cost value is formed by combining the changes in the safety intervals of each threat target, the maneuver range of the aircraft, the continuity of the action, and the satisfaction of flight constraints.

[0149] The joint candidate action that minimizes the overall cost and satisfies the flight constraints is selected as the final collision avoidance strategy, and horizontal and vertical collision avoidance commands are generated.

[0150] It should be noted that after completing the unified threat assessment, the unified set of threat targets is used as input to perform collaborative optimization of horizontal and vertical maneuvers, generating the optimal collision avoidance maneuver that satisfies the safety separation constraint. Unlike traditional methods that make horizontal and vertical collision avoidance decisions separately, this embodiment adopts a joint decision-making method, which simultaneously evaluates the collision avoidance effects of horizontal and vertical maneuvers during the same prediction process, and selects the collision avoidance strategy with the lowest overall cost in combination with flight constraints.

[0151] First, based on the set of threat targets output by the unified threat assessment module and the current flight status of the aircraft, a set of joint collision avoidance candidate actions is generated. The candidate actions include horizontal actions and vertical actions, where horizontal actions include maintaining heading, turning left and turning right, and vertical actions include maintaining altitude, climbing and descending.

[0152] Subsequently, the horizontal and vertical movements are combined to form the joint candidate movement set shown in Table 1:

[0153]

[0154] As shown in the table above, the joint candidate maneuver set includes horizontal and vertical maneuvers, with each maneuver number serving as an index. For each joint candidate maneuver, the aircraft's future flight trajectory is predicted at a fixed time step. For horizontal maneuvers, the aircraft's trajectory is updated based on the corresponding turn rate; for vertical maneuvers, the aircraft's altitude is updated based on the corresponding climb or descent rate.

[0155] At the same time, all threat targets update their future states according to the prediction results of the unified threat assessment module, and calculate the relative positional relationship between the local machine and each threat target in the future prediction time domain.

[0156] The trajectory prediction ends when the current forward prediction reaches the prediction termination time or when all threats meet the safety protection interval.

[0157] After the joint trajectory prediction is completed, a unified cost assessment is performed on each joint candidate action in the prediction time domain. This assessment is based on the predicted trajectory and comprehensively considers the changes in the safety intervals of each threat target, the range of the aircraft's maneuvering, the continuity of the action, and the satisfaction of flight constraints, so as to quantify the overall execution effect of the joint candidate action.

[0158] Specifically, the relative positional relationships between the aircraft and each threat target are calculated progressively within the prediction time domain, and a comprehensive cost value is formed based on the prediction results. First, the safety interval between the aircraft and each threat target is checked during the prediction process. When the predicted trajectory approaches or enters the target's safety protection boundary, a risk assessment value is calculated based on the conflict time and the degree of safety interval intrusion. Then, the heading change, altitude change, and maneuver duration generated by the joint candidate actions are evaluated. The larger the heading change angle, the higher the climb or descent rate, and the longer the maneuver duration, the higher the corresponding maneuver cost. Furthermore, to avoid frequent switching of collision avoidance direction within adjacent processing cycles, the current joint candidate action is compared with the action executed in the previous cycle. When the current action is opposite to the direction of the action in the previous cycle or the change amplitude is large, an action continuity penalty is added to maintain the stability and executability of the collision avoidance action. Finally, the comprehensive cost value of the joint candidate action is obtained by combining the results of the above safety risk assessment, maneuver amplitude assessment, and action continuity assessment. The comprehensive cost value of all joint candidate actions is compared, and the joint candidate action with the smallest comprehensive cost value and that meets the flight constraints is selected as the input for subsequent optimal action selection.

[0159] After evaluating all joint candidate actions, the combined costs are ranked, and the joint candidate action with the lowest total cost is selected as the final collision avoidance strategy.

[0160] If multiple candidate actions have the same overall cost, the action with the lower risk level, the action with the smaller maneuver range, and the action consistent with the previous cycle are selected first. The optimal horizontal action and the optimal vertical action are then output.

[0161] The final collision avoidance command is generated based on the optimal joint action. For targets that support V2V cooperative collision avoidance, corresponding coordination information is generated synchronously. Finally, the collision avoidance command is output to the display system, flight control system and communication system to realize the joint collision avoidance control of UAVs.

[0162] By constructing a candidate set of combined horizontal and vertical maneuvers and performing a comprehensive cost assessment of these combined maneuvers within the same prediction time domain, this embodiment can automatically balance the synergistic effects of horizontal steering and vertical maneuvering when both airborne conflicts and terrain constraints exist simultaneously. This avoids the problem of conflict between horizontal collision avoidance commands and vertical terrain warning commands in traditional methods. Furthermore, by introducing a maneuver continuity evaluation, frequent jitter in collision avoidance commands is effectively suppressed, improving command executability and flight stability.

