An unmanned aerial vehicle all-directional obstacle avoidance method based on millimeter wave radar and binocular laser module
By combining millimeter-wave radar and binocular laser modules, omnidirectional obstacle detection of drones is achieved. The threat level is distinguished according to the nature of the obstacle, and targeted obstacle avoidance actions are performed. This solves the problem of poor obstacle avoidance performance of drones under low visibility conditions, reduces energy consumption, and ensures endurance and safety.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINESE PEOPLES LIBERATION ARMY KET FORCE SERGEANT SCHOOL
- Filing Date
- 2023-07-05
- Publication Date
- 2026-05-29
AI Technical Summary
Existing drone obstacle avoidance technologies are ineffective in low visibility conditions, sensors are susceptible to interference and consume a lot of energy, making it difficult to achieve omnidirectional obstacle avoidance.
It employs a combination of millimeter-wave radar and binocular laser modules to achieve omnidirectional obstacle detection, distinguish the degree of threat based on the nature of the obstacle, and perform targeted obstacle avoidance actions, including low-frequency monitoring, increasing detection frequency, and flight path planning.
It enables omnidirectional obstacle avoidance for drones under low visibility conditions, reduces the energy consumption of detection equipment, ensures endurance, and improves the safety and efficiency of obstacle avoidance.
Smart Images
Figure CN117075146B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of obstacle avoidance for unmanned aerial vehicles (UAVs), and more particularly to an omnidirectional obstacle avoidance method for UAVs based on millimeter-wave radar and binocular laser modules. Background Technology
[0002] The autonomous obstacle avoidance function of drones mainly relies on the onboard ranging sensor to perceive in real time whether there are obstacles around the drone and the distance between the drone and the obstacle.
[0003] Omnidirectional obstacle avoidance for drones requires the ability to perceive surrounding obstacles and distance information without blind spots within a 360-degree horizontal range. Simultaneously, it provides obstacle and distance information within a 180-degree range in both the upper and lower planes. The omnidirectional obstacle avoidance function of drones is achieved through radar ranging sensors, vision technology, path planning technology, and other technologies.
[0004] Currently, common sensors used in drone obstacle avoidance technology include infrared sensors, ultrasonic sensors, laser sensors, and visual sensors. Civilian drones primarily use visual sensors for obstacle avoidance, but these have significant limitations in low-visibility conditions such as darkness, rain, and fog. Infrared and ultrasonic sensors can operate under these conditions, but they also have limitations in application, such as the ability of actively emitted infrared and ultrasonic waves to be absorbed by specific objects or being susceptible to interference. Summary of the Invention
[0005] The purpose of this invention is to overcome the shortcomings of the above-mentioned traditional technologies and provide an omnidirectional obstacle avoidance method for UAVs based on millimeter-wave radar and binocular laser modules, which performs targeted obstacle avoidance actions based on the different obstacle properties obtained by the obstacle avoidance detection module.
[0006] The objective of this invention is achieved through the following technical measures:
[0007] A method for omnidirectional obstacle avoidance of unmanned aerial vehicles (UAVs) based on millimeter-wave radar and binocular laser modules, characterized by the following steps:
[0008] S1. The UAV's flight control system determines the direction based on the destination information, and the obstacle avoidance detection module obtains the position and depth information of obstacles in that direction to plan the flight path;
[0009] S2. The obstacle avoidance detection module detects and obtains the distance between obstacles and the drone in real time, distinguishes different obstacle states, and performs targeted real-time obstacle avoidance. Based on whether the obstacle is moving, it distinguishes between stationary obstacles and moving obstacles. The obstacle avoidance detection module further distinguishes moving obstacles into departing obstacles, parallel obstacles, low-speed obstacles, and high-speed obstacles. Departing obstacles gradually increase in distance from the drone and their flight direction moves away from the drone and its path. Parallel obstacles maintain a stable distance from the drone and are in the same flight direction as the drone. Low-speed obstacles move at a low speed lower than the drone's flight speed toward the drone or its flight path. High-speed obstacles move at a high speed higher than the drone's flight speed. If the obstacle is identified as a stationary obstacle or a departing obstacle, proceed to S3; if it is identified as a parallel obstacle or a low-speed obstacle, proceed to S4; if it is identified as a high-speed obstacle, proceed to S5.
