Robot cooperative control method and control system

Abstract

The invention discloses a robot cooperative control method and system, and the method comprises the steps: carrying out the dynamic adjustment of the path planning of a plurality of robots based on the real-time sensor data in a complex environment, so as to achieve the efficient obstacle avoidance; communication links among different robots are intelligently managed, so that communication delay is reduced, and the precision of synchronous task execution is improved; the emergency response efficiency is optimized according to a task priority dynamic adjustment algorithm; fusion processing parameters of sensor data are adjusted to reduce environmental perception errors and improve perception accuracy. The method has the beneficial effects that the cooperative control precision among the multiple robots is improved, the multiple robots can be matched with one another to execute tasks in a complex environment, and the solving efficiency of the robots to sudden obstacles is improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a robot collaborative control method and control system. Background Art

[0002] With the development of industrial automation, more and more companies are introducing robots to perform tasks in order to improve production efficiency and reduce labor costs. However, in actual applications, a single robot often cannot meet complex operational requirements and needs to work with other robots or people to achieve specific goals. In this case, it is necessary to design and plan the collaborative working mode between robots.

[0003] At present, most existing robot collaborative control systems adopt centralized control mode, that is, all robots are controlled by a central controller. Although this control mode can ensure the stability of the overall system, since each robot needs to exchange information with the central controller through the communication network, it is prone to communication delays and data loss, which affects the performance and reliability of the entire system. In addition, when the system scale is large, the centralized control mode is also difficult to adapt to the ever-changing working environment and task requirements, lacking flexibility and scalability. Summary of the invention

[0004] The purpose of the present invention is to provide a robot collaborative control method to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] In a first aspect, the present invention provides a robot collaborative control method, comprising:

[0007] Based on real-time sensor data in complex environments, dynamic adjustments are made to the path planning of multiple robots to achieve efficient obstacle avoidance;

[0008] Intelligent management of communication links between different robots to reduce communication delays and improve the accuracy of synchronous task execution;

[0009] Dynamically adjust the algorithm based on task priority to optimize the efficiency of emergency response;

[0010] Adjust the fusion processing parameters of sensor data to reduce environmental perception errors and improve perception accuracy.

[0011] Preferably, the specific steps of dynamically adjusting the path planning of multiple robots based on real-time sensor data in a complex environment to achieve efficient obstacle avoidance include:

[0012] Monitor the location of dynamic and static obstacles in the environment;

[0013] Use artificial intelligence algorithms to predict potential collision points and generate alternative routes;

[0014] Adjust the robot trajectory based on the error value fed back by the sensor;

[0015] If the distance d is greater than the threshold T, the shortest paths of all affected robots are replanned, where d represents the minimum distance between the robot and the obstacle, and T represents the preset safety distance threshold.

[0016] Preferably, the dynamically adjusting the multi-robot path planning based on real-time sensor data in a complex environment further comprises the following steps:

[0017] Collect raw data from multiple sensor types;

[0018] Perform data denoising and preliminary filtering;

[0019] Use a multi-layer neural network model to evaluate the safety level S of each perception area;

[0020] If the safety level S is not less than the minimum acceptance standard M, the corresponding area is allowed to continue to be the candidate path selection range, where S represents the calculated safety index and M represents the lower limit of the tolerable risk level.

[0021] Preferably, the multi-robot path planning based on real-time sensor data analysis in a complex environment is specifically as follows:

[0022] Define the boundary R of the dangerous area and record the most recently detected abnormal event type A in the area;

[0023] Assign weights W to different areas and accumulate scores to evaluate the overall risk situation;

[0024] Decide whether to bypass a specific high-risk area based on the accumulated score;

[0025] If the total score Z at the current time t is greater than the warning threshold C, the comprehensive avoidance strategy is triggered to update the entire formation route, where t represents the current moment, Z represents the accumulated dangerous area impact factor, and C represents the triggering alarm threshold.

[0026] Preferably, a more sophisticated way of optimizing the multi-robot navigation system based on real-time sensor input is:

[0027] Collect the information returned by the latest laser scanner to build a 3D point cloud map F;

[0028] Compare the previous map information H and the newly created map G to calculate the change ratio P;

[0029] Determine the intensity of action to be taken based on the degree of change I;

[0030] If the ratio ρ of the change ΔH to H exceeds the specified limit L, the detection frequency is increased or additional resources are requested, where ρ = ΔH / H, ΔH is the difference between the current and previous maps, and L is the pre-set maximum change limit.

[0031] Preferably, the refinement and adjustment scheme based on the continuously updated real-time sensing results includes but is not limited to the following operations:

[0032] Accurately locate the current position of each member and mark Q in the electronic grid;

[0033] Analyze past trajectory data to mine pattern characteristics and learn potential risk signals;

[0034] Propose targeted preventive suggestionsJ to reduce the potential threat levelD;

[0035] If the behavior observed within the specified time window w deviates from the expected probability q by more than the normal fluctuation range N (i.e., q>N), a warning is issued and corresponding avoidance measures are prepared to be implemented. w, q, and N represent the length of the inspection cycle, the frequency of deviations, and the allowable range of variation, respectively.

