Multi-source safety event response control method for a composite robot

Abstract

This invention discloses a multi-source safety event response control method for composite robots, specifically relating to the field of robot safety handling control technology. It addresses the problem of weak connection and difficulty in forming closed-loop response control between existing multi-source anomaly information in chemical plants and the docking, approach, operation, and verification processes of composite robots. By uniformly timing and merging gas anomaly signals, temperature anomaly signals, equipment status anomaly signals, and electrical alarm signals within the target work unit, a multi-source event convergence record is formed. Based on this, an event response constraint package is generated, completing target docking position screening, operation anchor generation, object operation execution, result verification, and re-adjustment for failures. This achieves the effect of transforming dispersed anomaly information on-site into a continuous response link directly corresponding to the docking, approach, operation, and verification processes of the composite robot, solving the problem of disconnection between multi-source safety events and robot handling actions in existing technologies, making it difficult to form a stable closed-loop response control.

Description

Technical Field

[0001] This invention relates to the field of robot safety handling and control technology, specifically to a multi-source safety event response and control method for a composite robot. Background Technology

[0002] In chemical plant areas, valves, cabinets, filling interfaces, and equipment control panels are typically equipped with various monitoring methods, including gas monitoring, temperature monitoring, status acquisition, and electrical alarms, to detect safety incidents such as leaks, temperature rises, abnormal locations, and alarm linkage anomalies. Simultaneously, hybrid robots equipped with mobile chassis and robotic arms are increasingly being used for plant area inspections and on-site handling, replacing personnel in high-risk areas for observation, contact, and basic operations. Existing on-site systems can generally acquire information on gas anomalies, temperature anomalies, equipment status anomalies, and electrical alarms, and robots can perform fixed-point inspections, target recognition, or single-step actions. However, these information and actions mostly operate independently. The front-end monitoring focuses on anomaly detection, while the robot focuses on path positioning, target recognition, or pre-set action execution. There is a lack of continuous connection between the two, centered around the same object of concern and the same on-site boundary conditions.

[0003] In existing technologies, one approach involves the monitoring system reporting anomalies from different sources separately, with the higher-level system or on-duty personnel then determining, based on experience, whether to dispatch a robot or arrange for manual handling. Another approach involves the robot performing actions such as approaching, grasping, pressing, and rotating based on visual recognition results or preset task templates. While these methods can address anomaly collection and action execution separately, a common gap often exists in real-world chemical plant operations: multi-source safety event information is not consistently transformed into unified constraints directly corresponding to the robot's docking position, approach direction, operating posture, action amplitude, and result verification. Consequently, even if a combination of gas anomalies, temperature anomalies, equipment status anomalies, and electrical alarms occurs within the same site, the robot's subsequent docking, approach, and operation may still follow a generic template, making it difficult to establish a continuous link between the anomaly source, the object of handling, the operation path, and result confirmation within the same site boundary.

[0004] Therefore, the main problem faced by existing technologies is not whether a single anomaly can be detected, nor whether a robot can complete a single action, but rather the lack of a control method that can converge scattered anomaly information into a continuous response basis oriented towards specific objects when facing multi-source safety events in chemical plants, and further constrain the robot's docking, approach, operation, verification, and adjustment processes. Due to this deficiency, existing systems are prone to problems with poor coordination between multi-source safety events and robot actions, making it difficult to form a stable and continuous closed-loop response control under the same site constraints. This is precisely the key technical problem that needs to be addressed in the future. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a multi-source safety event response control method for composite robots, thereby solving the problems mentioned in the background section.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a multi-source safety event response control method for a composite robot, comprising:

[0007] S1. Collect gas abnormality signals, temperature abnormality signals, equipment status abnormality signals, and electrical alarm signals of the target work unit within a preset time window, merge them according to a unified time reference and work unit identifier, and generate a multi-source event convergence record.

[0008] S2. Extract risk boundaries, order of disposal objects, docking posture constraints, operation posture constraints, action range constraints, and result verification constraints from the multi-source event convergence records, and generate an event response constraint package.

[0009] S3. Filter the target docking position of the composite robot according to the event response constraint package, and generate the corresponding operation anchor record by combining the target docking position, the position of the object to be dealt with, and the operation posture constraint.

[0010] S4. Control the composite robot to reach the target docking position, drive the robotic arm to perform observation and confirmation, trial contact and restricted actions on the object to be handled in sequence, and form a corresponding object operation record;

[0011] S5. Retrieve the current status feedback based on the object operation record, execute actions to complete the verification and risk convergence verification according to the result verification constraints, and generate an event response judgment record;

[0012] S6. When the judgment conclusion of the event response judgment record is any one of the following: action not completed, risk not converged, or pending verification, adjust the target docking position, operation anchor position, and disposal object sequence according to the event response constraint package, and execute the next round of response control.

