Space-time grid multi-robot dynamic priority game cooperative control method and system
By employing a decentralized distributed architecture and a dynamic priority game-theoretic collaborative control method, the stability and efficiency issues of multi-robot systems are resolved, enabling efficient obstacle avoidance and autonomous collaborative operation, thereby improving the system's fault tolerance and adaptability.
Patent Information
- Application Number
- CN202610715104.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-05-21
- Publication Date
- 2026-07-10
AI Technical Summary
Existing multi-robot systems suffer from problems such as high risk of single point of failure, high communication load, large latency, statically fixed priorities, and weak deadlock resolution capabilities, resulting in unstable system operation and low efficiency.
A decentralized distributed architecture is adopted, and the robot’s autonomous and collaborative control is achieved through a local rolling spatiotemporal grid map, dynamic priority non-cooperative game, and deadlock automatic resolution mechanism. The robot uses onboard sensors to perceive the environment, constructs a local spatiotemporal grid map, combines Kalman filtering to predict dynamic obstacles, uses multi-factor weighted calculation to calculate dynamic priorities, and allocates path rights through a Nash equilibrium strategy.
It improves the environmental adaptability of multi-robot systems, reduces collision risks, enhances the stability and efficiency of collaborative control, strengthens system fault tolerance and response speed, and reduces deployment and maintenance costs.
Smart Images
Figure CN122363346A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of mobile robot control and artificial intelligence technology, specifically relating to a spatiotemporal grid-based multi-robot dynamic priority game collaborative control method and system. Background Technology
[0002] With the continuous improvement of industrial automation, multi-robot systems have been widely used in inspection and handling operations in complex scenarios such as substations, smart factories, and logistics warehouses, becoming the core carrier of intelligent industrial operation and maintenance. Currently, mainstream multi-robot collaborative control technologies generally adopt a centralized scheduling architecture, where all robot status information needs to be uploaded to a central server in real time. The server then uniformly completes path planning and command issuance, which is difficult to meet the reliable operation requirements of dynamic and complex industrial environments.
[0003] However, existing technologies have the following shortcomings: First, the risk of single point of failure is prominent, and failure of the central server or core communication node will directly lead to system paralysis; Second, the communication load is high and the latency is large. After the scale of robot expansion increases, the bandwidth consumption increases dramatically, and instruction lag is prone to occur in strong electromagnetic interference environments, causing safety hazards; Third, the priority and scheduling rules are static and fixed, and the right-of-way cannot be dynamically adjusted according to the urgency of the task, the robot's battery status, etc., resulting in low overall system efficiency; Fourth, the deadlock resolution capability is weak. Multiple robots are prone to looping and waiting in narrow passages and intersections, the centralized global replanning response is slow, and the on-site adaptability is poor.
[0004] Therefore, a new method is urgently needed. Summary of the Invention
[0005] The purpose of this invention is to provide a method and system for dynamic priority game-theoretic collaborative control of multi-robots in a spatiotemporal grid. This method achieves autonomous collaboration, efficient obstacle avoidance and stable operation of multi-robot clusters through a decentralized distributed architecture, spatiotemporal grid conflict prediction, dynamic priority non-cooperative game-theoretic approach and deadlock automatic resolution mechanism.
