A method and system for dynamic obstacle identification and avoidance in a drone hoisting process

Abstract

The present application relates to a kind of unmanned aerial vehicle hoisting process obstacle dynamic identification and avoidance method and system, it is related to the field of unmanned aerial vehicle hoisting, it includes: collection early warning image information and unmanned aerial vehicle current position;Based on early warning image information and pedestrian feature to obtain the pedestrian travel trajectory;Based on unmanned aerial vehicle current position to obtain hoisting dangerous area;According to pedestrian travel trajectory and hoisting dangerous area to determine the remaining time of pedestrian into hoisting dangerous area;According to pedestrian travel trajectory to determine the throw drop point;Control unmanned aerial vehicle to throw warning body towards throw drop point, and collect the stepping signal of warning body;In response to stepping signal to issue sound prompt, to prompt pedestrian to pause travel, and keep unmanned aerial vehicle continue to execute hoisting operation according to original flight line.The present application has the effect of considering the continuity and safety of hoisting operation.

Description

Technical Field

[0001] This invention relates to the field of drone hoisting, and in particular to a method and system for dynamic obstacle identification and avoidance during drone hoisting. Background Technology

[0002] Obstacle dynamic recognition and avoidance refers to the technology by which unmanned aerial vehicles (UAVs) perceive dynamic obstacles (such as pedestrians and vehicles) in the surrounding environment during hoisting operations and take corresponding measures to avoid collisions or interference.

[0003] Currently, drone hoisting operations typically rely on onboard vision or radar sensors to detect obstacles. When a pedestrian is detected entering the safe zone, existing systems generally handle the situation by hovering and waiting, flying around the obstacle, or directly interrupting the hoisting operation.

[0004] However, hovering or flying around significantly reduces lifting efficiency, and frequent interruptions to operations increase operational complexity. Furthermore, existing technologies lack the ability to proactively predict and warn pedestrian movement trends, often passively waiting for pedestrians to leave rather than guiding them to actively avoid the area. This makes it difficult to balance the continuity and safety of lifting operations, and therefore requires improvement. Summary of the Invention

[0005] To balance the continuity and safety of hoisting operations, this invention provides a method and system for dynamic obstacle identification and avoidance during UAV hoisting.

[0006] In a first aspect, the present invention provides a method for dynamic identification and avoidance of obstacles during the hoisting process of an unmanned aerial vehicle (UAV), employing the following technical solution: A method for dynamic obstacle identification and avoidance during drone hoisting includes: Collect early warning image information and the current location of the drone; When the warning image information contains preset pedestrian features, the pedestrian's current position, current speed, and gait phase information are obtained based on the warning image information and pedestrian features. By combining the pedestrian's current location, current speed, and gait phase information, the pedestrian's trajectory can be predicted. The hoisting hazard zone is determined based on the drone's current location and the preset hoisting hazard range. The remaining time for a pedestrian to enter the hoisting danger zone is determined based on the pedestrian's movement trajectory and the hoisting danger zone. When the remaining time is less than a preset first threshold, the landing point of the projectile is determined based on the pedestrian's trajectory. Control the drone to launch warning objects towards the launch point and collect the trampling signals of the warning objects; The drone responds to the stampede signal by emitting an audible alert to prompt pedestrians to stop and continues to perform the hoisting operation along its original flight path.

[0007] By adopting the above technical solution, pedestrians are identified by collecting early warning images and their position, speed and gait phase information are extracted. The gait phase information is combined with the prediction of pedestrian trajectories to improve the accuracy of trajectory prediction. The remaining time for pedestrians to enter the danger zone is calculated based on the dangerous range of the drone hoisting. When the time is less than the threshold, a warning object is actively launched to the landing point on the predicted trajectory. An audible prompt is issued after the pedestrian steps on it to guide the pedestrian to stop walking. This allows the drone to continue the original hoisting operation without hovering or flying around, thus balancing the continuity and safety of the hoisting operation.

[0008] Optionally, a method for recovering warning items may also be included: The length of the collection and recovery line and the distance between the warning device and the drone; The current relaxation length of the recovery line is calculated based on the release length value and the interval distance value; The real-time drive current value of the winch in the recovery module is collected, and the recovery resistance coefficient is determined based on the real-time drive current value. The target recovery rate value is obtained based on the current relaxation length and recovery resistance coefficient; The control recycling module recycles the warning body at the target recycling speed value.

[0009] By adopting the above technical solution, the current slack length is calculated by collecting the length of the retractable cable and the distance between the warning body and the cable. The retraction resistance coefficient is determined by combining the winch drive current, and the target retraction speed is dynamically adjusted. This achieves adaptive control of the cable tension during the retraction of the warning body, avoiding retraction jamming or tangling due to slackness or excessive tightness, and improving the reliability and efficiency of retraction.

[0010] Optional, also includes: Collect the real-time speed value of the recovery line and the real-time tension value of the recovery line after the recovery is started; The velocity deviation is calculated based on the real-time velocity value and the target recovery velocity value; The tension influence factor is calculated based on the real-time tension value and the preset benchmark tension value. The recovery speed correction is obtained based on the speed deviation and tension influence factor; The target recovery speed value is added to the recovery speed correction value to obtain the dynamic recovery speed value, and the recovery module is controlled to continue to recover the warning body at the dynamic recovery speed value.

[0011] By adopting the above technical solution, the correction amount of the recovery speed is calculated based on the deviation between the real-time speed and the target speed and the influence factor between the real-time tension and the reference tension, so as to realize the closed-loop adjustment of the recovery speed, so that the recovery process can remain stable when the resistance changes, prevent the cable from suddenly tightening or loosening, and improve the safety and stability of the recovery.

[0012] Optionally, methods for handling multi-user scenarios are also included: Collect the number of pedestrians in the hazardous area of ​​the hoisting operation; When the number of pedestrians is greater than 1, collect the entry time and location of each pedestrian. Priority targets are selected based on the time each pedestrian enters. The launch point is determined based on the priority launch target, and the drone is controlled to launch a warning object toward that launch point; The sound pressure level of the response generated after the priority target tramples the warning object, as well as the head deflection angle of other pedestrians, are collected. The effective sound pressure level at other pedestrian locations is calculated based on the response sound pressure level and the location of each pedestrian. The head deflection angle and the preset attention deflection threshold are used to determine the attention capture status of each other pedestrian; When the effective sound pressure value is greater than the preset alertness threshold and the attention capture state is captured, it is determined that the corresponding pedestrian has voluntarily avoided the situation, and the throwing action against the pedestrian is suppressed.