[0010] S3, the obstacle avoidance processing unit treats newly appearing stationary obstacles and departing obstacles in the flight direction as low-risk obstacles. It expands the model according to the circumcircle radius of the UAV and the circumcircle radius of the obstacle with a normal safety threshold to construct a collision behavior zone. It then plans an optimal flight path in real time to avoid the UAV from entering the collision behavior zone and avoid obstacles. This is updated for the UAV to fly, thus achieving real-time obstacle avoidance flight.
[0011] S4. The obstacle avoidance processing unit treats newly appearing parallel obstacles and low-speed obstacles in the flight direction as medium-risk obstacles. It performs expansion modeling based on the circumradius of the UAV and the obstacle using normal and medium-risk safety thresholds respectively. A medium-risk collision behavior zone is constructed based on the modeling data of the medium-risk safety threshold, and an optimal flight path that avoids the UAV entering the medium-risk collision behavior zone and avoids obstacles is planned in real time and updated for the UAV to fly, achieving real-time obstacle avoidance flight. If the medium-risk collision behavior zone completely blocks the flight direction and an optimal flight path that avoids the UAV entering the medium-risk collision behavior zone and avoids obstacles cannot be constructed, a collision behavior zone is constructed according to the normal safety threshold model, and an optimal flight path that avoids the UAV entering the collision behavior zone and avoids obstacles is planned in real time and updated for the UAV to fly, achieving real-time obstacle avoidance cautious flight. Simultaneously, the UAV's flight speed is reduced and the obstacle detection frequency of the obstacle avoidance module is increased to a medium frequency. If the collision behavior zone constructed according to the normal safety threshold model completely blocks the flight direction and an optimal flight path that avoids the UAV entering the collision behavior zone and avoids obstacles cannot be constructed, then the UAV stops moving forward, hovers in place, or returns to the starting point.
[0012] S5. The obstacle avoidance processing unit treats newly appearing high-speed obstacles in the flight direction as high-risk obstacles. It performs expansion modeling based on the circumradius of both the UAV and the obstacle, using both normal and high-risk safety thresholds. Based on the high-risk safety threshold modeling data, it constructs a high-risk collision behavior zone and plans an optimal flight path in real-time to avoid the UAV entering the high-risk collision behavior zone and bypassing the obstacle. This is updated to the UAV for flight, enabling real-time obstacle avoidance for medium-risk flight and increasing the obstacle detection frequency of the obstacle avoidance module to a high frequency. If the collision behavior zone constructed according to the high-risk safety threshold model completely blocks the flight direction and an optimal flight path cannot be constructed to avoid the UAV entering the collision behavior zone and bypassing the obstacle, then the collision behavior zone is constructed according to the normal safety threshold model. A flight path is set in real-time based on the flight direction furthest from or farthest from the high-speed obstacle and updated to the UAV for flight, enabling real-time obstacle avoidance for dangerous flight and increasing the obstacle detection frequency of the obstacle avoidance module to a high frequency. If the collision behavior zone constructed according to the normal safety threshold model completely blocks the flight direction and an optimal flight path cannot be constructed to avoid the UAV entering the collision behavior zone and bypassing the obstacle, then the UAV stops moving forward, hovers in place, or returns to the starting point.
[0013] S6. The drone's flight control system determines whether it has reached the destination. If it has reached the destination, it hovers or lands. If it has not reached the destination, it repeats from step S1.
[0014] As a preferred solution, the obstacle avoidance detection module in S2 uses the following method to detect the distance between the drone and obstacles in real time:
[0015] After detecting an obstacle, the obstacle avoidance detection module acquires the position vector of the obstacle relative to the drone through two consecutive frames and calculates the relative velocity vector between the obstacle and the drone. It then performs smoothing filtering on the relative velocity vectors of multiple frames to obtain the filtered relative velocity vector.
[0016] If the drone moves along its current heading and speed, the nearest distance vector between the drone and the obstacle is calculated. The nearest distance vector is obtained by the following formula:
[0017] ,
[0018] in, This represents the relative position vector between the two. This represents the distance vector between the two nearest points. This represents the unit velocity vector of the drone relative to the obstacle;
[0019] in, It is obtained from the following formula:
[0020] ,
[0021] in, Let be the velocity of the drone relative to the obstacle in the normal direction. The absolute velocity of the drone relative to the obstacle in the normal direction.
[0022] As a preferred embodiment, in S3-S5, the normal safety threshold is a threshold set according to the dynamic characteristics of the UAV. This threshold refers to the distance at which a stationary obstacle may pose a threat to the flight of the UAV, that is, the distance at which a collision may occur between the UAV and the stationary obstacle.