[0013] Furthermore, S1 includes:

[0014] The gas anomaly signal, temperature anomaly signal, equipment status anomaly signal, and electrical alarm signal are processed in the order of alignment, noise reduction, and completion.

[0015] When at least two types of valid abnormal signals are obtained under the same target work unit, the same preset time window, and the same rule version identifier, one currently valid multi-source event convergence record is generated.

[0016] Furthermore, S2 includes:

[0017] Extract risk boundaries and the order of action targets based on the convergence records of multi-source events;

[0018] The risk boundary is formed by the location of the abnormal measurement point corresponding to the gas abnormal signal, the temperature rise part corresponding to the temperature abnormal signal, the location of the abnormal object corresponding to the equipment status abnormal signal, and the alarm source location corresponding to the electrical alarm signal.

[0019] The order of objects to be dealt with is determined based on the number of categories that are jointly pointed to by valid abnormal signals, the on-site distance between the object to be dealt with and the center of the risk boundary, the order of objects that can be arranged in a continuous and close manner on the current docking surface, and the order of objects in the preset number of the target work unit.

[0020] Furthermore, S2 also includes:

[0021] Extract the operation posture constraints based on the order of the objects to be handled, the location of the objects to be handled, the docking posture constraints, the abnormal signals of the equipment status, and the object installation structure records;

[0022] Extract motion amplitude constraints based on the current status value of the equipment status anomaly signal, the object installation structure record, and the on-site operation baseline record;

[0023] The event response constraint package is written in the following order: risk boundary, order of objects to be dealt with, docking posture constraint, operation posture constraint, movement range constraint, and result verification constraint.

[0024] Furthermore, S3 includes:

[0025] The parking position range, chassis orientation range, approach direction requirements, and evacuation direction retention requirements are retrieved from the parking posture constraint record and compared item by item with the accessible area, restricted access area, and prohibited stay area in the risk boundary record to screen the target parking position;

[0026] Then, by combining the target docking position, the location of the object to be disposed of, and the operation posture constraints, an operation anchor record is generated. The operation anchor record includes the target docking position number, the location number of the object to be disposed of, the end approach direction, the target height range, and the wrist allowable posture range.

[0027] Furthermore, S4 includes:

[0028] After the on-site control side controls the composite robot to reach the target docking position, it drives the robotic arm to perform observation and confirmation, probing contact, and restricted actions in sequence around the object to be handled.

[0029] After observation and confirmation, the process proceeds to tentative contact; after successful tentative contact, the process proceeds to restricted actions.

[0030] Furthermore, S4 also includes:

[0031] After observation and confirmation, tentative contact, and execution of restricted actions are completed, the on-site control side generates an object operation record;

[0032] The object operation record includes the target docking position number, operation anchor record number, current object identification, object location number, observation confirmation result, trial contact result, restricted action direction, and object operation result identifier.

[0033] Furthermore, S5 includes:

[0034] When there are multiple object operation records corresponding to the same current object, only the current valid object operation record with the most recent write time is taken as the only upstream record. Records with object operation results marked as pending verification are not used as the basis for action completion verification.

[0035] Retrieve current status feedback during the confirmation period. Current status feedback includes object status feedback and abnormal signal feedback.

[0036] First, perform the action to complete the verification based on the result verification constraint, and then perform the risk convergence verification based on the result verification constraint.

[0037] The judgment conclusion is determined based on the action completion verification results and risk convergence verification results, and an event response judgment record is generated.

[0038] Furthermore, S6 includes:

[0039] When the judgment conclusion of the event response judgment record is any one of the following: action not completed, risk not converged, or pending verification, under the same target work unit, the same rule version, and the same event response constraint package number, the target docking position, operation anchor position, and order of handling objects shall be readjusted according to the current valid event response constraint package.

[0040] When the target docking position changes, the original operation anchor position record becomes invalid and needs to be readjusted;

[0041] Based on this, the next round of response control registration records will be generated.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. By uniformly merging and timing-synchronizing gas anomaly signals, temperature anomaly signals, equipment status anomaly signals, and electrical alarm signals within the target work unit, a multi-source event convergence record is formed. Based on this, risk boundaries, the order of handling objects, docking posture constraints, operating posture constraints, action amplitude constraints, and result verification constraints are further extracted to generate an event response constraint package. Based on this, target docking position screening, operation anchor generation, object operation execution, result verification, and re-adjustment for failures are completed. This achieves the effect of transforming dispersed anomaly information in the chemical site into a continuous response link directly corresponding to the actual docking, approach, operation, and verification process of the composite robot. This solves the core technical problem in the existing technology of the disconnect between multi-source safety events and robot handling actions, and the difficulty in forming a stable closed-loop response control under the same site constraints.