[0013] By adopting the above technical solution, in multi-pedestrian scenarios, warning objects are launched first to pedestrians who enter the danger zone. The sound generated after they step on the object and the head turning angle and effective sound pressure value of other pedestrians are used to comprehensively judge the attention capture status. The launching action is suppressed for pedestrians who have already avoided the danger zone, avoiding repeated deployment and improving the warning efficiency and targeting in multi-person scenarios.

[0014] Optionally, a group alert confirmation method based on the content of the response sound can also be included: Collect audio clips of the response sounds generated after the target is prioritized to step on the warning object; The fundamental frequency change rate and short-time energy value are extracted from the speech segment, and the intonation excitation index is calculated based on the fundamental frequency change rate and short-time energy value; The audio clips are matched with a pre-defined warning word list to obtain a matching score. The confidence level of the audio content is obtained based on the intonation arousal index and the matching score. When the confidence level of the sound content is greater than the preset content threshold, a group warning signal is generated, and the throwing action to all other pedestrians is suppressed based on the signal.

[0015] By adopting the above technical solution, the voice segments after the priority target is trampled are collected, the fundamental frequency change rate and short-time energy value are extracted to calculate the tone arousal index, and the confidence level is obtained by matching it with the warning word library. When the confidence level exceeds the threshold, a group warning effective signal is generated to suppress the throwing to other pedestrians. The warning effect is verified by the sound content to avoid invalid throwing due to misjudgment.

[0016] Optionally, a projectile suppression verification method may also be included: Collect the moment of head deflection and the trigger moment of the trampling signal from the priority projectile target; The reaction delay time is calculated based on the head deflection time and the trigger time. The pedestrian distance value is determined by combining the positions of each pedestrian, the priority target for projectile, and the early warning image information. The speed of line-of-sight propagation is calculated based on pedestrian distance and reaction delay time; When the speed of line-of-sight propagation falls within the preset normal propagation speed range, a projectile suppression signal is generated; In response to the projectile suppression signal, projectile actions against all other pedestrians who have not yet turned their heads are suppressed.

[0017] By adopting the above technical solution, the reaction delay time is calculated based on the head deflection time and the stamping signal trigger time. The speed of visual propagation is calculated in combination with the pedestrian distance. When the speed falls into the normal propagation range, it is determined to be an effective attention transfer, and a projectile suppression signal is generated to avoid repeated projection on pedestrians who have already observed the warning event, thereby improving the accuracy and rationality of projectile suppression.

[0018] Optionally, a method for determining the launch landing point may also be included: The direction of pedestrian movement is determined based on the pedestrian's trajectory. Gait phase period is determined based on gait phase information; The current supporting foot position is determined based on gait phase information and the pedestrian's current position; The stride length is obtained based on the pedestrian's current speed and gait phase period; Determine the half-step offset by combining the current support foot position and stride length; The landing point of the projectile is determined based on the pedestrian's direction of travel, half-step offset, and the current position of the supporting foot.

[0019] By adopting the above technical solution, the current supporting foot position and stride length are determined based on the gait phase, and the half-step offset is calculated in combination with the gait cycle. This accurately predicts the pedestrian's next foot landing point, so that the projectile landing point falls on the position where the pedestrian is about to step under the natural gait, thereby improving the probability of the warning object being stepped on and the accuracy of the warning timing.

[0020] Optionally, a method for synchronizing the launch timing may also be included: The current phase value is determined based on gait phase information; Predict the start time of the support phase based on the gait phase period and the current phase value; The estimated flight time is determined by combining the launch landing point and the current position of the drone; The ejection trigger time is calculated based on the support phase start time and the expected flight time. When the system time reaches the launch trigger moment, the launch module is controlled to emit an alarm.

[0021] By adopting the above technical solution, the starting time of the support phase is predicted based on the gait phase cycle and the current phase. The launch trigger time is calculated by combining the flight time of the warning object. This ensures that the warning object lands exactly at the pedestrian's footing point at the start of the support phase, achieving millisecond-level synchronization between the launch and the pedestrian's gait, thereby improving the success rate of trampling and the warning effect.

[0022] Optionally, an adaptive warning intensity method may also be included: Collect the radio frequency signal strength in the direction of pedestrians; Based on radio frequency signal strength, to extract radio frequency signal modulation features; The method is based on the modulation characteristics of radio frequency signals and a preset mobile communication feature database to determine whether a pedestrian is in an active call state. When the matching result indicates an active call status, the attention distraction coefficient is calculated based on the radio frequency signal strength and the preset reference strength value. Increase the volume of the warning device based on the attention distraction coefficient.

[0023] By adopting the above technical solution, radio frequency signals in the direction of pedestrians are collected and modulation features are extracted to identify whether they are in an active call state. If so, the attention dispersion coefficient is calculated based on the signal strength, and the volume of the warning sound is adaptively increased to compensate for the impact of the call on auditory attention, ensuring that the warning information can be effectively transmitted and improving the reliability of warnings in complex scenarios.

[0024] Secondly, this application provides a dynamic obstacle recognition and avoidance system during drone hoisting, employing the following technical solution: A dynamic obstacle recognition and avoidance system for unmanned aerial vehicle (UAV) hoisting processes includes: The data acquisition module is used to collect early warning image information, the current location of the drone, and trampling signals; The memory is used to store the program that implements a method for dynamic obstacle recognition and avoidance during the hoisting process of an unmanned aerial vehicle (UAV); The processor is used to load and execute programs stored in memory.