[0006] To achieve the above objectives, this invention provides a spatiotemporal grid-based multi-robot dynamic priority game cooperative control method and system, comprising the following steps: S1. Each inspection robot uses onboard sensors to perceive the surrounding environment in real time, constructs a local rolling spatiotemporal grid map, and simultaneously extracts task urgency, remaining battery power, remaining task time, and kinematic constraint factor state parameters; the local rolling spatiotemporal grid map is transmitted to the local unit S2 and S3, and the state parameters are transmitted to the local unit S4. S2. Based on the local rolling spatiotemporal grid map of S1, generate the desired trajectory with timestamps, expand the trajectory into a spatiotemporal pipeline representing uncertainty, package the robot identifier, desired trajectory and state parameters into a broadcast packet, and send it to the neighboring machine at a preset frequency; the broadcast packet is received by the neighboring machine S3. S3 receives the neighbor broadcast packet from S2, maps the neighbor's expected trajectory to the local rolling spatiotemporal grid, detects the overlap of the spatiotemporal pipeline space of the two sides on a time-by-time slice, and determines the potential conflict event if there is an overlap; transmits the conflict determination result and the neighbor's state parameters to the local S4 to trigger the game. S4. Calculate the dynamic priority coefficient based on the local state parameters of S1 and the neighboring state parameters of S3, and allocate road rights using a non-cooperative game Nash equilibrium strategy; when the two machines meet, compare the priority coefficients to determine the winner and the conceding party. The winner maintains the original trajectory, and the conceding party performs the least costly avoidance action; when the difference in priority coefficients is within a preset threshold, introduce a random disturbance correction coefficient and re-determine to resolve the deadlock; the road rights allocation result and avoidance command are transmitted to the local machine S5. S5. Based on the right-of-way result and avoidance command in S4, the winning party maintains its original trajectory and motion state, while the yielding party performs an avoidance action; after the conflict is resolved, the original mission trajectory is restored, and motion control commands are output to drive the chassis actuators.
[0007] Preferably, in S1, the local scrolling spatiotemporal grid map is a three-dimensional matrix with a time slice step of 0.5 seconds, covering a future time window of 0-5 seconds, and is updated at a frequency of 10Hz. Local scrolling spatiotemporal raster map satisfy: When spatial coordinates At any moment When occupied by obstacles When spatial coordinates At any moment When it is a passable area ; in, For the three-dimensional raster state values of the locally scrolling spatiotemporal raster map; These are the X and Y coordinates of the spatial plane. For each moment in a time slice.
[0008] Preferably, in S2, the desired trajectory is a set of pose points with timestamps. The spatiotemporal pipeline is a cylindrical trajectory envelope; the broadcast packet length does not exceed 256 bytes, and the transmission frequency is 10Hz.
[0009] Preferably, in S3, the condition for determining spatial overlap is: ; In the formula, For the first In a time slice, the robot With robots The center distance of the spatiotemporal pipeline; For the first Each discrete-time slice moment; The robot's safety radius; When the spacetime pipelines of the two robots are at any time When the conditions for spatial overlap are met, it is determined that spatial overlap exists; if only the trajectory uncertainty envelope contacts without forming spatial overlap, a low-speed warning is triggered, and the robot actively reduces its movement speed by 30% to shrink the trajectory uncertainty envelope.
[0010] Preferably, in S4, the dynamic priority coefficient is calculated using a multi-factor weighted average, which takes into account the urgency of the task, the remaining battery power, the remaining time of the task, and the kinematic constraint factor. The formula for calculating the dynamic priority coefficient is: ; In the formula, For robots At any moment Dynamic priority coefficient; For robots The urgency of the tasks includes fire alarm, fault review, routine inspection and return to base for charging. Among them, fire alarm is assigned a value of 10, fault review is assigned a value of 8, routine inspection is assigned a value of 5, and return to base for charging is assigned a value of 1. For robots At any moment The remaining battery power; For robots The maximum battery capacity, which is used to characterize the low-charge protection weight; For robots At any moment The remaining time, which is used to characterize the urgency of the time; For robots The kinematic constraint factor is used to characterize the difficulty of a robot to stop suddenly or change lanes. , , , These are preset weighting coefficients for task urgency, low battery protection, time urgency, and kinematic constraints, respectively.
[0011] Preferably, in S4, an anti-jitter threshold is set. The winning side locks the original trajectory, and the conceding side chooses the action with the minimum cost function from deceleration, waiting, and rerouting. The game process is as follows: Set anti-shake threshold The robot calculates dynamic priority coefficients. The game theory approach is used to determine the right-of-way between the conflicting parties. If the dynamic priority coefficient of robot A and the dynamic priority coefficient of robot B satisfy... If so, then robot A is determined to be the winner of the game, i.e., the one with the right of way, and robot B is determined to be the one who concedes, i.e., the one who avoids the path. If the dynamic priority coefficient of robot B and the dynamic priority coefficient of robot A satisfy... The system determines that robot B has the right-of-way and robot A has the right-of-way.
