CN-122018292-A - Event triggering-based multi-robot safety prediction control method and device

CN122018292ACN 122018292 ACN122018292 ACN 122018292ACN-122018292-A

Abstract

The invention relates to the technical field of mobile robot control, and discloses a multi-robot safety prediction control method and device based on event triggering, wherein the multi-robot safety prediction control method based on event triggering comprises the steps of acquiring real-time states of each robot and each obstacle in a target multi-robot system; based on the real-time state of each robot and each obstacle, forward propagation prediction is carried out by utilizing a graph neural network model to obtain a control obstacle function value, prediction uncertainty is calculated based on the control obstacle function value, the prediction uncertainty is compared with a preset event trigger threshold, if the prediction uncertainty accords with the preset event trigger threshold, a control sequence of the robot is updated to obtain a control variable at the current moment, and the robot is controlled to carry out obstacle avoidance control based on the control variable at the current moment. The invention improves the safety and stability of the multi-robot system and simultaneously remarkably improves the resource utilization efficiency.

Inventors

LIU YANJUN
WANG BIN
ZENG QIANG
LI SHU
LIU LEI
LI DAPENG
FU QIANG
CAO YUFEI

Assignees

辽宁工业大学

Dates

Publication Date: 20260512
Application Date: 20251230

Claims (10)

1. An event-triggering-based multi-robot safety prediction control method, which is characterized by comprising the following steps: acquiring real-time states of each robot and each obstacle in a target multi-robot system, wherein the target multi-robot system is a four-wheel independent drive four-wheel independent steering mobile robot system; Based on the real-time states of the robots and the obstacles, forward propagation prediction is carried out by using a graph neural network model, and a control obstacle function value is obtained; Calculating a prediction uncertainty based on the control obstacle function value, and comparing the prediction uncertainty with a preset event trigger threshold; If the prediction uncertainty accords with the preset event triggering threshold, updating a robot control sequence to obtain a control variable at the current moment; and controlling the robot to perform obstacle avoidance control based on the control variable at the current moment.
2. The method of claim 1, wherein the graph neural network model includes an input feature encoding module, an input mapping module, a message passing module, an attention module, a message processing module, and a graph reading module, wherein the performing forward propagation prediction using the graph neural network model based on the real-time states of each robot and each obstacle to obtain the control obstacle function value includes: the real-time states of the robots and the obstacles are encoded into original node features through the input feature encoding module, and original edge features are determined based on the original node features; mapping the original node characteristics to Gao Weiyin spaces through the input mapping module to obtain initial node characteristics; Calculating edge message features by the message passing module based on adjacent node features in the original edge features and the initial node features; Weighting calculation is carried out on the side message characteristics through the attention module, so that the side message characteristics after attention enhancement are obtained; the message processing module is used for carrying out message aggregation on the edge message characteristics after the attention is enhanced to obtain updated node characteristics; and aggregating and mapping the updated node characteristics through the map reading module to obtain the control obstacle function value.
3. The method according to claim 2, wherein the aggregating the attention-enhanced edge message features by the message processing module to obtain updated node features comprises: aggregating the edge message characteristics after attention enhancement according to the index of the target node to obtain neighborhood information characteristics of each node; fusing the neighborhood information features of the nodes with the initial node features to obtain fused node features; And carrying out layer normalization processing on the node characteristics after the fusion processing to obtain the updated node characteristics.
4. The method of claim 1, wherein the calculating a prediction uncertainty based on the control obstacle function value and comparing the prediction uncertainty to a preset event trigger threshold comprises: And acquiring a control obstacle mean value, calculating the prediction uncertainty based on the control obstacle function value and the control obstacle mean value, and comparing the prediction uncertainty with a preset event trigger threshold.
5. The method according to claim 1, wherein updating the robot control sequence to obtain the control variable at the current time if the prediction uncertainty meets the preset event trigger threshold comprises: constructing a kinematic model of the four-wheel independent driving four-wheel independent steering mobile robot; Establishing a robot safety prediction control model based on the kinematic model of the four-wheel independent driving four-wheel independent steering mobile robot; solving the robot safety prediction control model to obtain a control increment sequence at the current moment; and obtaining a control variable of the previous moment, and adding the first control variable in the control increment sequence of the current moment with the control variable of the previous moment to obtain the control variable of the current moment.
6. The method of claim 5, wherein the establishing a robot safety predictive control model based on the kinematic model of the four-wheel independently driven four-wheel independently steered mobile robot comprises: Establishing a robot motion prediction model based on the kinematic model of the four-wheel independent driving four-wheel independent steering mobile robot; based on the robot motion prediction model, taking the minimum cost value as a target, taking the control increment to be solved as an optimization variable, and constructing an objective function; constructing robot obstacle avoidance safety constraints based on the control obstacle function values; And obtaining control variable constraint, control increment constraint and robot operation terminal set constraint, and taking the robot obstacle avoidance safety constraint, the control variable constraint, the control increment constraint and the robot operation terminal set constraint as constraint conditions of the objective function to obtain the robot safety prediction control model.
7. An event trigger based multi-robot safety predictive control device, the device comprising: the system comprises an acquisition module, a control module and a control module, wherein the acquisition module is used for acquiring real-time states of each robot and each obstacle in a target multi-robot system, and the target multi-robot system is a four-wheel independent driving four-wheel independent steering mobile robot system; The prediction module is used for performing forward propagation prediction by utilizing the graph neural network model based on the real-time states of the robots and the obstacles to obtain control obstacle function values; A comparison module for calculating a prediction uncertainty based on the control obstacle function value and comparing the prediction uncertainty with a preset event trigger threshold; The updating module is used for updating the robot control sequence to obtain a control variable at the current moment if the prediction uncertainty accords with the preset event trigger threshold; and the control module is used for controlling the robot to perform obstacle avoidance control based on the control variable at the current moment.
8. An electronic device, comprising: A memory and a processor, the memory and the processor being communicatively connected to each other, the memory having stored therein computer instructions, the processor executing the computer instructions to perform the event trigger based multi-robot safety predictive control method of any one of claims 1 to 6.
9. A computer-readable storage medium having stored thereon computer instructions for causing a computer to execute the event trigger based multi-robot safety predictive control method of any one of claims 1 to 6.
10. A computer program product comprising computer instructions for causing a computer to perform the event trigger based multi-robot safety predictive control method of any one of claims 1 to 6.