[0036] Preferably, a more intelligent path adjustment mechanism is implemented based on real-time sensor data in a complex environment:

[0037] Integrate multiple information sources to output comprehensive environmental cognition B;

[0038] Apply adaptive filtering technology to improve positioning accuracy E;

[0039] Improve future hedging plans based on historical experienceC;

[0040] If the remaining visibility r is less than the minimum visibility V when encountering an unknown obstacle, switch to the backup low-light mode or sonar-assisted detection system, where r and V represent the actual visual conditions on site and the minimum requirements to ensure safe navigation, respectively.

[0041] In a second aspect, the present invention provides a robot collaborative control system, which is used to implement the method described in any of the above embodiments, including:

[0042] A path adjustment unit, the adjustment unit is used to dynamically adjust the path planning of multiple robots based on real-time sensor data in a complex environment to achieve efficient obstacle avoidance;

[0043] A link management unit, which intelligently manages the communication links between different robots to reduce communication delays and improve the accuracy of synchronous task execution;

[0044] An algorithm optimization unit, wherein the algorithm optimization unit dynamically adjusts the algorithm according to the task priority to optimize the emergency response efficiency;

[0045] A data processing unit adjusts the fusion processing parameters of the sensor data to reduce environmental perception errors and improve perception accuracy.

[0046] In a third aspect, the present invention provides an electronic device comprising at least one processor, wherein the processor is communicatively connected to at least one memory, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute any one of the methods described above.

[0047] In a fourth aspect, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to implement any of the above-mentioned methods when executed by a processor.

[0048] The disclosed embodiment of the present invention provides a robot collaborative control method, including: dynamically adjusting the path planning of multiple robots based on real-time sensor data in complex environments to achieve efficient obstacle avoidance; intelligently managing the communication links between different robots to reduce communication delays and improve the accuracy of synchronous task execution; dynamically adjusting the algorithm according to task priority to optimize the efficiency of emergency response; adjusting the fusion processing parameters of sensor data to reduce environmental perception errors and improve perception accuracy. The solution of the disclosed embodiment can solve the following problems:

[0049] 1. How to conduct real-time control of the path planning of multiple robots to solve the problem of efficient obstacle avoidance in complex environments.

[0050] 2. How to regulate the communication delay between different robots to solve the accuracy problem of synchronous task execution.

[0051] 3. How to regulate the priority adjustment in the task allocation algorithm to solve the problem of inefficient emergency response.

[0052] 4. How to adjust and regulate the fusion processing method of sensor data to solve the problem of excessive environmental perception errors.

[0053] 5. How to dynamically adjust the energy management strategy to solve the problem of insufficient battery life for long-term tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 is a flow chart of a method according to an embodiment of the present invention;

[0055] Figure 2This is one of the flow charts of path planning for multiple robots in a complex environment according to an embodiment of the present invention;

[0056] Figure 3 This is the second flow chart of path planning for multiple robots in a complex environment according to an embodiment of the present invention;

[0057] Figure 4 This is a flow chart of sensor data analysis and multi-robot path planning in a complex environment of the present invention;

[0058] Figure 5 This is a flowchart of the refinement method of robot navigation in an embodiment of the invention;

[0059] Figure 6 A flow chart of a sensor result adjustment scheme that is continuously updated and implemented for an embodiment of the present invention;

[0060] Figure 7 This is a flowchart of the intelligent path adjustment mechanism according to an embodiment of the present invention;

[0061] Figure 8 is a system module diagram of an embodiment of the present invention;

[0062] Fig. 9 This is a module diagram of an electronic device according to the present invention. DETAILED DESCRIPTION

[0063] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] See also Figure 1 The present invention provides a technical solution: a robot collaborative control method. The method comprises:

[0065] S100: Dynamically adjust multi-robot path planning to achieve efficient obstacle avoidance based on real-time sensor data in complex environments.

[0066] S200: Intelligently manage the communication links between different robots to reduce communication delays and improve the accuracy of synchronous task execution.

[0067] S300: Dynamically adjust the algorithm according to task priority to optimize the efficiency of emergency response.

[0068] S400: By adjusting the fusion processing parameters of sensor data, the environmental perception error is reduced and the perception accuracy is improved. By dynamically adjusting the energy consumption management strategy, the problem of insufficient endurance for long-term tasks is solved.

[0069] In one embodiment of the present invention, see Figure 2 , based on real-time sensor data in complex environments, the specific steps for dynamically adjusting the path planning of multiple robots to achieve efficient obstacle avoidance include:

[0070] The process of monitoring the location of dynamic and static obstacles in the environment relies on a variety of sensors installed on each robot, such as lidar, ultrasonic sensors, and cameras. These devices can identify fixed and moving obstacles in the current working area. After that, all the detection information is integrated to create a detailed map model for further analysis and processing.

[0071] The operation of predicting potential collision points and generating alternative paths using artificial intelligence algorithms is accomplished through a deep learning framework in one embodiment. The network accepts the map obtained from sensor fusion as an input parameter, and after internal multi-layer perceptron operations, it outputs the results: possible dangerous situations and corresponding solution path sets. Specifically, the AI ​​system will make a detailed assessment of each pair of units that are about to intersect or travel in opposite directions, thereby ensuring a good balance between safety and efficiency.

[0072] Adjusting the robot trajectory based on the error value fed back by the sensor involves feeding back the results of the planned execution to the decision-making process. For example, if the actual driving route of a robot deviates from the planned route, the Kalman filter method will be used to calculate the difference between the two, and the direction and speed parameters to be followed will be modified based on this difference, so that the operation is more in line with the expected settings.