[0012] Preferably, in S4, deadlock resolution is achieved by adding a 0-1 random number to adjust the priority coefficient and re-determining the game to break the symmetric state, specifically as follows: If the dynamic priority coefficients of both parties satisfy If a path conflict deadlock is detected, the deadlock resolution mechanism will be initiated: Set random disturbance coefficient The corrected formula is as follows: ; In the formula, For robots The original dynamic priority coefficient; This refers to the corrected dynamic priority coefficient; A uniformly distributed random number in the interval 0 to 1; The revised dynamic priority coefficients are then reintroduced into the game decision process, with no more than three iterations, to complete a stable determination of the right-of-way party and the yielding party.
[0013] This invention also provides a spatiotemporal grid-based multi-robot dynamic priority game cooperative control system, comprising: The local spatiotemporal grid perception construction module is used to control each inspection robot to perceive the surrounding environment in real time using onboard sensors, construct a local rolling spatiotemporal grid map, and simultaneously extract the robot's task urgency, remaining battery power, remaining task duration, and kinematic constraint factor state parameters. The spatiotemporal grid map is then output to the trajectory intent broadcasting module and the collision risk prediction module, and the state parameters are output to the dynamic priority game module. The trajectory intent broadcasting module, connected to the local spatiotemporal grid perception construction module, is used to generate a timestamped desired trajectory based on the local rolling spatiotemporal grid map, expand the trajectory into an uncertain spatiotemporal pipeline, package the robot identifier, desired trajectory and state parameters to form a broadcast packet, and send it to adjacent robot nodes within the communication range at a preset frequency. The collision risk prediction module is connected to the local spatiotemporal grid perception construction module and the trajectory intention broadcasting module. It is used to receive trajectory intention broadcast packets from adjacent robot nodes, map the expected trajectory of the neighboring robot to the local rolling spatiotemporal grid of the local robot, detect the spatial overlap of the spatiotemporal pipelines of both parties on a time-by-time slice and determine potential conflict events, and transmit the conflict determination results and the neighboring robot state parameters to the dynamic priority game module. The dynamic priority game module is connected to the local spatiotemporal grid perception construction module and the collision risk prediction module, respectively. It is used to calculate the dynamic priority coefficient by combining the state parameters of the local machine and the neighboring machine, complete the right-of-way allocation by adopting the Nash equilibrium strategy, compare the priority coefficients to determine the winner and the conceding party, introduce random perturbation correction coefficients to re-determine the deadlock state within the threshold range, and output the right-of-way allocation result and avoidance action instructions to the adaptive cooperative execution module. The adaptive collaborative execution module, connected to the dynamic priority game module, is used to control the winning party to maintain its original trajectory movement and the conceding party to perform the minimum cost avoidance action according to the right-of-way allocation result and avoidance instructions. After the conflict is resolved, the original task trajectory and running state of both parties are restored, and motion control instructions are output to drive the robot chassis execution mechanism.
[0014] Therefore, the present invention employs the above-mentioned spatiotemporal grid multi-robot dynamic priority game cooperative control method and system. Compared with the prior art, the technical solution of the present invention has the following beneficial effects: (1) By using airborne sensors to perceive the environment and construct a local spatiotemporal map containing static and dynamic obstacles, and combining Kalman filtering to predict the trajectory of dynamic obstacles, the problem that existing technologies only focus on static obstacles and lack dynamic prediction when multiple robots cooperate, resulting in high collision risk and poor environmental adaptability is overcome. This achieves the effects of accurately perceiving the inspection environment, predicting dynamic trends, reducing the probability of collision, and improving environmental adaptability. (2) The dynamic priority is calculated by multi-factor weighting, and the right-of-way is allocated by combining the Nash equilibrium strategy. The deadlock is resolved by random perturbation correction. This overcomes the problems of unreasonable right-of-way allocation, game deadlock, and low collaborative efficiency in the existing multi-robot collaboration. It achieves the effects of fair right-of-way allocation, quick resolution of decision-making deadlock, and improved collaborative control stability and efficiency. (3) The distributed modular design is adopted, and each robot can autonomously complete perception, decision-making and execution without the intervention of the central controller. This overcomes the problems of existing collaborative control relying on central control, high response delay, poor fault tolerance and insufficient adaptability, and achieves the effects of improving collaborative response speed, enhancing system fault tolerance, adapting to various distributed inspection scenarios and reducing deployment and maintenance costs.