[0025] In summary, this application includes at least one of the following beneficial technical effects: 1. By collecting early warning images, pedestrians are identified and their position, speed, and gait phase information are extracted. The gait phase information is combined with the prediction of pedestrian trajectories to improve the accuracy of trajectory prediction. The remaining time for pedestrians to enter the danger zone is calculated based on the dangerous range of the drone hoisting. When the time is less than the threshold, a warning object is actively launched to the landing point on the predicted trajectory. An audible prompt is issued after the pedestrian steps on it, guiding the pedestrian to stop walking. This allows the drone to continue the original hoisting operation without hovering or flying around, thus balancing the continuity and safety of the hoisting operation. 2. In multi-pedestrian scenarios, prioritize launching warning objects at the first pedestrian to enter the danger zone. Utilize the sound generated after the pedestrian steps on the object and the head turning angle and effective sound pressure value of other pedestrians to comprehensively judge the attention capture status. Suppress the launching action for pedestrians who have already voluntarily avoided the danger zone to avoid repeated deployment and improve the warning efficiency and targeting in multi-pedestrian scenarios. 3. First, determine the current supporting foot position and stride length based on the gait phase, and calculate the half-step offset in combination with the gait cycle to accurately predict the pedestrian's next foot landing point, so that the projectile landing point falls on the position that the pedestrian is about to step on under the natural gait, thereby improving the probability of the warning object being stepped on and the accuracy of the warning timing. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating a method for dynamic obstacle identification and avoidance during drone hoisting. Detailed Implementation

[0027] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0028] Reference Figure 1 This application discloses a method for dynamic obstacle identification and avoidance during drone hoisting, including the following steps: S10: Collect early warning image information and the current location of the drone.

[0029] Early warning image information refers to two-dimensional image data that is collected in real time by the visual sensors carried by the UAV, covering the hoisting operation area and its surroundings. This information is used to identify dynamic targets such as pedestrians and obstacles in the environment.

[0030] The current location of a drone refers to its coordinates in three-dimensional space, calculated using the drone's onboard Global Positioning System (GPS). These coordinates can be absolute coordinates expressed in latitude, longitude, and altitude.

[0031] S11: When the warning image information contains preset pedestrian features, obtain the pedestrian's current position, current speed, and gait phase information based on the warning image information and pedestrian features.

[0032] Pedestrian features refer to a set of visual features that are pre-trained using machine learning models (such as convolutional neural networks) to distinguish pedestrians from other moving objects. The specific network structure, training dataset, and parameters of this model are pre-trained and fixed by those skilled in the art based on the actual application scenario.

[0033] The pedestrian's current position refers to the three-dimensional spatial coordinates of the pedestrian in the world coordinate system (the same coordinate system as the drone's position), calculated by performing an inverse perspective transformation on the pedestrian area detected in the warning image information, and combining this with the drone's current position and camera attitude parameters. The specific matrix parameters of the inverse perspective transformation are obtained by calibrating the camera's intrinsic and extrinsic parameters using methods that are common knowledge in the field.

[0034] The pedestrian's current velocity refers to the three-dimensional velocity vector obtained by dividing the change in the pedestrian's current position across multiple consecutive frames of warning image information by the time interval between adjacent frames. To improve accuracy, algorithms such as Kalman filtering can be used to smooth the velocity. The specific filtering parameters are preset by those skilled in the art based on the frame rate and motion noise level.

[0035] Gait phase information refers to data that quantitatively describes the periodic state of alternating leg movements during a pedestrian's walking process. Specifically, by performing temporal analysis on the lower limb joint angles of a pedestrian in a continuous image sequence, a pre-defined gait cycle detection algorithm (such as peak detection based on joint angle trajectories) is used to identify the support phase (foot contacting the ground and bearing weight) and swing phase (foot leaving the ground and stepping forward) within each gait cycle, and outputs the phase at the current moment and its percentage of progress (0% to 100%) within a complete cycle. The specific algorithm parameters for gait detection are pre-calibrated by those skilled in the art based on the normal walking speed range of pedestrians.

[0036] When the warning image information contains pedestrian features, it indicates that there are pedestrians around the drone and avoidance is required. Therefore, it is necessary to first determine the pedestrian's current location, current speed, and gait phase information for subsequent steps.

[0037] S12: Combine the pedestrian's current position, current speed, and gait phase information to predict the pedestrian's trajectory.

[0038] A pedestrian trajectory is a spatiotemporal curve obtained by predicting the position sequence of a pedestrian over a number of seconds (e.g., 2 to 5 seconds) starting from the current moment. This prediction employs a Kalman filter based on a kinematic model combined with gait phase constraints: linear extrapolation is performed using the pedestrian's current speed and position, while the trajectory is synchronously corrected based on periodic speed fluctuations in the gait phase information (deceleration during the support phase and acceleration during the swing phase), thus obtaining a predicted trajectory that better conforms to the laws of human movement. Parameters such as prediction duration and state transition matrix are set by those skilled in the art based on the UAV's reaction time and the warning vehicle's flight time.

[0039] S13: Obtain the hoisting hazard area based on the current location of the drone and the preset hoisting hazard range.

[0040] The hoisting hazard zone refers to the spatial area parameters pre-defined by those skilled in the art, with reference to the drone's hoisting point.

[0041] The hoisting hazard zone refers to the actual geometric area on the ground or in space mapped from the current position of the drone as a reference point according to the hoisting hazard range parameters. It is used to determine whether a collision may occur if a pedestrian enters this area. If the hoisting hazard range includes a lower vertical height limit, then a pedestrian below that lower height limit is also considered to have entered the hazard zone. The specific geometric definition is set by those skilled in the art based on safety margins. The specific mapping method is common knowledge in the art.

[0042] S14: Determine the remaining time for a pedestrian to enter the hoisting hazard area based on the pedestrian's movement trajectory and the hoisting hazard area.

[0043] The remaining time refers to the time when the pedestrian's trajectory first intersects the boundary of the hoisting danger zone, calculating the time of the first intersection, and then subtracting the current time from the calculated time. If the trajectory does not intersect the zone, the remaining time is set to a constant much larger than the first threshold (indicating no risk). This calculation uses linear or quadratic interpolation methods based on discrete points on the trajectory. The specific interpolation algorithms and intersection methods are common knowledge in this field and will not be elaborated here.

[0044] S15: When the remaining time is less than the preset first threshold, determine the landing point of the projectile based on the pedestrian's trajectory.

[0045] The first threshold is a preset time constant. This threshold can be adjusted by the operator before the mission using ground station software, based on the actual scenario.

[0046] The launch landing point refers to the ground location where the UAV needs to launch the warning object. The method for determining the launch landing point will be explained in detail in subsequent sections S70 to S75, and will not be repeated here.

[0047] S16: Control the drone to launch warning objects towards the launch point and collect the trampling signals of the warning objects.