[0163] Example 2

[0164] Embodiment 2 of the present invention provides a small unmanned aerial vehicle (UAV) collision avoidance system with integrated terrain perception capabilities, comprising:

[0165] The target monitoring and environmental information processing module is used to receive multi-source environmental information and perform data validity detection, target status update and multi-source target fusion.

[0166] The unified target modeling module is used to convert aerial targets, terrain targets, and fixed obstacle targets into threat targets in a unified manner, and to establish a unified target set. Each target in the unified target set is represented by a unified threat description attribute.

[0167] The unified threat assessment module is used to perform status prediction, conflict time calculation, threat quantification, and threat ranking for each target in the unified target set, forming a threat target set.

[0168] The joint collision avoidance decision module is used to make joint collision avoidance decisions based on the set of threat targets, simultaneously evaluate the collision avoidance effects of horizontal and vertical maneuvers, and generate coordinated horizontal and vertical collision avoidance commands.

[0169] Example 3

[0170] Figure 4 This is a schematic diagram of the structure of an electronic device provided in Embodiment 3 of the present invention, as shown below. Figure 4 As shown, the electronic device includes a processor 21, a memory 22, an input device 23, and an output device 24; the number of processors 21 in the computer device can be one or more. Figure 4 Taking a processor 21 as an example; the processor 21, memory 22, input device 23, and output device 24 in an electronic device can be connected via a bus or other means. Figure 4 Taking the example of a connection between China and Israel via a bus.

[0171] The memory 22, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules. The processor 21 executes various functional applications and data processing of the electronic device by running the software programs, instructions, and modules stored in the memory 22, thereby realizing the collision avoidance method for small UAVs with integrated terrain perception capabilities as described in Embodiment 1.

[0172] The memory 22 may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a given function; the data storage area may store data created based on terminal usage. Furthermore, the memory 22 may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory, or other non-volatile solid-state storage device. In some instances, the memory 22 may further include memory remotely located relative to the processor 21, which can be connected to the electronic device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0173] Input device 23 can be used to receive user input such as ID and password. Output device 24 is used to output the network configuration page.

[0174] Example 4

[0175] Embodiment 4 of the present invention also provides a computer-readable storage medium, wherein the computer-executable instructions, when executed by a computer processor, are used to implement a small unmanned aerial vehicle collision avoidance method with integrated terrain perception capability as provided in Embodiment 1.

[0176] The storage medium containing computer-executable instructions provided in the embodiments of the present invention is not limited to the method operation provided in Embodiment 1, but can also execute related operations in the small UAV collision avoidance method with integrated terrain perception capability provided in any embodiment of the present invention.

[0177] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A collision avoidance method for small unmanned aerial vehicles (UAVs) integrating terrain perception capabilities, characterized in that, Includes the following steps: The UAV receives multi-source environmental information within its surveillance range, including aerial target information, its own status information, and terrain and obstacle environmental information. A unified target model is constructed for the targets corresponding to the aerial target information, the terrain targets corresponding to the terrain and obstacle environment information, and the fixed obstacle targets. Each target in the unified target set is represented by a unified threat description attribute. A unified threat assessment is performed on each target in the unified target set to predict the potential conflict time between each target and the local machine and determine the threat level. The targets are then sorted according to their threat levels to form a threat target set. Based on the set of threat targets, a joint collision avoidance decision is made, and the collision avoidance effects of horizontal and vertical maneuvers are evaluated simultaneously to generate coordinated horizontal and vertical collision avoidance commands.

2. The collision avoidance method for small unmanned aerial vehicles (UAVs) with integrated terrain perception capability according to claim 1, characterized in that, Prior to the unified target modeling, target monitoring and environmental information processing are also included, which includes: Receive multi-source environmental information, including ADS-B target information, V2V target information, non-cooperative target information, local navigation status information, and terrain and obstacle environment information; Perform data validity checks on the input data; Update the target's status, establish the target's trajectory, and continuously maintain the target's status; Multi-source targets are fused and associated based on their address, spatial location, velocity, and temporal consistency.

3. The collision avoidance method for small unmanned aerial vehicles (UAVs) with integrated terrain perception capability according to claim 2, characterized in that, The unified target modeling includes: Based on the results of environmental information processing, targets within the monitored range are classified into aerial targets, terrain targets, fixed obstacle targets, and local targets. Different types of targets are represented in a unified manner. Among them, the aerial targets retain dynamic information, the terrain targets and fixed obstacle targets are described using a static target model, and the local targets are represented using real-time updated status information. Establish a unified threat description attribute for all targets, which includes target number, target type, spatial status, security protection boundary, prediction model, and target effective status; Organize all objectives into a unified set of objectives.