[0015] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description
[0016] Figure 1 This is a flowchart of an embodiment of the spatiotemporal grid multi-robot dynamic priority game cooperative control method and system of the present invention; Figure 2 This is a schematic diagram of spatiotemporal conflict detection in an embodiment of the spatiotemporal grid multi-robot dynamic priority game cooperative control method and system of the present invention. Figure 3 This is a flowchart of the dynamic priority game in the first embodiment of the spatiotemporal grid multi-robot dynamic priority game cooperative control method and system of the present invention; Figure 4 This is a schematic diagram illustrating a dynamic priority game-based right-of-way allocation example in an intersection scenario, as described in Embodiment 1 of the Spatiotemporal Grid Multi-Robot Dynamic Priority Game Cooperative Control Method and System of the present invention. Detailed Implementation
[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Unless otherwise defined, the technical or scientific terms used in the present invention should have the ordinary meaning understood by those skilled in the art.
[0018] Example 1 like Figures 1-4 As shown, this embodiment provides a spatiotemporal grid multi-robot dynamic priority game cooperative control method and system. It should be understood that the specific parameters, models and protocols mentioned in this embodiment are only examples to help those skilled in the art understand the present invention, and are not intended to limit the present invention.
[0019] The spatiotemporal grid multi-robot dynamic priority game cooperative control method and system of the present invention includes the following steps: S1. The robot collects positioning data from LiDAR, vision sensors, IMU, and SLAM to construct a local rolling spatiotemporal grid map.
[0020] The generated local rolling spatiotemporal grid map is passed to steps S2 and S3; at the same time, the robot's task urgency, remaining battery power and current motion status are collected and passed to step S4.
[0021] In this step, the robot uses the SLAM (Simultaneous Localization and Mapping) algorithm to locate its own position and fuses LiDAR point cloud data to construct a local cost map. Unlike traditional two-dimensional planar maps, this map is a three-dimensional grid map containing two-dimensional spatial coordinates and discrete time slices. The time slice step size is a preset 0.5s, covering a preset 5s time window in the future, and is updated at a preset frequency of 10Hz. It includes static obstacle grids and dynamic obstacle prediction occupancy grids.
[0022] The robot performs SLAM localization at a preset frequency of 50Hz to acquire real-time two-dimensional planar pose information, and fuses LiDAR and visual sensor data at a preset frequency of 10Hz to complete static obstacle recognition and marking. It then uses a Kalman filter algorithm to predict the motion trajectories of surrounding dynamic obstacles such as pedestrians, marking the grid areas occupied by dynamic obstacles in the future as impassable regions. This constructs a three-dimensional locally rolling spatiotemporal grid map containing X-axis and Y-axis spatial coordinates and a T-axis time slice. ,satisfy: Representing spatial coordinates At any moment Occupied by obstacles; Representing spatial coordinates At any moment This is a passable area. For the three-dimensional raster state values of the locally scrolling spatiotemporal raster map; These are the X and Y coordinates of the spatial plane. For each moment in a time slice.
[0023] The map is updated and scrolled at a preset frequency of 10Hz. Simultaneously, four status parameters are extracted: the robot's current movement speed, task urgency, remaining battery percentage, and remaining task time. These parameters provide basic data support for subsequent trajectory planning, conflict detection, and dynamic priority calculation.