[0048] The warning object is a flexible, harmless, and retrievable physical object (such as a silicone elastic ball with built-in accelerometers and pressure sensors), connected to a winch on the bottom of the drone via a recovery line. Upon launch, a projectile module (such as a compressed pneumatic or electromagnetic catapult) propels the warning object along a calculated trajectory toward the impact point. The initial velocity and angle are calculated by the controller based on the launch distance and the drone's current altitude using a trajectory calculation that takes into account air resistance and the downwash from the drone's rotor blades. The specific trajectory model is common knowledge in the field.

[0049] A trampling signal refers to the characteristic physical quantity generated when a warning object is stepped on by a pedestrian. This signal is identified by a high-amplitude, short-duration pulse waveform (e.g., amplitude greater than a preset acceleration threshold and duration less than 0.3 seconds) detected by an accelerometer inside the warning object. The range and sampling frequency of the accelerometer are pre-calibrated by those skilled in the art based on the weight of the warning object and the hardness of the ground.

[0050] S17: Respond to the stamping signal by emitting an audible alert to prompt pedestrians to stop walking and to allow the drone to continue its hoisting operation along the original flight path.

[0051] Audio alerts refer to warning sounds with specific frequencies and rhythms emitted by the drone's onboard speaker.

[0052] After issuing an audible warning, the drone does not need to take any additional evasive action because the warning object has successfully attracted the pedestrian's attention and caused them to stop or detour. Therefore, it can continue to complete the hoisting task without changing its existing flight path and speed.

[0053] It also includes methods for recovering warning items: S20: The length of the collection and recovery line and the distance between the warning device and the drone.

[0054] The release length value refers to the total length of the recovery line actively released by the winch drum from the initial storage state to the current moment, as measured in real time by the winch encoder in the recovery module.

[0055] The interval distance value refers to the three-dimensional straight-line distance between the current location of the warning object and the reference point of the drone body, as measured by the laser rangefinder on the drone. This distance is read by the laser rangefinder.

[0056] S21: Calculate the current relaxation length of the recovery line based on the release length value and the interval distance value.

[0057] The current slack length refers to the difference between the actual path length of the retrieval line from the winch to the warning object and the straight-line distance between the warning object and the drone. This difference represents the extra length occupied by the retrieval line in space due to bending, sagging, or redundancy. The specific calculation formula is: Current slack length = Release length value - Interval distance value. When this value is positive, it indicates that there is excess slack in the retrieval line, and the winch needs to tighten this amount of slack before starting retrieval; when this value is negative or zero, it indicates that the retrieval line is in a taut state, and careful retrieval is required to prevent jamming.

[0058] S22: Collect the real-time drive current value of the winch in the recovery module, and determine the recovery resistance coefficient based on the real-time drive current value.

[0059] The real-time drive current value refers to the instantaneous current flowing into the motor windings, which is fed back in real time by the electronic speed controller driving the winch motor. The current value is obtained through a sampling resistor connected in series in the motor power supply circuit.

[0060] The recovery resistance coefficient is a dimensionless parameter representing the total resistance of the current recovery path, obtained by calibrating and converting the real-time drive current value. This coefficient is derived from a pre-set current resistance mapping table, which records the recovery resistance coefficients corresponding to different real-time drive current values. The current resistance mapping table is pre-determined by those skilled in the art through extensive calibration experiments and is embedded in the controller.

[0061] S23: Obtain the target recovery rate value based on the current relaxation length and recovery resistance coefficient.

[0062] The target recovery speed refers to the recovery line speed that the winch should achieve. This speed value is derived from a preset speed mapping table, which records the target recovery speed values ​​corresponding to different combinations of current slack lengths and recovery resistance coefficients. The speed mapping table is preset by those skilled in the art based on the recovery line material, winch motor characteristics, and typical ground resistance relationships.

[0063] S24: Control the recycling module to recycle the warning body at the target recycling speed value.

[0064] The control module retrieves the warning object at the target retrieval speed value, thereby completing the retrieval action of the warning object.

[0065] Also includes: S30: Collects the real-time speed value of the recovery line and the real-time tension value of the recovery line after recovery is started.

[0066] The real-time speed value refers to the current actual linear speed of the recovery line, which is obtained by measuring the drum angular velocity in real time through the winch encoder and converting it according to the drum radius (pre-set by those skilled in the art).

[0067] Real-time tension value refers to the instantaneous value of the tension borne by the recovery line, which is measured in real time by a tension sensor installed on the recovery line path.

[0068] S31: Calculate the speed deviation based on the real-time speed value and the target recovery speed value.

[0069] Speed ​​deviation refers to the difference between the real-time speed value and the target recovery speed value. The specific calculation formula is: Speed ​​Deviation = Target Recovery Speed ​​Value - Real-time Speed ​​Value. A positive deviation value indicates that the actual speed is lower than the target speed, requiring acceleration; a negative deviation value indicates that the actual speed is higher than the target speed, requiring deceleration.

[0070] S32: Calculate the tension influence factor based on the real-time tension value and the preset benchmark tension value.

[0071] The reference tension value refers to the pre-calibrated tension of the recovery line under ideal conditions with no additional resistance during the alarm body recovery process. This value is obtained by averaging multiple no-load recovery tests and is preset and fixed in the controller by those skilled in the art based on the recovery line material and winch characteristics.

[0072] The tension influence factor is a dimensionless coefficient used to quantify the impact of current recovery resistance on recovery speed control, obtained by comparing the real-time tension value with the reference tension value. The specific calculation formula is: Tension Influence Factor = (Real-time Tension Value - Reference Tension Value) / Reference Tension Value. A factor greater than 0 indicates the presence of additional resistance, requiring a reduction in recovery speed; a factor less than 0 indicates that the resistance is less than the reference (e.g., a slack recovery line), allowing for an increase in speed.

[0073] S33: Obtain the recovery speed correction amount based on the speed deviation and tension influence factor.

[0074] The recovery speed correction is a compensation value calculated based on the current speed deviation and the degree of tension influence, used to adjust the target recovery speed value. The specific calculation method is: Recovery speed correction = Speed ​​deviation × Proportional coefficient + Tension influence factor × Tension compensation coefficient. The proportional coefficient and tension compensation coefficient are pre-calibrated by those skilled in the art based on the response speed and stability requirements of the closed-loop control system.