[0023] As a preferred embodiment, in S4, the medium-risk safety threshold is a threshold set based on the dynamic characteristics of the UAV. This threshold refers to the distance at which an obstacle with a flight speed lower than the UAV's flight speed may pose a threat to the UAV's flight. In other words, the time period during which the UAV can actively avoid a moving obstacle that is actively approaching the UAV is a reaction time unit. The distance of the medium-risk obstacle flying towards the UAV within this reaction time unit, plus the normal safety threshold, is the medium-risk safety threshold.
[0024] As a preferred embodiment, in S5, the high-risk safety threshold is a threshold set based on the dynamic characteristics of the UAV. This threshold refers to the distance at which an obstacle with a flight speed higher than that of the UAV may pose a threat to the UAV's flight. In other words, the time period during which the UAV can actively avoid a moving obstacle that is actively approaching the UAV is a reaction time unit. The distance of the high-risk obstacle flying towards the UAV within this reaction time unit plus the normal safety threshold is the medium-risk safety threshold.
[0025] As a preferred embodiment, the calculation of the collision behavior zone in S2-S5 involves the UAV moving along its current heading and speed, determining the distance at which the UAV may be unable to avoid colliding with the obstacle based on the UAV's dynamic characteristics, and defining the collision behavior zone based on the obstacle's position and this distance.
[0026] As a preferred solution, the real-time obstacle avoidance calculation method for UAVs in S2-S5 is as follows:
[0027] Step 1: Use the obstacle avoidance detection module to obtain the position and depth information of obstacles in the flight direction;
[0028] Step 2: Construct the collision behavior zone. This involves expanding the model based on the radius of the drone's circumscribed circle and the dimensions of the obstacle to establish the collision behavior zone. The goal is to prevent the drone from entering this zone. The calculation formula is as follows:
[0029] L=R+r+λ
[0030] Where L is the length of the collision behavior zone, R is the radius of the circumscribed circle of the UAV, r is the radius of the circumscribed circle of the obstacle, and λ is the threshold. The threshold is determined by the obstacle avoidance detection module in real time by detecting the distance between the obstacle and the UAV and distinguishing different obstacle states.
[0031] Step 3: Perform optimal path planning based on the UAV's direction of motion and its distance from the edge of the obstacle after expansion modeling;
[0032] When the first obstacle is detected, it is determined whether the drone has entered the collision behavior zone. If it has entered the collision behavior zone, the drone will autonomously switch to hover mode and wait for the obstacle avoidance system to find the optimal path before continuing to move forward at full speed. If it has not entered the collision behavior zone, it will continue to fly and calculate the next waypoint.
[0033] As a preferred solution, the obstacle avoidance detection module uses a binocular laser sensor on the UAV to measure forward obstacle distance, a three-dimensional millimeter-wave radar on the UAV to measure obstacles above, below, and behind the UAV, and a left- and right-oriented millimeter-wave radar on the UAV to measure obstacles to the left and right.
[0034] As a preferred solution, in S2, a binocular laser sensor enables forward obstacle ranging, a three-directional millimeter-wave radar enables obstacle ranging in the up, down, and back directions, and two unidirectional millimeter-wave radars enable obstacle ranging in the left and right directions, respectively.
[0035] Due to the adoption of the above technical solution, the beneficial effects of the present invention compared with the prior art are as follows:
[0036] This application discloses an omnidirectional obstacle avoidance method for unmanned aerial vehicles (UAVs) based on millimeter-wave radar and a binocular laser module. Specifically, it utilizes a binocular laser sensor for forward obstacle ranging, a three-directional millimeter-wave radar for upward, downward, and backward obstacle ranging, and two unidirectional millimeter-wave radars for left and right obstacle ranging. This enables omnidirectional obstacle detection for the UAV, and the method determines the nature of the obstacle and its threat level to the UAV's flight based on the detection results, taking avoidance actions according to different threat levels. The avoidance actions vary depending on the level of threat. Generally, for low-threat stationary obstacles or obstacles far from the drone, low-frequency monitoring is the primary method, without the need for specific flight path avoidance. For moving obstacles approaching or parallel to the drone, the detection frequency is increased, and the obstacle avoidance route is replanned based on potential risks to proactively increase the distance from potential threats. For moving obstacles flying erratically and at higher speeds, or moving obstacles whose flight paths intersect with the drone's and pose a collision risk, a highly vigilant approach is adopted, with high-frequency detection of obstacle movement and the planning of routes away from the obstacle to minimize the risk of collision. Using the obstacle avoidance scheme of this application, although the drone carries a large number of detection devices, including binocular laser sensors and multiple millimeter-wave radars, in most cases, the drone is in a low-risk state during flight. These detection devices can perform scanning operations at a lower frequency, significantly reducing the increased energy consumption caused by multiple detection devices. This ensures that the drone's range under normal flight conditions is not significantly less than that of a drone with fewer detection devices, guaranteeing both safety and endurance.