[0024] S2. Based on the local rolling spatiotemporal grid map generated in step S1, generate a predicted trajectory and trajectory pipeline. Pack data such as robot ID, predicted trajectory, task urgency, remaining battery power, and current motion status into a broadcast packet and send it to adjacent robot nodes within a preset communication range via a wireless ad hoc network at a preset frequency of 10Hz.
[0025] In this step, a desired trajectory is generated based on the local rolling spatiotemporal raster map constructed in step S1. The trajectory is a set of pose points with timestamps. .
[0026] By combining sensor errors and control accuracy, the nominal trajectory is extended into a cylindrical trajectory pipe with a radius of 0.2 meters, forming a trajectory uncertainty envelope. The robot's unique ID, current pose, current speed, desired trajectory, trajectory pipe radius, task urgency, remaining battery percentage, and remaining task time are packaged into a broadcast packet with a length of ≤256 bytes. Using a wireless ad hoc network point-to-point communication mode, the data packet is sent to adjacent robot nodes within the communication range at a frequency of 10Hz, providing external data for conflict detection and game decision-making for adjacent robots.
[0027] S3. The robot receives broadcast packets from neighboring robot nodes, maps the predicted trajectories of the neighboring robots to a local rolling spatiotemporal grid map, and detects spatiotemporal overlap of trajectory pipelines on a time-slice basis. If overlap exists, it is determined to be a potential path conflict, and the state data of the neighboring robots is passed to step S4. In this step, the robot parses the trajectory intent broadcast packets sent by neighboring robot nodes and extracts the desired trajectory. The trajectory of the neighboring machine is mapped onto the local rolling spatiotemporal grid. The trajectory points are expanded into cylindrical spatiotemporal pipes to represent trajectory uncertainty. The spatial projection relationship between the spatiotemporal pipes of the local and neighboring machines is detected by time-slice analysis. The spatial overlap determination condition is as follows: ; In the formula, For the first In a time slice, the robot With robots The center distance of the spatiotemporal pipeline; For the first Each discrete-time slice moment; The safe radius for the robot.
[0028] When the spatial overlap determination condition is met, that is, when the two spatiotemporal pipelines are determined to be in time overlap... If spatial overlap exists, and only the trajectory uncertainty envelope contacts without forming a spatial overlap, a low-speed warning is triggered and the movement speed is actively reduced by 30% to shrink the trajectory uncertainty envelope. If there is a spatial overlap area between the spatiotemporal pipelines of the local robot and the adjacent robot within any time slice, it is judged as a potential conflict event, triggering the subsequent game process.
[0029] S4. Based on the potential path conflict determination results output in step S3 and the state parameters of the local robot and neighboring robots, a multi-factor weighted evaluation function is used to calculate the dynamic priority coefficient, and the priority coefficient is dynamically corrected. The corrected coefficient is then substituted back into the game determination process to complete the stable decision-making of the right-of-way party and the avoidance party.
[0030] In this step, a multi-factor weighted evaluation function combined with the Nash equilibrium strategy in a non-cooperative game model is used to calculate the dynamic priority coefficient and complete the right-of-way allocation, thus calculating the robot's... The dynamic priority coefficient is calculated using the following formula: ; In the formula, For robots At any moment Dynamic priority coefficient; For robots The urgency of the tasks includes fire alarm, fault review, routine inspection and return to base for charging. Among them, fire alarm is assigned a value of 10, fault review is assigned a value of 8, routine inspection is assigned a value of 5, and return to base for charging is assigned a value of 1. For robots At any moment The remaining battery power; For robots The maximum battery capacity, which is used to characterize the low-charge protection weight; For robots At any moment The remaining time, which is used to characterize the urgency of the time; For robots The kinematic constraint factor is used to characterize the difficulty of a robot to stop suddenly or change lanes. , , , These are preset weighting coefficients for task urgency, low battery protection, time urgency, and kinematic constraints, respectively.