[0073] If the distance d is greater than the threshold T, the shortest path for all affected robots is replanned. In this step, d represents the closest distance measured between the robot and the obstacle, and T is a pre-specified standard value to ensure safety. In general, the ideal selection range is 0.5 to 2 meters (depending on the specific application scenario). Once the comparison between the two exceeds the limit, the optimization program is activated immediately to avoid damage caused by direct impact. This mechanism can effectively solve the problem of new threats brought about by dynamic scene changes. For example, when an unmanned vehicle suddenly encounters a change in the fence of the construction area ahead during the automatic delivery of goods, it will respond reasonably immediately and select the optimal detour strategy.

[0074] In summary, these steps are interconnected to form a closed-loop intelligent control system, which achieves the goal of coordinating and commanding multiple robots in complex environments.

[0075] In one embodiment of the present invention, see Figure 3 ,dynamically adjusting multi-robot path planning based on real-time sensor data in complex environments further includes the following steps:

[0076] Collecting raw data from multiple sensor types. This process involves collecting data output from multiple sensor devices, including visual sensors, distance sensors, and inertial measurement units. The goal is to build a comprehensive, accurate, and real-time map of the environment to assist in subsequent processing. For example, in a warehouse automation environment, this information is used to depict the location of goods and surrounding obstacles.

[0077] Perform data denoising and preliminary filtering. This step mainly reduces errors caused by environmental factors, removes unnecessary interference components and retains useful information. Specifically, high-frequency noise signals are removed by low-pass filtering and other technical means. In one embodiment, for the distance data obtained by the laser rangefinder used in the warehouse handling scenario, the data after this step shows a more accurate target boundary.

[0078] The safety level S of each sensing area is evaluated using a multi-layer neural network model. This stage relies on deep learning algorithms to analyze the previously preprocessed data and score each monitoring block according to the set standards. The safety index S mentioned here quantifies the danger level of the location, and its value is usually between 0 and 1 (0 represents the most dangerous state and 1 represents the safest), and the lowest tolerable level is the threshold M; in actual application, the optimal value of M is defined according to demand. If S ≥ M, it is considered to meet the standard and continue to be considered as a candidate route; if it does not meet the standard, it is necessary to detour to avoid obstacles or find another way.

[0079] After determining the safety index, it is the final step. When it is detected that a certain distance meets the condition that S is not less than the predetermined lower limit, it can be normally included in the feasible trajectory set. This means that if there is no potential threat in a certain direction (or the risk level is within an acceptable range), it will be retained as one of the references for the next move. For example, in the operation process of the delivery robot, assuming that the calculated hazard factor of the current corridor in front is 0.95 greater than 0.6 (here it is assumed that this number is the safety critical point recognized by the system), then this channel is still available and the robot can choose to go straight instead of changing the established route.

[0080] In the design of the entire solution framework, all parameter settings are based on the principle of ensuring smooth, stable and efficient overall operation while ensuring the best quality of task completion. Therefore, each of the above steps is closely linked, and together they achieve the ability to respond and adapt in a timely manner to uncertain and changing external conditions.

[0081] In one embodiment of the present invention, see Figure 4 The present invention is based on real-time sensor data analysis in complex environments. The multi-robot path planning is as follows:

[0082] Define the boundary of the dangerous area R and record the type of abnormal event A detected most recently in the area. This step is to ensure a clear understanding and positioning of the area where risks may exist. R is the specific physical range determined based on sensor feedback, while A is used to mark special situations or types of emergencies that occur in this area, such as obstacle intrusion, equipment failure, and other information.

[0083] Next, weights W are assigned to different areas and the accumulated scores are used to evaluate the overall risk situation. The importance and potential threat level of each part are reflected by assigning corresponding values ​​to different characteristics, and then a comprehensive risk index is generated by combining these values. The setting of weight W depends on the criticality of each section in the global path planning. Generally, appropriate scores are selected from 1 to 10. The cumulative scoring mechanism can accurately evaluate the safety level of the path and guide the next step.

[0084] Based on the above cumulative score, it is decided whether to bypass a specific high-risk area. When the cumulative score is positive and reaches a certain level, it indicates that the risk of the area is high. If there is a more reliable alternative, the system will rearrange the robot team to avoid the potential risk area within the cost and time range. For example, in one embodiment, assuming that the cumulative score in an area is 8, it means that its relative risk is relatively high and should be avoided; conversely, if the score is less than or equal to 7, continue to move forward along the original path.

[0085] If the total score Z at the current time t is greater than the warning threshold C, the risk avoidance strategy will be activated immediately to dynamically update and adjust the entire formation route, where Z reflects the impact factor of all monitoring indicators, and C refers to the alarm standard pre-set to define the danger threshold. The ideal maximum limit of Z is set to 20 (usually not exceeding this value); if it exceeds this limit, it means that the existing route is extremely undesirable and the action plan must be changed to ensure the successful completion of the mission and the non-damage of assets.