[0037] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. Attached Figure Description
[0038] Appendix Figure 1 This is a side view of a UAV based on millimeter-wave radar and binocular laser module according to the present invention.
[0039] Appendix Figure 2 This is a three-dimensional structural diagram of a UAV based on millimeter-wave radar and binocular laser module according to the present invention. Detailed Implementation
[0040] Example: As attached Figure 1 and Figure 2As shown, this embodiment discloses an omnidirectional obstacle avoidance method for unmanned aerial vehicles (UAVs) based on millimeter-wave radar and binocular laser modules. The UAV based on millimeter-wave radar and binocular laser modules adopts an omnidirectional obstacle avoidance design. Its omnidirectional obstacle avoidance module includes a binocular laser sensor 1 for forward obstacle ranging, a three-directional millimeter-wave radar 4 for obstacle ranging above, below, and behind the UAV, and a left-direction millimeter-wave radar 2 and a right-direction millimeter-wave radar 3 for obstacle ranging to the left and right of the UAV.
[0041] As attached Figure 1 and Figure 2 As shown, the binocular laser sensor 1 achieves forward obstacle ranging, the three-directional millimeter-wave radar 4 achieves obstacle ranging in the up, down, and rear directions, and the left-direction millimeter-wave radar 2 and right-direction millimeter-wave radar 3, as two unidirectional millimeter-wave radars, achieve obstacle ranging in the left and right directions respectively. The three-directional millimeter-wave radar 4 consists of three millimeter-wave radars: an upward millimeter-wave radar 43, a downward millimeter-wave radar 42, and a rearward millimeter-wave radar 41. The upward millimeter-wave radar 43, the downward millimeter-wave radar 42, and the rearward millimeter-wave radar 41 respectively achieve obstacle ranging in the top, bottom, and rear directions of the UAV.
[0042] The method for omnidirectional obstacle avoidance of UAVs based on millimeter-wave radar and binocular laser modules, which uses the above-mentioned omnidirectional obstacle avoidance module for real-time detection of the obstacle avoidance detection module, includes the following steps:
[0043] S1. The UAV's flight control system determines the direction based on the destination information, and the obstacle avoidance detection module obtains the position and depth information of obstacles in that direction to plan the flight path;
[0044] S2. The obstacle avoidance detection module detects and obtains the distance between obstacles and the drone in real time, distinguishes different obstacle states, and performs targeted real-time obstacle avoidance. Based on whether the obstacle is moving, it distinguishes between stationary obstacles and moving obstacles. The obstacle avoidance detection module further distinguishes moving obstacles into departing obstacles, parallel obstacles, low-speed obstacles, and high-speed obstacles. Departing obstacles gradually increase in distance from the drone and their flight direction moves away from the drone and its path. Parallel obstacles maintain a stable distance from the drone and are in the same flight direction as the drone. Low-speed obstacles move at a low speed lower than the drone's flight speed toward the drone or its flight path. High-speed obstacles move at a high speed higher than the drone's flight speed. If the obstacle is identified as a stationary obstacle or a departing obstacle, proceed to S3; if it is identified as a parallel obstacle or a low-speed obstacle, proceed to S4; if it is identified as a high-speed obstacle, proceed to S5.
[0045] S3, the obstacle avoidance processing unit treats newly appearing stationary obstacles and departing obstacles in the flight direction as low-risk obstacles. It expands the model according to the circumcircle radius of the UAV and the circumcircle radius of the obstacle with a normal safety threshold to construct a collision behavior zone. It then plans an optimal flight path in real time to avoid the UAV from entering the collision behavior zone and avoid obstacles. This is updated for the UAV to fly, thus achieving real-time obstacle avoidance flight.