[0031] Set anti-shake threshold The robot calculates dynamic priority coefficients. The game theory approach is used to determine the right-of-way between the conflicting parties. If the dynamic priority coefficient of robot A and the dynamic priority coefficient of robot B satisfy... If so, then robot A is determined to be the winner of the game, i.e., the one with the right of way, and robot B is determined to be the one who concedes, i.e., the one who avoids the path. If the dynamic priority coefficient of robot B and the dynamic priority coefficient of robot A satisfy... The system determines that robot B has the right-of-way and robot A has the right-of-way.
[0032] The right-of-way player's strategy space is locked to maintain its original trajectory, while the yielding player's strategy space is to decelerate. ,wait , rewind A cost function is constructed based on the time cost, energy cost, and path deviation cost of the action. Choose one of the cost functions The minimum action to be performed is expressed by the formula: ; In the formula, The overall cost of the avoidance maneuver; Time cost represents the additional time required to perform the corresponding avoidance action; Energy cost represents the additional electricity consumption generated during the avoidance maneuver. The path deviation cost represents the degree of spatial deviation between the avoidance trajectory and the original desired trajectory. , , These are preset weighting coefficients corresponding to time, energy consumption, and path deviation.
[0033] If the dynamic priority coefficients of both parties satisfy The system is determined to be in a path conflict deadlock state, and the deadlock resolution mechanism is initiated: Set random disturbance coefficient A random perturbation factor is introduced to correct the dynamic priority coefficient. The correction formula is as follows: ; In the formula, For robots The original dynamic priority coefficient; This refers to the corrected dynamic priority coefficient; A uniformly distributed random number in the interval 0 to 1; By substituting the revised dynamic priority coefficients back into the game decision process, and with ≤3 iterations, the deadlock symmetry can be broken, and a stable decision between the right-of-way party and the yielding party can be achieved.
[0034] S5. The robot with the right-of-way maintains its original trajectory; the robot trying to avoid the collision first reduces its speed. If a collision cannot be avoided, it waits. If the collision still cannot be resolved, it performs local path replanning. After the path conflict is resolved, the robot resumes its original mission trajectory and motion state.
[0035] In this step, the robot with the right-of-way maintains its original desired trajectory and motion state and only makes a ±10% speed adjustment. It does not need to perform any avoidance actions. The robot that avoids the path performs the avoidance operation according to the minimum cost action selected in step S4. After the path conflict is completely resolved, both the robot with the right-of-way and the robot that avoids the path resume their original task trajectory and normal motion state, continue to perform the inspection task, and output chassis motion control commands such as linear velocity, angular velocity, steering angle, hovering, and path replanning, which directly drive the robot actuator to complete the coordinated action.
[0036] Example 2 The difference between this embodiment and Embodiment 1 is that the deadlock resolution mechanism is replaced by a timing priority method, and random perturbation is not used.
[0037] When the difference in priority coefficients between the two parties When the robot is powered on, the task is started. The robot with the earlier timestamp is determined to have the right-of-way, and the robot with the later timestamp is determined to give way. This method does not require random numbers, is more stable, and is suitable for industrial scenarios with higher deterministic requirements.
[0038] Therefore, the present invention adopts the above-mentioned spatiotemporal grid multi-robot dynamic priority game cooperative control method and system. The method realizes autonomous cooperation, efficient obstacle avoidance and stable operation of multi-robot clusters through a decentralized distributed architecture, spatiotemporal grid conflict prediction, dynamic priority non-cooperative game and deadlock automatic resolution mechanism.
[0039] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.