[0086] Finally, an example is given to more clearly illustrate how the above method can be applied in actual scenarios. Suppose there is a formation of three logistics robots that work together to carry items to a designated location. Each individual carries multiple detectors to obtain intelligence about the surrounding situation and feed it back to the control center. When a sensor finds a pile of unknown objects near the target location, it quickly transmits the relevant information back to the central processor. According to the comparison of the pre-stored data, it is known that such objects have a high probability of hindering or even damaging the robot's movement. Therefore, after a series of calculations, it is found that the score of the nearby small area has exceeded the warning line C, so the obstacle avoidance plan is immediately initiated-that is, the entire team is ordered to drive along another predetermined path to bypass suspicious obstacles until the forward track in the normal state is restored to continue the task of transporting materials. This cycle is repeated until all steps are completed to ensure the safe and effective implementation of the established operations.

[0087] In one embodiment of the present invention, see Figure 5 , a more sophisticated way to optimize a multi-robot navigation system based on real-time sensor input is:

[0088] First, during the robot collaboration process, the latest information returned by the laser scanner is collected to construct a 3D point cloud map F. In this way, the latest environmental information can be obtained, including the location and size of walls and obstacles, so as to accurately reflect the surrounding space conditions. This process ensures that subsequent navigation instructions rely on the latest and detailed scene description.

[0089] Next, the previously stored map information H and the newly created map G are compared to calculate the change ratio P. Specifically, this change ratio is used to measure the degree of change in the environment between two modelings. In practice, assuming that the shelf position is suddenly moved in a warehouse, a higher P value will be obtained. This indicates that the environment has changed significantly and must be responded to accordingly.

[0090] After that, the action intensity I that needs to be taken is determined based on the above-mentioned degree of change. In other words, the task strategy or movement mode is adjusted according to the actual extent of the environmental change. If it is determined that the environment has only changed slightly, the original route should be maintained; if large-scale obstacles or other abnormal situations occur, the risk should be avoided immediately or a new route should be planned.

[0091] Furthermore, when the change ΔH (the difference between the current and previous maps) is divided by H (the size of the original map data), that is, the value of ρ exceeds the given threshold L, the system will increase the detection frequency or request external additional resource support. The parameter ρ defines the metric for the percentage of map changes: ρ = ΔH / H; here ΔH means the size of the change in the map structure between the current and the last record; the range is [0, +∞]; and the optimal value is usually determined by the reliability requirements and processing capabilities in the specific application scenario. For example, in an emergency (such as the detection of an unknown dangerous object), L can be set to a lower value (such as 5% to 10%) in advance to facilitate rapid decision-making. This is because even slight changes in geographical morphology can have a significant impact on high-precision positioning tasks. When encountering extreme events (such as huge ruins after the collapse of a building), ρ may quickly soar to a level far beyond the specified limit L, and all available resources need to be mobilized immediately to cope with sudden changes.

[0092] In one embodiment, imagine a team of three handling robots working in a complex factory workshop. These devices are equipped with multiple lidar sensors to monitor the physical environment around them, and they continuously use this data to maintain a real-time 3D map of their area. In the early hours of the morning one day, without prior notice, the staff began to adjust the large machinery on the production line, causing some key aisles to become narrower and many temporary obstacles to appear. At this time, the previous periodic refresh rate may not be able to meet the needs of instant perception. Therefore, by introducing the above mechanism: as the difference ΔH obtained from each refresh continues to increase and triggers the condition of ρ>L, not only the communication frequency between members is strengthened, but also a group of manpower is promptly obtained from the monitoring command room to assist in clearing the accumulation of items in the blocked area, thereby ensuring the continuous operation efficiency of the entire system without too much interference.

[0093] In one embodiment of the present invention, see Figure 6 , based on the continuously updated real-time sensing results, the refined adjustment plan includes but is not limited to the following operations:

[0094] The first operation to be performed is to accurately locate the current position of each member and mark Q in the electronic grid. This step is intended to ensure that the exact position of each robot participating in the collaborative task is identified in the digital grid map. This requires obtaining the specific coordinates of the robot through sensors and then mapping them to the grid model to complete the setting of Q. For example, in one embodiment, if a group of small drones used for search and rescue operations are conducting a joint survey of a complex terrain, the precise positioning and calibration of these individuals can assist the command and control system to grasp the area where each drone is located, so as to better deploy each equipment for subsequent actions.

[0095] Another important process is to analyze the past trajectory data mining pattern characteristics O in order to identify behavioral patterns and potential danger signals from the past mobility history. In this process, data mining technology is used to parse the stored path records and learn the patterns, trends and potential hidden safety hazards. For example, in practical applications, assuming that the location, time period and surrounding conditions where failures or accidents occur frequently in certain specific environments will be paid special attention. Once the relevant risk factors are found, they will be immediately transferred to the next stage for processing. In addition, the understanding of parameter O refers to the summary of the inherent correlation shown by a series of time-space distribution points, which helps to make further decisions based on the ability to predict future conditions.

[0096] In order to effectively respond to the various uncertain factors that come one after another, targeted prevention suggestions J will be put forward to reduce the potential threat level D. This link focuses on planning response strategies in advance to prevent the occurrence of known and foreseeable problems. For example, in an implementation scenario, if monitoring finds that obstacles frequently appear in a certain direction, the suggestion can be to guide the robot to avoid this area or slow down and drive carefully. At the same time, for the D value, it measures the severity of a certain degree of crisis, and the range is from high risk 0.9 to low safety 0.1. The appropriate level is selected according to the specific task requirements and environmental conditions. For example, when it is set at 0.3, a more balanced match between safety and response speed can be achieved.