[0044] 2. By performing actions based on the result verification constraints after object operation, the verification and risk convergence verification are completed. If the verification fails, the target docking position, operation anchor position and the order of the objects to be handled are readjusted based on the original event response constraint package to form the next round of response control. This achieves the effect of continuously correcting the handling path and improving the on-site response targeting and handling reliability without deviating from the current target operation unit and existing constraint boundaries. This solves the problem in the existing technology that the robot lacks an adaptive correction mechanism for risk convergence results after completing a single operation, and that the action is completed but the risk is not eliminated or the handling is interrupted. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the overall process of a multi-source safety event response control method for a composite robot;

[0046] Figure 2 A schematic diagram is generated for extracting event response constraint packages;

[0047] Figure 3 This is a schematic diagram of the next round of response control in the case of failure. Detailed Implementation

[0048] 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, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] Example: Combined with Appendix Figures 1-3 This embodiment provides a multi-source safety event response control method for a composite robot, including:

[0050] S1. Collect abnormal gas signals, abnormal temperature signals, abnormal equipment status signals, and electrical alarm signals from the target work unit within a preset time window. Merge these signals according to a unified time reference and work unit identifier to generate a multi-source event convergence record. The specific implementation is as follows:

[0051] After the composite robot enters the chemical plant area for operation, it does not directly enter the disposal scheduling. Instead, it first forms a multi-source event convergence record around the target work unit. The target work unit is taken from the site and is the smallest independent disposal area that can be continuously observed or operated by the composite robot after one stop. Under one rule version, it corresponds to only one work unit identifier and does not cover two valve positions, two panel cabinets, two filling interfaces, or two control panel operation positions at the same time. Its corresponding on-site position can be one valve group operation position, one panel cabinet operation position, one filling interface position, or one equipment control panel operation position. The execution subject is the joint acquisition side of the field acquisition device on the composite robot body and the original monitoring equipment in the plant area. The action period is the continuous running period after the composite robot is in work standby, inspection passing by, or receiving on-site abnormal acquisition push. The upstream source is limited to the gas monitoring point, temperature monitoring point, equipment status acquisition point, and electrical alarm acquisition point associated with the current target work unit. It does not mix data across work units.

[0052] This step first collects gas anomaly signals from the target work unit within a preset time window. The preset time window is a fixed-length collection interval starting from the collection time of the first valid anomaly record. Before the collection interval closes, subsequent valid anomaly records belonging to the same target work unit are added to the current processing link. After the preset end time is reached, the link is closed, and subsequent newly added anomaly records are no longer merged into the current record but are transferred to the next preset time window for reprocessing. The gas anomaly signal only takes the original anomaly record formed by the on-site gas monitoring device according to the current rule version. The anomaly conclusion is not recalculated in this step. Specifically, it can be collected by a fixed gas probe, a mobile supplementary probe, or a robot-carried gas detection head. The on-site record name is uniformly "Gas Anomaly Signal", and the concentration unit is recorded as volume fraction or mass concentration. The collection rhythm maintains the original reporting rhythm of the source device. The first record specifies the probe source, installation position, collection time, corresponding work unit identifier, and rule version identifier. If there is an offline breakpoint within the time window, only missing markers are allowed to be filled in, and concentration values ​​are not allowed to be filled in.

[0053] Subsequently, temperature anomaly signals of the target work unit within a preset time window are collected. The temperature anomaly signals are only the original anomaly records generated by the on-site temperature measuring device according to the current rule version. No further anomaly is derived from the original temperature sequence in this step. The source can be a thermal imager, a point-type temperature measuring element, or a temperature acquisition device on the surface of the equipment. The temperature unit is uniformly ℃. When writing, the original measuring point number, acquisition rhythm, acquisition time, corresponding work unit identifier, and rule version identifier are retained. In chemical sites, the temperature rise around the valve stuffing box, the temperature rise at the busbar connection of the panel cabinet, and the temperature rise at the connection of the filling joint can all be used as the on-site landing points of the temperature anomaly signals. When sorting, first align according to the acquisition time, and then check whether the location of the measuring point is within the boundary of the target work unit. Records outside the boundary are not included in the current sorting link.