Claims
1. A spatiotemporal grid-based multi-robot dynamic priority game-theoretic cooperative control method, characterized in that, Includes the following steps: S1. Each inspection robot uses onboard sensors to perceive the surrounding environment in real time, constructs a local rolling spatiotemporal grid map, and simultaneously extracts task urgency, remaining battery power, remaining task time, and kinematic constraint factor state parameters; the local rolling spatiotemporal grid map is transmitted to the local unit S2 and S3, and the state parameters are transmitted to the local unit S4. S2. Based on the local rolling spatiotemporal grid map of S1, generate the desired trajectory with timestamps, expand the trajectory into a spatiotemporal pipeline representing uncertainty, package the robot identifier, desired trajectory and state parameters into a broadcast packet, and send it to the neighboring machine at a preset frequency; the broadcast packet is received by the neighboring machine S3. S3. Receive the neighbor broadcast packet from S2, map the neighbor's expected trajectory to the local rolling spatiotemporal grid, and detect the overlap of the spatiotemporal pipeline space of both parties on a time-by-time slice basis. If there is an overlap, it is determined to be a potential conflict event. The conflict determination result and the neighboring machine's state parameters are transmitted to the local S4 to trigger the game; S4. Calculate the dynamic priority coefficient based on the local state parameters of S1 and the neighboring state parameters of S3, and allocate road rights using a non-cooperative game Nash equilibrium strategy; when the two machines meet, compare the priority coefficients to determine the winner and the conceding party. The winner maintains the original trajectory, and the conceding party performs the least costly avoidance action; when the difference in priority coefficients is within a preset threshold, introduce a random disturbance correction coefficient and re-determine to resolve the deadlock; the road rights allocation result and avoidance command are transmitted to the local machine S5. S5. Based on the right-of-way result and avoidance command in S4, the winning party maintains its original trajectory and motion state, while the yielding party performs an avoidance action; after the conflict is resolved, the original mission trajectory is restored, and motion control commands are output to drive the chassis actuators.
2. The spatiotemporal grid multi-robot dynamic priority game cooperative control method according to claim 1, characterized in that, In S1, the local scrolling spatiotemporal grid map is a three-dimensional matrix with a time slice step of 0.5 seconds, covering a future time window of 0-5 seconds, and is updated at a frequency of 10Hz. Local scrolling spatiotemporal raster map satisfy: When spatial coordinates At any moment When occupied by obstacles When spatial coordinates At any moment When it is a passable area ; in, For the three-dimensional raster state values of the locally scrolling spatiotemporal raster map; These are the X and Y coordinates of the spatial plane. For each moment in a time slice.
3. The spatiotemporal grid multi-robot dynamic priority game cooperative control method according to claim 2, characterized in that, In S2, the desired trajectory is a set of pose points with timestamps. The spatiotemporal pipeline is a cylindrical trajectory envelope; the broadcast packet length does not exceed 256 bytes, and the transmission frequency is 10Hz.
4. The spatiotemporal grid multi-robot dynamic priority game cooperative control method according to claim 3, characterized in that, In S3, the condition for determining spatial overlap is: ； In the formula, For the first In a time slice, the robot With robots The center distance of the spatiotemporal pipeline; For the first Each discrete-time slice moment; The robot's safety radius; When the spacetime pipelines of the two robots are at any time When the conditions for spatial overlap are met, it is determined that spatial overlap exists; if only the trajectory uncertainty envelope contacts without forming spatial overlap, a low-speed warning is triggered, and the robot actively reduces its movement speed by 30% to shrink the trajectory uncertainty envelope.
5. The spatiotemporal grid multi-robot dynamic priority game cooperative control method according to claim 4, characterized in that, In S4, the dynamic priority coefficient is calculated using a multi-factor weighted average, which takes into account the urgency of the task, the remaining power, the remaining time of the task, and the kinematic constraint factor. The formula for calculating the dynamic priority coefficient is: ； In the formula, For robots At any moment Dynamic priority coefficient; For robots The urgency of the tasks includes fire alarm, fault review, routine inspection and return to base for charging. Among them, fire alarm is assigned a value of 10, fault review is assigned a value of 8, routine inspection is assigned a value of 5, and return to base for charging is assigned a value of 1. For robots At any moment The remaining battery power; For robots The maximum battery capacity, which is used to characterize the low-charge protection weight; For robots At any moment The remaining time, which is used to characterize the urgency of the time; For robots The kinematic constraint factor is used to characterize the difficulty of a robot to stop suddenly or change lanes. , , , These are preset weighting coefficients for task urgency, low battery protection, time urgency, and kinematic constraints, respectively.