[0097] Finally, a time window w will be set up during the detection process to observe whether there are abnormal activities beyond the normal limit N. If the frequency of deviation q exceeds the limit of the allowable change level N (i.e. q>N), the system determines that there is a potential risk and needs to immediately trigger an alarm to prepare to initiate risk avoidance measures to resolve the current crisis situation. The three letter variables in the formula are explained as follows: w represents the length of the continuous uninterrupted observation period, and the default recommended unit period is 2-5 minutes; q reflects the actual cumulative number of deviation events, and the counting method calculates the cumulative value based on the fixed interval refresh mechanism; finally, N refers to the amplitude threshold of the fluctuation range within the normal range. Considering the differences between different types of tasks, N is worth adjusting based on specific analysis of the actual situation. However, maintaining a small error ratio is conducive to maintaining the stability of the system. Therefore, in most cases, it is recommended that the initial setting is less than or equal to 0.05.

[0098] In one embodiment of the present invention, see Figure 7 , implement a more intelligent path adjustment mechanism based on real-time sensor data in complex environments:

[0099] The first step is to integrate multiple information sources to output comprehensive environmental cognition. This mechanism builds a comprehensive environmental model by integrating data from multiple sensors such as cameras, lidar, and infrared sensors to enhance the robot's perception accuracy and response speed in non-deterministic scenarios. Specifically, in one embodiment, different types of sensor data are transmitted to the central processing unit for analysis and fusion, thereby providing a detailed and reliable input basis for decision-making. This information covers details such as spatial structure, object location and its attributes.

[0100] Then, the adaptive filtering technology is applied to improve the positioning accuracy. In this step, Kalman filtering or other similar adaptive algorithms are adopted to eliminate or correct the noise and outliers existing in the sensor readings. Formula E represents minimizing the estimation error by dynamically updating the trade-off coefficient k(n): [E = lim_{n→∞}{E{x(k)} - \hat{x}(k)}], where k represents the time iteration step, [x(k)] represents the true value, and [\hat{x}(k)] is the estimated output. The parameter values here depend on the specific noise level and system dynamic characteristics; usually, the best trade-off strategy that can minimize the prediction variance is selected to make the position calculation more stable and reliable, ensuring that the robot always maintains efficient travel trajectory control.

[0101] Next is to improve the future hazard avoidance plan based on historical experience. By analyzing the memory bank of previous execution results and applying machine learning models, the obstacle avoidance strategy is automatically optimized. For example, after successfully taking certain turning or deceleration measures to avoid threats in a similar environmental layout in the past, similar countermeasures will be preferentially recommended in the future, thereby enhancing safety and efficiency. This requires storing a large amount of case data as a reference and using classification and clustering algorithms to extract patterns, and making more intelligent and reasonable pre-judgments and adjustment action designs based on this.

[0102] The last item is to switch to the standby low-light mode or sonar-assisted detection system when an unknown obstacle is encountered and the remaining visibility r is less than the set value V. In this scenario, r can refer to the actual visual range length index (in meters), and V sets the safety threshold standard for activating the auxiliary sensing device. That is, if [r < V] (such as sudden attenuation of on-site lighting or smoke obstructing the line of sight and other unexpected factors), the system should immediately switch to using more robust sensing tools to ensure the ability to continue operating and ensure the integrity of personnel and property. At this time, sonar or night vision equipment can play a key role in escorting the smooth completion of the task.

[0103] In an embodiment of the present invention, the considerations of path replanning in the robot's autonomous behavior norms in complex environments are further deepened: First, a special obstacle avoidance plan X is set for the task background. This step aims to analyze and identify potential dangerous obstacles in the operating environment, such as sudden collapse areas, temporary accumulations, or moving objects. For this factor, this means pre-evaluating the working场所 of the robot and possible changes on the target path, and formulating emergency measures or alternative path strategies according to the possible risks. For example, in one embodiment, when deploying a group of automatic guided vehicles (AGVs) in a logistics warehouse to perform material handling tasks, once signs of a front shelf collapse are detected, the plan X is immediately called to trigger an emergency stop, and a warning message is sent to other vehicles through the background dispatching center to prompt a change in the walking channel.

[0104] Then, a differentiated speed adjustment policy Y is formulated in combination with the terrain undulation characteristics. Different geographical conditions will cause changes in driving safety and stability. Higher moving speeds are allowed on flat sections to improve operation efficiency; while on non-uniform ground surfaces such as slopes and stepped surfaces, the traveling speed needs to be reduced to avoid accidental rollovers or skidding. For example, when operating a multi-rotor UAV in a mountain patrol scenario, its control system judges the height difference and tilt angle at the current location based on high-precision mapping data. When it is confirmed that it is passing through a narrow aerial corridor between steep cliffs in the valley, it automatically slows down the flight speed according to the principle to ensure a smooth crossing.

[0105] Based on the above two analysis results, a capability framework for rapidly iterating and locally fine-tuning the scheme V based on real-time road conditions is constructed. This means continuously collecting the surrounding information provided by the latest perception layer, including sources such as visual sensing images and radar echo point cloud data, in order to timely correct the position error problems caused by new variables not considered before. Specifically, when a city cleaning robot encounters an unforeseeable pothole area with accumulated water during street cleaning, relying on the actual contour feature parameters returned by a small laser sensor installed under the front end, after the algorithm optimization engine re-evaluates whether the current position coordinates deviate from the established track, a millisecond-level precise correction action is taken to prevent collisions with fixed structures or pedestrians.