[0075] S34: Add the target recovery speed value to the recovery speed correction amount to obtain the dynamic recovery speed value, and control the recovery module to continue to recover the warning body at the dynamic recovery speed value.

[0076] The dynamic recovery speed value refers to the new speed setpoint used for real-time control of the winch, obtained by algebraically summing the current target recovery speed value and the calculated recovery speed correction. The calculation formula is: Dynamic recovery speed value = Target recovery speed value + Recovery speed correction. The controller uses this dynamic recovery speed value as the new control target, causing the actual speed of the recovery line to follow this dynamic value. This compensates for speed deviations caused by nonlinear factors such as ground friction and cable bending during the recovery process, improving the stability and reliability of the recovery.

[0077] It also includes methods for handling multi-user scenarios: S40: Collect the number of pedestrians in the hazardous area of ​​the hoisting operation.

[0078] The number of pedestrians refers to the total number of individual pedestrians currently located within the hoisting hazard area, calculated by combining early warning image information with a pedestrian detection algorithm. The pedestrian detection algorithm is pre-trained and fixed by those skilled in the art based on the application scenario.

[0079] S41: When the number of pedestrians is greater than 1, collect the entry time and location of each pedestrian.

[0080] The pedestrian entry time refers to the remaining time from the current moment until each pedestrian is expected to enter the hoisting hazard area. The method for calculating this remaining time is the same as that for determining the remaining time for pedestrians to enter the hoisting hazard area in S14.

[0081] The position of each pedestrian refers to the three-dimensional spatial coordinates of each pedestrian in the world coordinate system at the current moment, and the method of obtaining it is the same as the current position of the pedestrian in S11.

[0082] S42: Select priority launch targets based on the entry time of each pedestrian.

[0083] The priority target is the pedestrian with the shortest entry time (i.e., remaining time). If multiple pedestrians have the same entry time, the one closest to the drone is selected as the priority target.

[0084] S43: Determine the launch landing point based on the priority launch target, and control the UAV to launch a warning object towards the launch landing point.

[0085] The method for determining this target is the same as in S15, except that the target pedestrian is replaced with a priority target for launching. The control procedure for launching the warning device is the same as in S16, and will not be described in detail here.

[0086] S44: Collect the response sound pressure level and the head deflection angle of other pedestrians after the priority target is stepped on the warning object.

[0087] The sound pressure level response (SPR) refers to the sound pressure level of the sound emitted by a pedestrian, such as a gasp or footsteps, after the warning object has been deployed to the target and then stepped on. The SPR is calculated from the effective value of the microphone signal using a method known in the field.

[0088] Head tilt angle refers to the angle at which another pedestrian's head rotates relative to the positive direction of their body. This angle is obtained by acquiring key head posture points (such as the eyes and the tip of the nose) using visual sensors and calculating the rotation matrix.

[0089] S45: Calculate the effective sound pressure level at other pedestrian locations based on the response sound pressure level and the location of each pedestrian.

[0090] The effective sound pressure level (SPL) is the estimated sound pressure level after distance attenuation when the response sound generated by the prioritized target propagates to the location of other pedestrians. The specific calculation formula is: Effective SPL = Response Sound Pressure Level − 20 × log10(Distance / Reference Distance), where the distance is the Euclidean distance between other pedestrians and the prioritized target, and the reference distance is taken as 1 meter. This calculation is based on a spherical wave diffusion attenuation model, ignoring obstruction and reflection, and is performed in real time by the controller.

[0091] S46: Determine the attention capture status of each other pedestrian based on the head deflection angle and the preset attention deflection threshold.

[0092] Attention deflection threshold refers to the angular limit at which a pedestrian's attention has been turned toward the primary target. This threshold is preset by a person skilled in the art based on the typical angle at which humans naturally turn their heads to observe others.

[0093] Attention capture status refers to the binary judgment result obtained by comparing the head deflection angle of each other pedestrian with an attention deflection threshold. When the head deflection angle is greater than or equal to the attention deflection threshold, the status is recorded as "captured"; otherwise, it is recorded as "not captured". This status is used to characterize whether the pedestrian has visually noticed the abnormal reaction of the priority target.

[0094] S47: When the effective sound pressure value is greater than the preset alertness threshold and the attention capture state is captured, it is determined that the corresponding pedestrian has voluntarily avoided the situation, and the throwing action against the pedestrian is suppressed.

[0095] The alertness threshold refers to the lower limit of the sound pressure level at which other pedestrians can perceive the response sound of the prioritized projectile target through hearing. This threshold is preset by those skilled in the art based on the ambient background noise level and the characteristics of human auditory sensitivity.

[0096] Suppressing the launching action against this pedestrian means that for other pedestrians who meet the above conditions, the system marks them as having autonomously avoided the obstacle, and no longer performs launching point determination, warning object launch, or subsequent related operations against them. This suppression logic is independently judged for each person in a multi-person scenario, and launching permissions are only retained for other pedestrians who do not meet the conditions.

[0097] It also includes a group alert confirmation method based on the content of the response sound: S50: Collect audio segments of the response sound generated after the target is prioritized to step on the warning object.

[0098] A voice clip refers to a continuous audio data segment, typically 1 to 3 seconds long, collected by the drone's onboard microphone after the pedestrian has stepped on the warning object following its initial deployment. This segment is usually captured automatically using an acoustic activity detection algorithm.

[0099] S51: Extract the fundamental frequency change rate and short-time energy value from the speech segment, and calculate the intonation excitation index based on the fundamental frequency change rate and short-time energy value.

[0100] The fundamental frequency change rate refers to the rate at which the fundamental frequency of vocal cord vibration changes over time in a speech segment. It is obtained by extracting the fundamental frequency trajectory using the cepstral method and then calculating the mean value of the fundamental frequency difference between adjacent frames.

[0101] Short-time energy value refers to the sum of squares of signal energy of a speech segment within a preset short-time window (e.g., 20 milliseconds). It is obtained by summing the squares of the amplitudes of the sampling points within the window.

[0102] The intonation arousal index is a dimensionless parameter used to quantify the degree of emotional arousal in pedestrians, obtained by weighted fusion of the fundamental frequency change rate and short-time energy value. The specific calculation formula is: Intonation Arousal Index = α × Normalized Fundamental Frequency Change Rate + β × Normalized Short-Time Energy Value, where α and β are preset weighting coefficients, pre-set by those skilled in the art based on the contribution of acoustic characteristics to emotion recognition.