[0046] S4. The obstacle avoidance processing unit treats newly appearing parallel obstacles and low-speed obstacles in the flight direction as medium-risk obstacles. It performs expansion modeling based on the circumradius of the UAV and the obstacle using normal and medium-risk safety thresholds respectively. A medium-risk collision behavior zone is constructed based on the modeling data of the medium-risk safety threshold, and an optimal flight path that avoids the UAV entering the medium-risk collision behavior zone and avoids obstacles is planned in real time and updated for the UAV to fly, achieving real-time obstacle avoidance flight. If the medium-risk collision behavior zone completely blocks the flight direction and an optimal flight path that avoids the UAV entering the medium-risk collision behavior zone and avoids obstacles cannot be constructed, a collision behavior zone is constructed according to the normal safety threshold model, and an optimal flight path that avoids the UAV entering the collision behavior zone and avoids obstacles is planned in real time and updated for the UAV to fly, achieving real-time obstacle avoidance cautious flight. Simultaneously, the UAV's flight speed is reduced and the obstacle detection frequency of the obstacle avoidance module is increased to a medium frequency. If the collision behavior zone constructed according to the normal safety threshold model completely blocks the flight direction and an optimal flight path that avoids the UAV entering the collision behavior zone and avoids obstacles cannot be constructed, then the UAV stops moving forward, hovers in place, or returns to the starting point.
[0047] S5. The obstacle avoidance processing unit treats newly appearing high-speed obstacles in the flight direction as high-risk obstacles. It performs expansion modeling based on the circumradius of both the UAV and the obstacle, using both normal and high-risk safety thresholds. Based on the high-risk safety threshold modeling data, it constructs a high-risk collision behavior zone and plans an optimal flight path in real-time to avoid the UAV entering the high-risk collision behavior zone and bypassing the obstacle. This is updated to the UAV for flight, enabling real-time obstacle avoidance for medium-risk flight and increasing the obstacle detection frequency of the obstacle avoidance module to a high frequency. If the collision behavior zone constructed according to the high-risk safety threshold model completely blocks the flight direction and an optimal flight path cannot be constructed to avoid the UAV entering the collision behavior zone and bypassing the obstacle, then the collision behavior zone is constructed according to the normal safety threshold model. A flight path is set in real-time based on the flight direction furthest from or farthest from the high-speed obstacle and updated to the UAV for flight, enabling real-time obstacle avoidance for dangerous flight and increasing the obstacle detection frequency of the obstacle avoidance module to a high frequency. If the collision behavior zone constructed according to the normal safety threshold model completely blocks the flight direction and an optimal flight path cannot be constructed to avoid the UAV entering the collision behavior zone and bypassing the obstacle, then the UAV stops moving forward, hovers in place, or returns to the starting point.
[0048] S6. The drone's flight control system determines whether it has reached the destination. If it has reached the destination, it hovers or lands. If it has not reached the destination, it repeats from step S1.
[0049] The method used by the obstacle avoidance detection module in S2 to detect the distance between the drone and obstacles in real time is as follows:
[0050] After detecting an obstacle, the obstacle avoidance detection module acquires the position vector of the obstacle relative to the drone through two consecutive frames and calculates the relative velocity vector between the obstacle and the drone. It then performs smoothing filtering on the relative velocity vectors of multiple frames to obtain the filtered relative velocity vector.
[0051] If the drone moves along its current heading and speed, the nearest distance vector between the drone and the obstacle is calculated. The nearest distance vector is obtained by the following formula:
[0052] ,
[0053] in, This represents the relative position vector between the two. This represents the distance vector between the two nearest points. This represents the unit velocity vector of the drone relative to the obstacle;
[0054] in, It is obtained from the following formula:
[0055] ,
[0056] in, Let be the velocity of the drone relative to the obstacle in the normal direction. The absolute velocity of the drone relative to the obstacle in the normal direction.
[0057] In S3-S5, the normal safety threshold is a threshold set according to the dynamic characteristics of the UAV. This threshold refers to the distance at which a stationary obstacle may pose a threat to the flight of the UAV, that is, the distance at which a collision may occur between the UAV and the stationary obstacle.
[0058] In S4, the medium-risk safety threshold is a threshold set based on the dynamic characteristics of the UAV. This threshold refers to the distance at which an obstacle with a flight speed lower than the UAV's flight speed may pose a threat to the UAV's flight. In other words, the time period during which the UAV can actively avoid a moving obstacle that is actively approaching the UAV is a reaction time unit. The distance of the medium-risk obstacle flying toward the UAV within this reaction time unit plus the normal safety threshold is the medium-risk safety threshold.