[0106] If the path curvature radius Rm is less than the maximum steering ability Rs of the vehicle (expression: Rm < Rs), then the forward speed is slowed down to prevent overturning or out-of-control phenomena. Among them, Rm represents the parameter of the curved road surface radius, which depends on the shape and scale characteristics of the turning section. There is no absolute limit range, but in the ideal state, it is as close as possible to the target curve arc length value; Rs explicitly represents the limit parameter of the single locomotive steering angle, which is generally set by the manufacturer at the factory to give a definite value. For example, an omnidirectional mobile platform equipped with an off-road wheeled service manipulator has the characteristic of a small inner wheel spacing, thus obtaining a larger minimum turning diameter performance. That is, its Rs is usually in the range of 1 meter to 2.5 meters. Therefore, during the process of delivering packages through a narrow aisle, whenever it is predicted that it is about to enter a relatively sharp corner area, the speed will be actively reduced according to this relationship logic to avoid potential risks, and at the same time, the actual trajectory deviation degree will be dynamically monitored to prepare for timely taking supplementary intervention measures to ensure the successful completion of the mission journey.

[0107] In one embodiment of the present invention, in a multi-robot formation motion scenario, the collaboration efficiency is improved by enhancing perception capabilities and fine-tuning regulation, specifically: introducing edge computing nodes to speed up the local computing speed Zc. This means that a small computing device with efficient processing capabilities is installed close to the robot. These edge nodes can quickly process local information without relying on a distant central processor, thereby reducing the time cost of data round trips and also reducing the communication load. In one embodiment, multiple cleaning robots need to coordinate cleaning paths that cover a large office area. If each sensor and actuator is equipped with a specific algorithm processing chip as a local decision-making unit, the route can be changed in real time when a ground obstacle is detected, thereby improving efficiency without the need to frequently contact a remote data center for instructions.

[0108] To ensure the stability and accuracy of communication, a dual redundant communication channel redundancy design is implemented to improve the anti-interference ability Kt. Here, Kt reflects the probability that the system can still transmit effectively when facing different environmental noise or signal attenuation. The value range is usually from 0 to 1, and the expected optimal value of high reliability is close to 1. If a single channel is interfered with or fails, another backup channel can seamlessly take over to ensure continuous data transmission quality. For example, consider a search and rescue robot team shuttling through a complex disaster area for search activities; sending critical information such as distress signs on two frequency bands at the same time can significantly enhance the probability of successful positioning of survivors and the consistency of instructions issued by the command center.

[0109] Dynamically adjust the resource allocation level Lb according to the complexity of the task, which involves the scheduling of resources such as power and CPU time slices. Before the task starts, the planning algorithm will evaluate the workload of each stage and then allocate reasonable initial resources. As the environment changes or emergencies occur, the progress and status will be continuously monitored to increase or decrease the energy and other auxiliary tools allocated to each unit robot in a timely manner. Assume that a swarm of drones completes the aerial material delivery operation. Specifically, the pilot aircraft estimates the flight energy consumption based on the current weather conditions and cargo weight, and flexibly replenishes the power battery of each team member according to the actual situation to prevent a member from being exhausted midway and causing the overall transportation plan to be interrupted.

[0110] When the ratio of computing resource utilization u to total available memory blocks c is greater than or equal to 1, that is, (u / c geq1), it means that all storage space allocated to the subsystem has been exhausted and there are additional applications that have not been met. If the load pressure g is close to the maximum carrying capacity s (g is approximately s), it indicates that the support limit is about to be exceeded and optimization measures must be taken immediately. The compression algorithm is started to release some idle or unimportant memory fragments so that more important processes can get priority to obtain running resources to ensure that important workflows are not negatively affected. For example, during the deployment of a large-scale warehouse, a robot army of many types of mobile sorting goods between shelves. Due to a series of high-frequency requirements such as video stream analysis, positioning and navigation, the service site servers near the main roads may be in an oversaturated operation state. Therefore, whenever it approaches the critical point, it triggers automatic image size reduction or simplification of non-mainline code logic fragments to reduce the burden on the server, ensure that the picking speed is not affected, and keep the entire automated logistics operation system running smoothly.

[0111] In one embodiment of the present invention, group work performance is optimized for highly uncertain workplace characteristics:

[0112] First, a virtual fence is built to limit the activity radius. By defining and deploying digital fences, the boundaries within which the robot group can move freely in a specific task are determined to prevent it from exceeding the preset working range. For example, in a logistics warehouse scenario, the fence can ensure that the robot transporting goods can only move within the specified area without mistakenly entering the prohibited area or affecting other operations.

[0113] Secondly, achieve all-weather coverage of multi-dimensional stereoscopic visual monitoring. Use high-precision cameras or other sensing devices to form a multi-level, all-round monitoring system to obtain real-time three-dimensional spatial information and perform environmental perception. In one embodiment, multiple cameras in a factory workshop can continuously collect image data from different angles, and feed it back to the main control unit after algorithm processing, thereby improving the ability to identify objects and obstacles.