6. The spatiotemporal grid multi-robot dynamic priority game cooperative control method according to claim 5, characterized in that, In S4, an anti-jitter threshold is set. The winning side locks the original trajectory, and the conceding side chooses the action with the minimum cost function from deceleration, waiting, and rerouting. The specific game process is as follows: Set anti-shake threshold The robot calculates dynamic priority coefficients. The game theory approach is used to determine the right-of-way between the conflicting parties. If the dynamic priority coefficient of robot A and the dynamic priority coefficient of robot B satisfy... If so, then robot A is determined to be the winner of the game, i.e., the one with the right of way, and robot B is determined to be the one who concedes, i.e., the one who avoids the path. If the dynamic priority coefficient of robot B and the dynamic priority coefficient of robot A satisfy... The system determines that robot B has the right-of-way and robot A has the right-of-way.
7. The spatiotemporal grid multi-robot dynamic priority game cooperative control method according to claim 6, characterized in that, In S4, deadlock resolution is achieved by adding a 0-1 random number to adjust the priority coefficient and re-playing the game to break the symmetric state. Specifically: If the dynamic priority coefficients of both parties satisfy If a path conflict deadlock is detected, the deadlock resolution mechanism will be initiated: Set random disturbance coefficient The corrected formula is as follows: ； In the formula, For robots The original dynamic priority coefficient; This refers to the corrected dynamic priority coefficient; A uniformly distributed random number in the interval 0 to 1; The revised dynamic priority coefficients are then reintroduced into the game decision process, with no more than three iterations, to complete a stable determination of the right-of-way party and the yielding party.
8. A spatiotemporal grid multi-robot dynamic priority game cooperative control system, applied to the spatiotemporal grid multi-robot dynamic priority game cooperative control method according to any one of claims 1-7, characterized in that, include: The local spatiotemporal grid perception construction module is used to control each inspection robot to perceive the surrounding environment in real time using onboard sensors, construct a local rolling spatiotemporal grid map, and simultaneously extract the robot's task urgency, remaining battery power, remaining task duration, and kinematic constraint factor state parameters. The spatiotemporal grid map is then output to the trajectory intent broadcasting module and the collision risk prediction module, and the state parameters are output to the dynamic priority game module. The trajectory intent broadcasting module, connected to the local spatiotemporal grid perception construction module, is used to generate a timestamped desired trajectory based on the local rolling spatiotemporal grid map, expand the trajectory into an uncertain spatiotemporal pipeline, package the robot identifier, desired trajectory and state parameters to form a broadcast packet, and send it to adjacent robot nodes within the communication range at a preset frequency. The collision risk prediction module is connected to the local spatiotemporal grid perception construction module and the trajectory intention broadcasting module. It is used to receive trajectory intention broadcast packets from adjacent robot nodes, map the expected trajectory of the neighboring robot to the local rolling spatiotemporal grid of the local robot, detect the spatial overlap of the spatiotemporal pipelines of both parties on a time-by-time slice and determine potential conflict events, and transmit the conflict determination results and the neighboring robot state parameters to the dynamic priority game module. The dynamic priority game module is connected to the local spatiotemporal grid perception construction module and the collision risk prediction module, respectively. It is used to calculate the dynamic priority coefficient by combining the state parameters of the local machine and the neighboring machine, complete the right-of-way allocation by adopting the Nash equilibrium strategy, compare the priority coefficients to determine the winner and the conceding party, introduce random perturbation correction coefficients to re-determine the deadlock state within the threshold range, and output the right-of-way allocation result and avoidance action instructions to the adaptive cooperative execution module. The adaptive collaborative execution module, connected to the dynamic priority game module, is used to control the winning party to maintain its original trajectory movement and the conceding party to perform the minimum cost avoidance action according to the right-of-way allocation result and avoidance instructions. After the conflict is resolved, the original task trajectory and running state of both parties are restored, and motion control instructions are output to drive the robot chassis execution mechanism.