4. The collision avoidance method for small unmanned aerial vehicles (UAVs) with integrated terrain perception capability according to claim 3, characterized in that, The unified threat assessment includes: The state prediction is performed on each target in the unified target set; wherein, for aerial targets, the future trajectory is predicted using a motion prediction model, and for terrain and fixed obstacle targets, the relative motion relationship between the machine and the target is predicted. The target conflict time is calculated based on the predicted target status. For aerial targets, the estimated time to enter the safety protection zone is calculated based on relative position, relative speed and motion trend. For terrain and fixed obstacles, the estimated collision time is calculated based on the flight trajectory of the aircraft and the spatial position of the obstacle. The threat level is calculated by considering the time of target conflict, security protection boundary, target movement trend, and target effective status. All targets are sorted according to their threat level, and the N targets with the highest threat level are selected to form a threat target set.

5. The collision avoidance method for small unmanned aerial vehicles (UAVs) with integrated terrain perception capability according to claim 4, characterized in that, The unified target set includes ground obstacle targets; wherein, the target conflict time calculation for the ground obstacle targets includes: When the drone's current altitude is lower than the height of the obstacle plus the safety separation boundary, it is determined that it has entered the danger zone, and the conflict time is set to the minimum value. When the drone's vertical speed approaches zero, it is determined that it has no maneuverability in the vertical direction, and the conflict time is set to the maximum value. In other cases, a one-dimensional uniform motion model is used to estimate the time to reach the safe separation boundary, and boundary constraint processing is performed based on the estimation results; Based on the time of the conflict and the current vertical speed of the drone, the equivalent height of the obstacle is forward-corrected.

6. The collision avoidance method for small unmanned aerial vehicles (UAVs) with integrated terrain perception capability according to claim 4, characterized in that, The unified target set includes point-shaped obstacle targets; wherein, the target conflict time calculation for the point-shaped obstacle targets includes: Based on the current direction of movement of the drone, the distance to the obstacle is projected forward, and the projected distance of the obstacle in the current direction of movement is calculated. When the projected distance is greater than the current distance, it is determined that the drone is moving away from the obstacle. When the drone has entered the target's safe interval, the conflict time is set to the minimum value; Set the conflict time to the maximum value when the drone is moving away from the obstacle; In other cases, the time to reach the target safety interval is estimated based on the current horizontal approach velocity, and boundary constraints are applied; Based on the time of the conflict and the current vertical speed of the drone, the equivalent height of the obstacle is forward-corrected.

7. The collision avoidance method for small unmanned aerial vehicles (UAVs) with integrated terrain perception capability according to claim 1, characterized in that, The joint collision avoidance decision includes: A set of joint collision avoidance candidate actions is generated based on the set of threat targets and the current flight status of the aircraft. The candidate actions include a combination of horizontal and vertical actions. The horizontal actions include maintaining the heading, turning left and turning right, and the vertical actions include maintaining altitude, climbing and descending. For each joint candidate action, the future flight trajectory of the machine is predicted with a fixed time step, while all threat targets update their future states and the relative positional relationship between the machine and each threat target in the future prediction time domain is calculated. For each joint candidate action, a joint cost assessment is performed in the prediction time domain, and a comprehensive cost value is formed by combining the changes in the safety intervals of each threat target, the maneuver range of the aircraft, the continuity of the action, and the satisfaction of flight constraints. The joint candidate action that minimizes the overall cost and satisfies the flight constraints is selected as the final collision avoidance strategy, and horizontal and vertical collision avoidance commands are generated.

8. A small unmanned aerial vehicle (UAV) collision avoidance system integrating terrain perception capability, used to implement the small UAV collision avoidance method integrating terrain perception capability as described in any one of claims 1 to 7, characterized in that, include: The target monitoring and environmental information processing module is used to receive multi-source environmental information and perform data validity detection, target status update and multi-source target fusion. The unified target modeling module is used to convert aerial targets, terrain targets, and fixed obstacle targets into threat targets in a unified manner, and to establish a unified target set. Each target in the unified target set is represented by a unified threat description attribute. The unified threat assessment module is used to perform status prediction, conflict time calculation, threat quantification, and threat ranking for each target in the unified target set, forming a threat target set. The joint collision avoidance decision module is used to make joint collision avoidance decisions based on the set of threat targets, simultaneously evaluate the collision avoidance effects of horizontal and vertical maneuvers, and generate coordinated horizontal and vertical collision avoidance commands.

9. An electronic device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the small unmanned aerial vehicle collision avoidance method with integrated terrain perception capability as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the collision avoidance method for small unmanned aerial vehicles with integrated terrain perception capability as described in any one of claims 1 to 7.

Images