[0103] S52: Match the audio segment with a preset warning word library to obtain a matching score.

[0104] A warning vocabulary database refers to a pre-stored collection of words or phrases containing semantics such as surprise, warning, and reminder, such as "Ah," "Be careful," "Danger," and "Get out of the way." This vocabulary database is pre-constructed by those skilled in the art based on the common language response habits of people in emergency situations.

[0105] The matching score refers to the similarity metric obtained by fuzzy matching of a speech segment converted into text through speech recognition with words in a warning word database. The value ranges from 0 to 1. A higher matching score indicates a greater likelihood that the pedestrian's voice contains warning words. The specific matching algorithm is common knowledge in this field.

[0106] S53: Confidence of audio content is obtained based on intonation excitability index and matching degree score.

[0107] Voice content confidence refers to the comprehensive credibility of a pedestrian's response voice as a warning, obtained by weighted fusion of the intonation arousal index and the matching score. The value ranges from 0 to 1. The calculation formula is: Voice content confidence = γ × intonation arousal index + δ × matching score, where γ and δ are preset weighting coefficients, pre-set by those skilled in the art based on the importance of the two features in warning judgment.

[0108] S54: When the confidence level of the sound content is greater than the preset content threshold, generate a group warning signal and suppress the throwing action to all other pedestrians based on the signal.

[0109] The content threshold is a threshold value used to determine whether the confidence level of sound content is sufficient to confirm the effectiveness of a group warning. This threshold was obtained by those skilled in the art through statistical analysis of a large amount of experimental data.

[0110] A group warning valid signal is a digital trigger signal generated by the controller when the confidence level of the sound content exceeds a content threshold. This signal indicates that the response sound of the priority target is sufficient to make other pedestrians perceive the danger and may actively avoid it, so the system no longer needs to throw warning objects to other pedestrians one by one. In response to this signal, the controller marks the throwing actions of all other pedestrians as suppressed until the current hoisting task is completed or a new pedestrian enters the danger zone.

[0111] It also includes a method for verifying projectile suppression: S60: Collects the moment of head deflection and the trigger moment of the trampling signal from the priority projectile target.

[0112] The head deflection moment refers to the timestamp recorded by the system when the head deflection angle of another pedestrian first reaches or exceeds the preset attention deflection threshold (see S46). This moment is obtained by continuously tracking the head posture angle and at the frame moment when the angle value exceeds the threshold.

[0113] The trigger time refers to the precise timestamp at which the stamping signal (see S16) is collected by the controller after the target stamping warning object is launched first.

[0114] S61: Calculate the reaction delay time based on the head deflection time and the trigger time.

[0115] Reaction delay time refers to the time interval between when the target is kicked and a sound is emitted (trigger moment) and when the pedestrian begins to turn their head (head deflection moment). The formula is: Reaction delay time = Head deflection moment - Trigger moment.

[0116] S62: Combine pedestrian locations, priority targets, and warning image information to determine pedestrian distance values.

[0117] The pedestrian distance value refers to the straight-line distance between other pedestrians and the priority target. It is calculated using the Euclidean distance formula based on the positions of each pedestrian and the coordinates of the priority target. Warning image information is used to assist in confirming that there are no obstructions between the two (e.g., using image recognition to determine if there are obstacles in the line-of-sight path). If obstruction exists, the distance value is not considered valid input (or is marked as unavailable). Obstruction determination methods are common knowledge in this field and will not be elaborated upon here.

[0118] S63: Calculate the speed of line of sight propagation based on pedestrian distance and reaction delay time.

[0119] The speed of visual attention propagation refers to the equivalent speed at which a pedestrian's attention "spreads" when the pedestrian prioritizes visual attention to a target. The formula is: Speed ​​of visual attention propagation = Pedestrian distance / Reaction delay time.

[0120] S64: When the line-of-sight propagation speed falls within the preset normal propagation speed range, a projectile suppression signal is generated.

[0121] The normal propagation speed range refers to a pre-defined interval consisting of a lower and upper limit, based on the natural reaction speed of humans turning their heads after hearing a sudden sound. This range is obtained by those skilled in the art through experimental measurements of typical reaction times of pedestrians traveling at different distances.

[0122] A projection suppression signal is a control flag generated by the controller when the speed of line-of-sight propagation is within the normal range. This signal indicates that the other pedestrian's gaze shift is in accordance with natural propagation patterns and is not an accidental action; therefore, it can be concluded that they have consciously noticed the unusual event.

[0123] S65: In response to the projectile suppression signal, suppress projectile actions against all other pedestrians who have not yet turned their heads.

[0124] In response to the projectile suppression signal, the controller cancels subsequent operations such as projectile landing point determination, warning object launch, and retrieval for all other pedestrians who have not yet experienced head deflection. Pedestrians who have already experienced head deflection were dealt with in S47 or prior to this step. This suppression logic is only effective for the current detection cycle; if a new pedestrian subsequently enters the hoisting hazard area, a reassessment will be performed.

[0125] It also includes methods for determining the launch landing point: S70: Determine the direction of pedestrian movement based on the pedestrian's trajectory.

[0126] The pedestrian's direction of travel refers to the direction of movement of the pedestrian at the current moment, extracted from the pedestrian's trajectory. Specifically, it is calculated by taking the difference between the pedestrian's current position and the previous position on the trajectory, normalizing it, and then obtaining the direction of travel vector.

[0127] S71: Determine the gait phase period based on gait phase information.

[0128] Gait phase period refers to the time interval between two adjacent points of the same phase during a pedestrian's walking process. Specifically, it is calculated by extracting the timestamps of two consecutive points of the same phase from the timestamp sequence of gait phase information and calculating their difference.

[0129] S72: Determine the current supporting foot position based on gait phase information and the pedestrian's current position.

[0130] The current supporting foot position refers to the coordinates of the ground projection point of the foot that is currently in contact with and bearing weight on the ground. The specific determination method is as follows: Based on gait phase information, the foot currently in the supporting phase (left or right) is identified. Then, combining the pedestrian's current position, gait phase period, and historical trajectory, a kinematic model is used to calculate the relative offset between this foot and the pedestrian's center of mass, thereby obtaining the world coordinates of the supporting foot. This model is pre-calibrated by those skilled in the art based on average human body dimensions.