[0059] In S5, the high-risk safety threshold is a threshold set based on the dynamic characteristics of the UAV. This threshold refers to the distance at which an obstacle with a flight speed higher than that of the UAV may pose a threat to the UAV's flight. In other words, the time period during which the UAV can actively avoid a moving obstacle that is actively approaching the UAV is a reaction time unit. The distance of the high-risk obstacle flying toward the UAV within this reaction time unit plus the normal safety threshold is the medium-risk safety threshold.
[0060] The calculation of the collision behavior zone described in S2-S5: The UAV moves along the current heading and speed, and the distance at which the UAV may be unable to avoid colliding with the obstacle is determined based on the UAV's dynamic characteristics. The collision behavior zone is then defined based on the obstacle's position and this distance.
[0061] Real-time obstacle avoidance calculation method for UAVs in S2-S5:
[0062] Step 1: Use the obstacle avoidance detection module to obtain the position and depth information of obstacles in the flight direction;
[0063] Step 2: Construct the collision behavior zone. This involves expanding the model based on the radius of the drone's circumscribed circle and the dimensions of the obstacle to establish the collision behavior zone. The goal is to prevent the drone from entering this zone. The calculation formula is as follows:
[0064] L=R+r+λ
[0065] Where L is the length of the collision behavior zone, R is the radius of the circumscribed circle of the UAV, r is the radius of the circumscribed circle of the obstacle, and λ is the threshold. The threshold is determined by the obstacle avoidance detection module in real time by detecting the distance between the obstacle and the UAV and distinguishing different obstacle states.
[0066] Step 3: Perform optimal path planning based on the UAV's direction of motion and its distance from the edge of the obstacle after expansion modeling;
[0067] When the first obstacle is detected, it is determined whether the drone has entered the collision behavior zone. If it has entered the collision behavior zone, the drone will autonomously switch to hover mode and wait for the obstacle avoidance system to find the optimal path before continuing to move forward at full speed. If it has not entered the collision behavior zone, it will continue to fly and calculate the next waypoint.
[0068] The technical principle of this invention is:
[0069] This application discloses an omnidirectional obstacle avoidance method for unmanned aerial vehicles (UAVs) based on millimeter-wave radar and a binocular laser module. Specifically, it utilizes a binocular laser sensor for forward obstacle ranging, a three-directional millimeter-wave radar for upward, downward, and backward obstacle ranging, and two unidirectional millimeter-wave radars for left and right obstacle ranging. This enables omnidirectional obstacle detection for the UAV, and the method determines the nature of the obstacle and its threat level to the UAV's flight based on the detection results, taking avoidance actions according to different threat levels. The avoidance actions vary depending on the level of threat. Generally, for low-threat stationary obstacles or obstacles far from the drone, low-frequency monitoring is the primary method, without the need for specific flight path avoidance. For moving obstacles approaching or parallel to the drone, the detection frequency is increased, and the obstacle avoidance route is replanned based on potential risks to proactively increase the distance from potential threats. For moving obstacles flying erratically and at higher speeds, or moving obstacles whose flight paths intersect with the drone's and pose a collision risk, a highly vigilant approach is adopted, with high-frequency detection of obstacle movement and the planning of routes away from the obstacle to minimize the risk of collision. Using the obstacle avoidance scheme of this application, although the drone carries a large number of detection devices, including binocular laser sensors and multiple millimeter-wave radars, in most cases, the drone is in a low-risk state during flight. These detection devices can perform scanning operations at a lower frequency, significantly reducing the increased energy consumption caused by multiple detection devices. This ensures that the drone's range under normal flight conditions is not significantly less than that of a drone with fewer detection devices, guaranteeing both safety and endurance.