Description

Event triggering-based multi-robot safety prediction control method and device Technical Field The invention relates to the technical field of mobile robot control, in particular to a multi-robot safety prediction control method and device based on event triggering. Background The multi-robot system is increasingly widely applied to actual scenes such as warehouse logistics, intelligent inspection, unmanned distribution and the like. In these scenarios, the multi-robot system can achieve efficient task execution and resource utilization through task decomposition and collaborative decision-making. However, with the increase of the number of robots and the complexity of the environment, how to increase the operation efficiency and the resource utilization rate of the system on the premise of ensuring the task completion and the safety is the focus of current research. Disclosure of Invention The invention provides a multi-robot safety prediction control method and device based on event triggering, which are used for solving the problems that a multi-robot system is difficult to realize in centralized control, the computational complexity is rapidly increased and communication and computational resources are limited when the robot scale is enlarged and the environmental complexity is improved. In a first aspect, the present invention provides a method for controlling multi-robot safety prediction based on event triggering, the method comprising: acquiring real-time states of each robot and each obstacle in a target multi-robot system, wherein the target multi-robot system is a four-wheel independent driving four-wheel independent steering mobile robot system; Based on the real-time state of each robot and each obstacle, forward propagation prediction is carried out by using a graph neural network model, so as to obtain a control obstacle function value; calculating a prediction uncertainty based on the control obstacle function value, and comparing the prediction uncertainty with a preset event trigger threshold; If the prediction uncertainty accords with a preset event trigger threshold, updating the robot control sequence to obtain a control variable at the current moment; and controlling the robot to perform obstacle avoidance control based on the control variable at the current moment. According to the event-triggered multi-robot safety prediction control method, forward propagation prediction is conducted on real-time states of each robot and each obstacle in a four-wheel independent-drive four-wheel independent-steering mobile robot system by utilizing a graph neural network model to obtain a control obstacle function value, prediction uncertainty is calculated based on the control obstacle function value, the prediction uncertainty is compared with a preset event trigger threshold, if the prediction uncertainty accords with the preset event trigger threshold, a robot control sequence is updated, distributed optimization is conducted only when needed based on an event trigger mechanism, network bandwidth occupation and node calculation load are effectively reduced, resource utilization efficiency of the system is improved, calculation complexity is reduced, and communication and calculation resource consumption of the system can be remarkably reduced on the premise that safety and collaborative task completion of the multi-robot system are guaranteed by utilizing the model prediction control method based on the graph neural network control obstacle function and the event trigger. In a second aspect, the present invention provides an event trigger-based multi-robot safety prediction control device, the device comprising: the system comprises an acquisition module, a control module and a control module, wherein the acquisition module is used for acquiring real-time states of each robot and each obstacle in a target multi-robot system, and the target multi-robot system is a four-wheel independent driving four-wheel independent steering mobile robot system; the prediction module is used for performing forward propagation prediction by utilizing the graph neural network model based on the real-time states of each robot and each obstacle to obtain a control obstacle function value; the comparison module is used for calculating the prediction uncertainty based on the control obstacle function value and comparing the prediction uncertainty with a preset event trigger threshold; the updating module is used for updating the robot control sequence to obtain a control variable at the current moment if the prediction uncertainty accords with a preset event trigger threshold; And the control module is used for controlling the robot to perform obstacle avoidance control based on the control variable at the current moment. In a third aspect, the invention provides an electronic device, which comprises a memory and a processor, wherein the memory and the processor are in communication connection, the memory stores compu