[0131] S73: Stroke length is obtained based on the pedestrian's current speed and gait phase period.

[0132] Stride length refers to the distance a pedestrian covers in one step. The formula is: Stride length = Pedestrian's current speed × Gait phase period.

[0133] S74: Determine the half-step offset by combining the current support foot position and stride length.

[0134] Half-step offset refers to half the stride length. The calculation formula is: Half-step offset = Stride length / 2.

[0135] S75: Determine the launch landing point based on the pedestrian's direction of travel, half-step offset, and current support foot position.

[0136] The calculation formula is: Projectile landing point = Current supporting foot position + Half-step offset × Pedestrian's walking direction vector. Wherein, the pedestrian's walking direction vector is a unit vector. The landing point is located half a step in front of the pedestrian's current supporting foot; when the pedestrian takes a step forward, their front foot will naturally step on the warning object.

[0137] It also includes methods for synchronizing the launch timing: S80: Determine the current phase value based on gait phase information.

[0138] The current phase value refers to the percentage of progress a pedestrian is in within the gait cycle at the current moment, extracted from gait phase information. This value is obtained by temporal analysis of joint angles in consecutive image frames using a gait detection algorithm. The specific algorithm is trained and solidified in advance by those skilled in the art.

[0139] S81: Predict the start time of the support phase based on the gait phase period and the current phase value.

[0140] The support phase start time refers to the time when the pedestrian's next support phase begins. The prediction formula is: Support phase start time = Current time + Gait phase period × (1 − Current phase value / 100).

[0141] S82: Combine the launch landing point and the current position of the drone to determine the estimated flight time.

[0142] The estimated flight time refers to the anticipated duration from launch to landing of the warning object. It is calculated as follows: the straight-line distance between the launch landing point and the current position of the UAV is calculated, divided by the average initial launch velocity of the warning object, and multiplied by a preset trajectory correction factor. This correction factor and the average initial launch velocity of the warning object are pre-calibrated by those skilled in the art.

[0143] S83: Calculate the launch trigger time based on the support phase start time and the expected flight time.

[0144] The launch trigger time refers to the system time point at which the UAV should launch the warning object when it lands at the start of the support phase. The calculation formula is: Launch trigger time = Start of support phase - Estimated flight time. If the calculated result is less than the current time, the current time is taken as the trigger time.

[0145] S84: When the system time reaches the launch trigger moment, control the launch module to emit an alarm.

[0146] The controller compares the system clock with the launch trigger time in real time. When the system time is greater than or equal to the launch trigger time, it sends a launch command to the launch module to control the launch module to launch the warning body.

[0147] It also includes an adaptive warning intensity method: S90: Collects the radio frequency signal strength in the direction of pedestrians.

[0148] Radio frequency signal strength refers to the electromagnetic radiation level value emitted by electronic devices carried by a pedestrian, collected by a directional antenna mounted on a drone in the direction of the pedestrian.

[0149] S91: Extracting radio frequency signal modulation features based on radio frequency signal strength.

[0150] Radio frequency (RF) signal modulation characteristics refer to the set of parameters extracted from RF signals that reflect the signal modulation method and transmission mode, including but not limited to the time-domain fluctuation pattern of the signal envelope, frequency-domain peak interval, and pulse period. These characteristics are extracted through spectrum analysis and demodulation algorithms, the specific algorithms of which are common knowledge in the field.

[0151] S92: Determine whether a pedestrian is in an active call state based on the radio frequency signal modulation characteristics and a preset mobile phone communication feature library.

[0152] A mobile communication feature library refers to a set of radio frequency signal modulation feature templates pre-stored in the system for active call states under different communication standards. This library is established by those skilled in the art by collecting signal samples from real call scenarios and extracting features.

[0153] An active call state refers to a situation where a pedestrian is making a voice call using their mobile phone, and in this state, the pedestrian's attention is distracted. The determination method is as follows: the collected radio frequency signal modulation features are matched with templates in the feature library. If the matching degree exceeds a preset threshold (set in advance by those skilled in the art), it is determined to be an active call state.

[0154] S93: When the matching result indicates an active call status, calculate the attention distraction coefficient based on the radio frequency signal strength and the preset reference strength value.

[0155] The reference strength value refers to the background strength of the radio frequency signal measured at a typical distance when a pedestrian is not using a mobile phone. This value is pre-calibrated by those skilled in the art through environmental scanning.

[0156] The attention distraction coefficient is a dimensionless parameter used to quantify the degree to which a pedestrian's attention is distracted by a phone call. The formula is: Attention Distraction Coefficient = (RF Signal Strength − Reference Strength Value) / Reference Strength Value, with the result limited to between 0 and 1. A higher coefficient indicates that the pedestrian's attention is more likely to be occupied by their phone.

[0157] S94: Increase the volume of the warning sound based on the attention distraction coefficient.

[0158] Increasing the volume of the warning device based on the attention distraction coefficient means multiplying a preset baseline volume by (1 + attention distraction coefficient) to obtain the target volume. The controller sends this target volume to the onboard speaker, making the warning device's sound louder than usual, thus penetrating pedestrians' attention. The maximum volume is limited by the drone's power supply and safety regulations, and the upper limit is preset by those skilled in the art. The baseline volume is also preset by those skilled in the art.

[0159] Based on the same inventive concept, embodiments of the present invention provide a dynamic obstacle recognition and avoidance system during drone hoisting, comprising: The data acquisition module is used to acquire early warning image information, the current position of the drone, trampling signals, release length value, interval distance value, real-time drive current value, real-time speed value, real-time tension value, number of pedestrians, entry time of each pedestrian, position of each pedestrian, response sound pressure level, head deflection angle, voice segments, head deflection time, trigger time, and radio frequency signal strength. The memory is used to store the program that implements a method for dynamic obstacle recognition and avoidance during the hoisting process of an unmanned aerial vehicle (UAV); The processor is used to load and execute programs stored in memory.