Claims
1. A method for omnidirectional obstacle avoidance of unmanned aerial vehicles (UAVs) based on millimeter-wave radar and binocular laser modules, characterized in that, Includes the following steps: S1. The UAV's flight control system determines the direction based on the destination information, and the obstacle avoidance detection module obtains the position and depth information of obstacles in that direction to plan the flight path; S2. The obstacle avoidance detection module detects and obtains the distance between obstacles and the drone in real time, distinguishes different obstacle states, and performs targeted real-time obstacle avoidance. Based on whether the obstacle is moving, it distinguishes between stationary obstacles and moving obstacles. The obstacle avoidance detection module further distinguishes moving obstacles into departing obstacles, parallel obstacles, low-speed obstacles, and high-speed obstacles. Departing obstacles gradually increase in distance from the drone and their flight direction moves away from the drone and its path. Parallel obstacles maintain a stable distance from the drone and are in the same flight direction as the drone. Low-speed obstacles move at a low speed lower than the drone's flight speed toward the drone or its flight path. High-speed obstacles move at a high speed higher than the drone's flight speed. If the obstacle is identified as a stationary obstacle or a departing obstacle, proceed to S3; if it is identified as a parallel obstacle or a low-speed obstacle, proceed to S4; if it is identified as a high-speed obstacle, proceed to S5. S3, the obstacle avoidance processing unit treats newly appearing stationary obstacles and departing obstacles in the flight direction as low-risk obstacles. It expands the model according to the circumcircle radius of the UAV and the circumcircle radius of the obstacle with a normal safety threshold to construct a collision behavior zone. It then plans an optimal flight path in real time to avoid the UAV from entering the collision behavior zone and avoid obstacles. This is updated for the UAV to fly, thus achieving real-time obstacle avoidance flight. S4. The obstacle avoidance processing unit treats newly appearing parallel obstacles and low-speed obstacles in the flight direction as medium-risk obstacles. It performs expansion modeling based on the circumradius of the UAV and the obstacle using normal and medium-risk safety thresholds respectively. A medium-risk collision behavior zone is constructed based on the modeling data of the medium-risk safety threshold, and an optimal flight path that avoids the UAV entering the medium-risk collision behavior zone and avoids obstacles is planned in real time and updated for the UAV to fly, achieving real-time obstacle avoidance flight. If the medium-risk collision behavior zone completely blocks the flight direction and an optimal flight path that avoids the UAV entering the medium-risk collision behavior zone and avoids obstacles cannot be constructed, a collision behavior zone is constructed according to the normal safety threshold model, and an optimal flight path that avoids the UAV entering the collision behavior zone and avoids obstacles is planned in real time and updated for the UAV to fly, achieving real-time obstacle avoidance cautious flight. Simultaneously, the UAV's flight speed is reduced and the obstacle detection frequency of the obstacle avoidance module is increased to a medium frequency. If the collision behavior zone constructed according to the normal safety threshold model completely blocks the flight direction and an optimal flight path that avoids the UAV entering the collision behavior zone and avoids obstacles cannot be constructed, then the UAV stops moving forward, hovers in place, or returns to the starting point. S5. The obstacle avoidance processing unit treats newly appearing high-speed obstacles in the flight direction as high-risk obstacles. It performs expansion modeling based on the circumradius of both the UAV and the obstacle, using both normal and high-risk safety thresholds. Based on the high-risk safety threshold modeling data, it constructs a high-risk collision behavior zone and plans an optimal flight path in real-time to avoid the UAV entering the high-risk collision behavior zone and bypassing the obstacle. This is updated to the UAV for flight, enabling real-time obstacle avoidance for medium-risk flight and increasing the obstacle detection frequency of the obstacle avoidance module to a high frequency. If the collision behavior zone constructed according to the high-risk safety threshold model completely blocks the flight direction and an optimal flight path cannot be constructed to avoid the UAV entering the collision behavior zone and bypassing the obstacle, then the collision behavior zone is constructed according to the normal safety threshold model. A flight path is set in real-time based on the flight direction furthest from or farthest from the high-speed obstacle and updated to the UAV for flight, enabling real-time obstacle avoidance for dangerous flight and increasing the obstacle detection frequency of the obstacle avoidance module to a high frequency. If the collision behavior zone constructed according to the normal safety threshold model completely blocks the flight direction and an optimal flight path cannot be constructed to avoid the UAV entering the collision behavior zone and bypassing the obstacle, then the UAV stops moving forward, hovers in place, or returns to the starting point. S6. The drone's flight control system determines whether it has reached the destination. If it has reached the destination, it hovers or lands. If it has not reached the destination, it repeats from step S1.
2. The omnidirectional obstacle avoidance method for unmanned aerial vehicles based on millimeter-wave radar and binocular laser module according to claim 1, characterized in that: The method used by the obstacle avoidance detection module in S2 to detect the distance between the drone and obstacles in real time is as follows: After detecting an obstacle, the obstacle avoidance detection module acquires the position vector of the obstacle relative to the drone through two consecutive frames and calculates the relative velocity vector between the obstacle and the drone. It then performs smoothing filtering on the relative velocity vectors of multiple frames to obtain the filtered relative velocity vector. If the drone moves along its current heading and speed, the nearest distance vector between the drone and the obstacle is calculated. The nearest distance vector is obtained by the following formula: ， in, This represents the relative position vector between the two. This represents the distance vector between the two nearest points. This represents the unit velocity vector of the drone relative to the obstacle; in, It is obtained from the following formula: ， in, Let be the velocity of the drone relative to the obstacle in the normal direction. The absolute velocity of the drone relative to the obstacle in the normal direction.