[0114] Furthermore, a flexible response process is established based on the theory of environmental adaptation. This process involves analyzing and learning environmental characteristics to flexibly respond to unpredictable changes, and designing a mechanism to ensure that the system can adjust its strategy in time to continue to operate normally when conditions change. Specifically, if a construction site is delayed due to a sudden change in weather, the control system will quickly adjust the plan based on existing experience, such as reallocating human and material resources or delaying the start time of non-critical processes.

[0115] Then, when an emergency occurs and the overall system performance degradation index p falls into the warning interval Pd (p represents the percentage of performance reduction in the process of completing a specified task; Pd means a pre-set risk level critical point). Once it is found that p is not greater than or equal to Pd, protective measures will be immediately initiated and the surrounding area will be re-surveyed to determine the latest status. For this formula, the value of p is usually between 0% and 100%, and the optimal choice of Pd depends on the fault tolerance rate and task requirements in the actual application scenario. If an abnormal situation suddenly occurs at a construction site (such as interference from an external UFO), and the detection instrument records that the current operating quality is significantly reduced and is lower than the pre-determined standard limit, corresponding measures should be taken immediately to ensure the safety of personnel and property while also avoiding losses in a larger range.

[0116] These steps work together to enhance group work efficiency and reliability, allowing them to operate efficiently even in complex and changing situations.

[0117] In one embodiment of the present invention, the key links of the present invention for further improving the path adjustment in the construction of task-driven robot cluster autonomous system for complex environments are: using incremental reinforcement learning algorithm to train the ability to cope with sudden changes. This method improves the ability of task-driven robots to quickly adapt to emergencies in complex environments by dynamically adjusting parameter weights and updating state-action value functions. For example, in a disaster response scenario, after receiving new obstacle information, the robot can instantly re-plan the route without re-initializing the entire learning process. This approach can not only improve response efficiency, but also reduce energy consumption.

[0118] Use deep convolutional networks to gain a deeper understanding of the physical space components. This step involves extracting environmental features from sensor data through deep learning techniques to build a more accurate three-dimensional model, help identify object shapes, textures and other characteristics, and optimize obstacle avoidance and path selection accordingly. In one embodiment, when one or more robots in the cluster encounter unexpected surface changes (such as puddles or building ruins), they can use the extracted spatial structure knowledge to make safe detour decisions.

[0119] Weigh the actual value Mv of various alternative options based on the cost-benefit ratio. This process refers to calculating the cost-benefit ratio of the current solution relative to other available options before taking each step, and selecting the most economical and feasible solution based on this. Specifically, in the task of delivering goods, faced with different transportation times and potential loss risks on multiple possible delivery routes, the result that minimizes the time while maintaining the best integrity rate of the goods will be evaluated and recommended to the execution unit as a high-priority path.

[0120] If the growth rate of the number of new potential risk sources m reaches r (r = (m-m_prev) / t_interval), exceeding the established tolerance level e (such as r>e), the approach will be stopped immediately and reported to the remote control center for confirmation of the follow-up action plan. Here, m, m_prev and t_interval represent the number of the latest statistical risk results, the number of results of the last inspection and the detection interval, respectively, and the range is usually 0 to positive infinity; among them, e generally depends on the risk tolerance limit of the actual application scenario, and a value of about 0.1 is usually considered to be a reasonable threshold setting to ensure timeliness but not over-sensitivity. This setting is intended to prevent the dangerous situation from deteriorating while buying the necessary time for human intervention, that is, to ensure that the robot can automatically take evasive measures and send warning notifications to the operator when it detects abnormally increased uncertainties.

[0121] Also, see Figure 8 This embodiment provides a robot collaborative control system, which is used to implement the method described in any of the above embodiments, including:

[0122] A path adjustment unit 100, which is used to dynamically adjust the path planning of multiple robots based on real-time sensor data in a complex environment to achieve efficient obstacle avoidance;

[0123] A link management unit 200, which intelligently manages the communication links between different robots to reduce communication delays and improve the accuracy of synchronous task execution;

[0124] An algorithm optimization unit 300, which dynamically adjusts the algorithm according to the task priority to optimize the emergency response efficiency;

[0125] The data processing unit 400 adjusts the fusion processing parameters of the sensor data to reduce the environmental perception error and improve the perception accuracy.

[0126] See also Fig. 9 , Fig. 9 A schematic diagram of an electronic device 20 that can implement an embodiment of the present invention is shown, and the electronic device is intended to represent various forms of control devices. The components shown herein, their connections and relationships, and their functions are only examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0127] The electronic device 20 includes at least one processor 21, and a memory connected to the at least one processor 21, such as a read-only memory (ROM) 22, a random access memory (RAM) 23, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 22 or the computer program loaded from the storage unit 28 to the random access memory (RAM) 13. In the RAM 23, various programs and data required for the operation of the electronic device 20 can also be stored. The processor 21, ROM 22 and RAM 23 are connected to each other through a bus 24. An input / output (I / O) interface 25 is also connected to the bus 24.

[0128] A number of components in the electronic device 20 are connected to the I / O interface 25, including: an input unit 26, such as a keyboard, a mouse, etc.; an output unit 27, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 29, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 29 allows the electronic device 20 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0129] The processor 21 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 21 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 21 performs the various methods and processes described above.