[0160] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0161] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for dynamic obstacle identification and avoidance in a UAV hoisting process, characterized in that, include: Collect early warning image information and the current location of the drone; When the warning image information contains preset pedestrian features, the pedestrian's current position, current speed, and gait phase information are obtained based on the warning image information and pedestrian features. By combining the pedestrian's current location, current speed, and gait phase information, the pedestrian's trajectory can be predicted. The hoisting hazard zone is determined based on the drone's current location and the preset hoisting hazard range. The remaining time for a pedestrian to enter the hoisting danger zone is determined based on the pedestrian's movement trajectory and the hoisting danger zone. When the remaining time is less than a preset first threshold, the landing point of the projectile is determined based on the pedestrian's trajectory. Control the drone to launch warning objects towards the launch point and collect the trampling signals of the warning objects; The drone responds to the stampede signal by emitting an audible alert to prompt pedestrians to stop and continues to perform the hoisting operation along its original flight path.

2. The method for dynamic obstacle identification and avoidance during UAV hoisting as described in claim 1, characterized in that, It also includes methods for recovering warning items: The length of the collection and recovery line and the distance between the warning device and the drone; The current relaxation length of the recovery line is calculated based on the release length value and the interval distance value; The real-time drive current value of the winch in the recovery module is collected, and the recovery resistance coefficient is determined based on the real-time drive current value. The target recovery rate value is obtained based on the current relaxation length and recovery resistance coefficient; The control recycling module recycles the warning body at the target recycling speed value.

3. The method for dynamic obstacle identification and avoidance during UAV hoisting as described in claim 2, characterized in that, Also includes: Collect the real-time speed value of the recovery line and the real-time tension value of the recovery line after the recovery is started; The velocity deviation is calculated based on the real-time velocity value and the target recovery velocity value; The tension influence factor is calculated based on the real-time tension value and the preset benchmark tension value. The recovery speed correction is obtained based on the speed deviation and tension influence factor; The target recovery speed value is added to the recovery speed correction value to obtain the dynamic recovery speed value, and the recovery module is controlled to continue to recover the warning body at the dynamic recovery speed value.

4. The method for dynamic obstacle identification and avoidance during UAV hoisting as described in claim 1, characterized in that, It also includes methods for handling multi-user scenarios: Collect the number of pedestrians in the hazardous area of ​​the hoisting operation; When the number of pedestrians is greater than 1, collect the entry time and location of each pedestrian. Priority targets are selected based on the time each pedestrian enters. The launch point is determined based on the priority launch target, and the drone is controlled to launch a warning object toward that launch point; The sound pressure level of the response generated after the priority target tramples the warning object, as well as the head deflection angle of other pedestrians, are collected. The effective sound pressure level at other pedestrian locations is calculated based on the response sound pressure level and the location of each pedestrian. The head deflection angle and the preset attention deflection threshold are used to determine the attention capture status of each other pedestrian; When the effective sound pressure value is greater than the preset alertness threshold and the attention capture state is captured, it is determined that the corresponding pedestrian has voluntarily avoided the situation, and the throwing action against the pedestrian is suppressed.

5. The method for dynamic obstacle identification and avoidance during UAV hoisting as described in claim 4, characterized in that, It also includes a group alert confirmation method based on the content of the response sound: Collect audio clips of the response sounds generated after the target is prioritized to step on the warning object; The fundamental frequency change rate and short-time energy value are extracted from the speech segment, and the intonation excitation index is calculated based on the fundamental frequency change rate and short-time energy value; The audio clips are matched with a pre-defined warning word list to obtain a matching score. The confidence level of the audio content is obtained based on the intonation arousal index and the matching score. When the confidence level of the sound content is greater than the preset content threshold, a group warning signal is generated, and the throwing action to all other pedestrians is suppressed based on the signal.

6. The method for dynamic obstacle identification and avoidance during UAV hoisting as described in claim 4, characterized in that, It also includes a method for verifying projectile suppression: Collect the moment of head deflection and the trigger moment of the trampling signal from the priority projectile target; The reaction delay time is calculated based on the head deflection time and the trigger time. The pedestrian distance value is determined by combining the positions of each pedestrian, the priority target for projectile, and the early warning image information. The speed of line-of-sight propagation is calculated based on pedestrian distance and reaction delay time; When the speed of line-of-sight propagation falls within the preset normal propagation speed range, a projectile suppression signal is generated; In response to the projectile suppression signal, projectile actions against all other pedestrians who have not yet turned their heads are suppressed.

7. The method for dynamic obstacle identification and avoidance during UAV hoisting as described in claim 1, characterized in that, It also includes methods for determining the launch landing point: The direction of pedestrian movement is determined based on the pedestrian's trajectory. Gait phase period is determined based on gait phase information; The current supporting foot position is determined based on gait phase information and the pedestrian's current position; The stride length is obtained based on the pedestrian's current speed and gait phase period; Determine the half-step offset by combining the current support foot position and stride length; The landing point of the projectile is determined based on the pedestrian's direction of travel, half-step offset, and the current position of the supporting foot.

8. The method for dynamic obstacle identification and avoidance during UAV hoisting as described in claim 7, characterized in that, It also includes methods for synchronizing the launch timing: The current phase value is determined based on gait phase information; Predict the start time of the support phase based on the gait phase period and the current phase value; The estimated flight time is determined by combining the launch landing point and the current position of the drone; The ejection trigger time is calculated based on the support phase start time and the expected flight time. When the system time reaches the launch trigger moment, the launch module is controlled to emit an alarm.

9. The method for dynamic obstacle identification and avoidance during UAV hoisting as described in claim 1, characterized in that, It also includes an adaptive warning intensity method: Collect the radio frequency signal strength in the direction of pedestrians; Based on radio frequency signal strength, to extract radio frequency signal modulation features; The method is based on the modulation characteristics of radio frequency signals and a preset mobile communication feature database to determine whether a pedestrian is in an active call state. When the matching result indicates an active call status, the attention distraction coefficient is calculated based on the radio frequency signal strength and the preset reference strength value. Increase the volume of the warning device based on the attention distraction coefficient.

10. A dynamic obstacle recognition and avoidance system during unmanned aerial vehicle (UAV) hoisting process, characterized in that, include: The data acquisition module is used to collect early warning image information, the current location of the drone, and trampling signals; A memory for storing a program that implements a method for dynamic obstacle recognition and avoidance during unmanned aerial vehicle (UAV) hoisting as described in any one of claims 1 to 9; The processor is used to load and execute programs stored in memory.

Images