3. The omnidirectional obstacle avoidance method for UAVs based on millimeter-wave radar and binocular laser module according to claim 1, characterized in that: In S3-S5, the normal safety threshold is a threshold set according to the dynamic characteristics of the UAV. This threshold refers to the distance at which a stationary obstacle may pose a threat to the flight of the UAV, that is, the distance at which a collision may occur between the UAV and the stationary obstacle.
4. The omnidirectional obstacle avoidance method for UAVs based on millimeter-wave radar and binocular laser module according to claim 1, characterized in that: In S4, the medium-risk safety threshold is a threshold set based on the dynamic characteristics of the UAV. This threshold refers to the distance at which an obstacle with a flight speed lower than the UAV's flight speed may pose a threat to the UAV's flight. In other words, the time period during which the UAV can actively avoid a moving obstacle that is actively approaching the UAV is a reaction time unit. The distance of the medium-risk obstacle flying toward the UAV within this reaction time unit plus the normal safety threshold is the medium-risk safety threshold.
5. The omnidirectional obstacle avoidance method for unmanned aerial vehicles based on millimeter-wave radar and binocular laser module according to claim 1, characterized in that: In S5, the high-risk safety threshold is a threshold set based on the dynamic characteristics of the UAV. This threshold refers to the distance at which an obstacle with a flight speed higher than that of the UAV may pose a threat to the UAV's flight. In other words, the time period during which the UAV can actively avoid a moving obstacle that is actively approaching the UAV is a reaction time unit. The distance of the high-risk obstacle flying toward the UAV within this reaction time unit plus the normal safety threshold is the medium-risk safety threshold.
6. The omnidirectional obstacle avoidance method for unmanned aerial vehicles based on millimeter-wave radar and binocular laser module according to claim 1, characterized in that: The calculation of the collision behavior zone described in S2-S5: The UAV moves along the current heading and speed, and the distance at which the UAV may be unable to avoid colliding with the obstacle is determined based on the UAV's dynamic characteristics. The collision behavior zone is then defined based on the obstacle's position and this distance.
7. The omnidirectional obstacle avoidance method for unmanned aerial vehicles based on millimeter-wave radar and binocular laser module according to claim 1, characterized in that: Real-time obstacle avoidance calculation method for UAVs in S2-S5: Step 1: Use the obstacle avoidance detection module to obtain the position and depth information of obstacles in the flight direction; Step 2: Construct the collision behavior zone. This involves expanding the model based on the radius of the drone's circumscribed circle and the dimensions of the obstacle to establish the collision behavior zone. The goal is to prevent the drone from entering this zone. The calculation formula is as follows: L=R+r+λ Where L is the length of the collision behavior zone, R is the radius of the circumscribed circle of the UAV, r is the radius of the circumscribed circle of the obstacle, and λ is the threshold. The threshold is determined by the obstacle avoidance detection module in real time by detecting the distance between the obstacle and the UAV and distinguishing different obstacle states. Step 3: Perform optimal path planning based on the UAV's direction of motion and its distance from the edge of the obstacle after expansion modeling; When the first obstacle is detected, it is determined whether the drone has entered the collision behavior zone. If it has entered the collision behavior zone, the drone will autonomously switch to hover mode and wait for the obstacle avoidance system to find the optimal path before continuing to move forward at full speed. If it has not entered the collision behavior zone, it will continue to fly and calculate the next waypoint.
8. A method for omnidirectional obstacle avoidance of a UAV based on millimeter-wave radar and a binocular laser module according to any one of claims 1 to 7, characterized in that: The obstacle avoidance detection module uses a binocular laser sensor on the drone to measure forward obstacles, a three-dimensional millimeter-wave radar on the drone to measure obstacles above, below, and behind the drone, and a left- and right-oriented millimeter-wave radar on the drone to measure obstacles to the left and right.
9. The omnidirectional obstacle avoidance method for unmanned aerial vehicles based on millimeter-wave radar and binocular laser module according to claim 8, characterized in that: In S2, a binocular laser sensor enables forward obstacle ranging, a three-directional millimeter-wave radar enables obstacle ranging in the up, down, and back directions, and two unidirectional millimeter-wave radars enable obstacle ranging in the left and right directions, respectively.