[0130] In some embodiments, the method of the above embodiment may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 28. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 20 via the ROM 22 and / or the communication unit 29. When the computer program is loaded into the RAM 23 and executed by the processor 21, one or more steps in the method described above may be performed. Alternatively, in other embodiments, the processor 21 may be configured to perform the method of the above embodiment in any other appropriate manner (e.g., by means of firmware).

[0131] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0132] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0133] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0134] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and

[0135] Input from the user may be received in any form, including acoustic input, voice input, or tactile input.

[0136] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0137] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0138] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A robot collaborative control method, characterized in that: include: Based on real-time sensor data in complex environments, dynamic adjustments are made to the path planning of multiple robots to achieve efficient obstacle avoidance; Intelligent management of communication links between different robots to reduce communication delays and improve the accuracy of synchronous task execution; Dynamically adjust the algorithm based on task priority to optimize the efficiency of emergency response; Adjust the fusion processing parameters of sensor data to reduce environmental perception errors and improve perception accuracy.

2. A robot collaborative control method according to claim 1, characterized in that: The specific steps of dynamically adjusting the path planning of multiple robots based on real-time sensor data in a complex environment to achieve efficient obstacle avoidance include: Monitor the location of dynamic and static obstacles in the environment; Use artificial intelligence algorithms to predict potential collision points and generate alternative routes; Adjust the robot trajectory based on the error value fed back by the sensor; If the distance d is greater than the threshold T, the shortest paths of all affected robots are replanned, where d represents the minimum distance between the robot and the obstacle, and T represents the preset safety distance threshold.

3. A robot collaborative control method according to claim 2, characterized in that: The dynamic adjustment of multi-robot path planning based on real-time sensor data in a complex environment further comprises the following steps: Collect raw data from multiple sensor types; Perform data denoising and preliminary filtering; Use a multi-layer neural network model to evaluate the safety level S of each perception area; If the safety level S is not less than the minimum acceptance standard M, the corresponding area is allowed to continue to be the candidate path selection range, where S represents the calculated safety index and M represents the lower limit of the tolerable risk level.

4. A robot collaborative control method according to claim 3, characterized in that: The multi-robot path planning based on real-time sensor data analysis in a complex environment is specifically as follows: Define the boundary R of the dangerous area and record the most recently detected abnormal event type A in the area; Assign weights W to different areas and accumulate scores to evaluate the overall risk situation; Decide whether to bypass a specific high-risk area based on the accumulated score; If the total score Z at the current time t is greater than the warning threshold C, the comprehensive avoidance strategy is triggered to update the entire formation route, where t represents the current moment, Z represents the accumulated dangerous area impact factor, and C represents the triggering alarm threshold.

5. A robot collaborative control method according to claim 4, characterized in that: A more sophisticated approach to optimizing a multi-robot navigation system based on real-time sensor input is: Collect the information returned by the latest laser scanner to build a 3D point cloud map F; Compare the previous map information H and the newly created map G to calculate the change ratio P; Determine the intensity of action to be taken based on the degree of change I; If the ratio ρ of the change ΔH to H exceeds the specified limit L, the detection frequency is increased or additional resources are requested, where ρ = ΔH / H, ΔH is the difference between the current and previous maps, and L is the preset maximum change limit.

6. A robot collaborative control method according to claim 5, characterized in that: The refinement and adjustment scheme based on the continuously updated real-time sensing results includes but is not limited to the following operations: Accurately locate the current position of each member and mark Q in the electronic grid; Analyze past trajectory data to mine pattern characteristics and learn potential risk signals; Propose targeted preventive suggestionsJ to reduce the potential threat levelD; If the behavior observed within the specified time window w deviates from the expected probability q by more than the normal fluctuation range N (i.e., q>N), a warning is issued and corresponding avoidance measures are prepared to be implemented. w, q, and N represent the length of the inspection cycle, the frequency of deviations, and the allowable range of variation, respectively.

7. A robot collaborative control method according to claim 1, characterized in that: The implementation of a more intelligent path adjustment mechanism based on real-time sensor data in complex environments: Integrate multiple information sources to output comprehensive environmental cognition B; Apply adaptive filtering technology to improve positioning accuracy E; Improve future hedging plans based on historical experienceC; If the remaining visibility r is less than the minimum visibility V when encountering an unknown obstacle, switch to the backup low-light mode or sonar-assisted detection system, where r and V represent the actual visual conditions on site and the minimum requirements to ensure safe navigation, respectively.

8. A robot collaborative control system, used to implement the method according to any one of claims 1 to 7, characterized in that: include: A path adjustment unit, the adjustment unit is used to dynamically adjust the path planning of multiple robots based on real-time sensor data in a complex environment to achieve efficient obstacle avoidance; A link management unit, which intelligently manages the communication links between different robots to reduce communication delays and improve the accuracy of synchronous task execution; An algorithm optimization unit, wherein the algorithm optimization unit dynamically adjusts the algorithm according to the task priority to optimize the emergency response efficiency; A data processing unit adjusts the fusion processing parameters of the sensor data to reduce environmental perception errors and improve perception accuracy.

9. An electronic device, characterized in that: The method comprises at least one processor, wherein the processor is communicatively connected to at least one memory, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the method according to any one of claims 1 to 7 when the processor executes the computer instructions.

Images