[0054] Then, collect the abnormal equipment status signals of the target work unit within the preset time window. The abnormal equipment status signals are only taken from the status deviation records related to the object corresponding to the target work unit. When the object corresponding to the target work unit has its own status acquisition device, the status record formed by the device is used as the source. When the object corresponding to the target work unit does not have its own status acquisition device, the visual status recognition record formed within the same preset time window is retrieved as the source. The visual status recognition record only gives the object's current position, current gear, or current open / closed state, and does not directly generate abnormal conclusions. The abnormal conclusions are still written after comparing with the target state according to the current rule version. In the valve scenario, it can be valve position inconsistency, handle stop abnormality, or interface looseness. In the cabinet scenario, it can be trolley position abnormality, switch position abnormality, or indicator status abnormality. In the equipment panel scenario, it can be button state not switched or knob gear stop abnormality. When writing the record, the status source, original unit, collection time, corresponding object position, corresponding work unit identifier, and rule version identifier are specified simultaneously. For visual status recognition records, only missing markers are allowed to be filled in. Automatic continuation of historical status is not allowed when recognition fails.

[0055] Then, electrical alarm signals of the target work unit within a preset time window are collected. The electrical alarm signals are only the original field alarm records that are pre-correlated between the source equipment and the target work unit. The alarm level and alarm content are not rewritten in this step. The fields written include at least the alarm source equipment, alarm code, occurrence time, recovery time, corresponding work unit identifier and rule version identifier. Near the opening and closing operation position, control circuit disconnection alarm, position contact abnormality alarm and partial discharge associated alarm can all fall into this signal source.

[0056] After the above four types of signals are collected, they are merged according to the unified time reference and the work unit identifier. The unified time reference is the current standard time published by the field control side. The local collection records of the composite robot are uniformly converted to the current standard time before being written. When the original record of the source device only carries the local time, the time conversion is completed first according to the most recent valid time offset, and then it enters the merging. Records that cannot be converted to the unified time reference do not enter the current merging link.

[0057] The order of processing before merging is fixed as follows: alignment, denoising, and completion. Alignment only processes the correspondence between the acquisition time and the work unit, without rewriting the original record values. Denoising only removes records with out-of-range measurements, incomparable times, untraceable sources, and failed work unit mappings, without smoothing and rewriting fluctuation records that are still within the legal range. Completion only adds missing markers, source missing descriptions, and rule version identifiers, without adding the four core fields: gas concentration, temperature, equipment status, and electrical alarm content. If any core field is missing and cannot be filled in within the current preset time window, the processing of that record is stopped and it is transferred to the record area to be checked, without being forcibly merged into the convergence result. Two rule version identifiers are not allowed to be mixed in the same multi-source event convergence record. The rule version identifier must at least correspond to the anomaly establishment caliber, work unit mapping caliber, time conversion caliber, denoising caliber, completion caliber, and deduplication caliber used in the current step.

[0058] The physical meaning of merging is to combine gas anomaly signals, temperature anomaly signals, equipment status anomaly signals, and electrical alarm signals obtained within the same target work unit and the same preset time window into a single on-site anomaly observation result. Merging does not cross units, time windows, or rule versions; it uses a combination of sequential writing and deduplication writing. Only one valid record is retained for the same source, the same acquisition time, and the same record content. Repeated reporting only increases the arrival count and the most recent arrival time, without generating new core records. Within the same target work unit and the same preset time window, at least two... When a valid anomaly signal of type 1 is obtained, a multi-source event convergence record is generated. When only one valid anomaly signal of type 2 is obtained, a single-source supplementary record is formed and not written into the multi-source event convergence record. When a second valid anomaly signal of type 2 enters and completes merging, an initial version of the current multi-source event convergence record is generated first. Before the preset time window closes, any subsequent newly added valid anomaly signals are only appended to the extended segment of this record. After the preset time window closes, the record is sealed and no further appends are accepted. Under the same target work unit, the same preset time window, and the same rule version identifier, only one currently valid multi-source event convergence record is allowed to be generated.

[0059] The resulting multi-source event convergence record includes at least the target work unit identifier, preset time window start and end times, gas anomaly signal set, temperature anomaly signal set, equipment status anomaly signal set, electrical alarm signal set, rule version identifier, generation time, and evidence trace index. The evidence trace index points at least to the original gas record, original temperature record, original equipment status record, original electrical alarm record, and current processing log that participated in the current merging. Its storage location can be set in the local work record area of ​​the composite robot and synchronously written to the field control side record library. After successful writing, the record number, writing time, and version lock result are returned for direct retrieval in the next stage.

[0060] This step only collects, organizes, merges, and writes on-site anomaly records within the authorized acquisition range. It does not issue control commands to on-site equipment or rewrite the original anomaly conclusions already formed by the source equipment.

[0061] When a slight evaporation occurs at a filling interface on site, a gas abnormality signal can be obtained first, followed by an abnormal temperature signal on the outer wall of the connector, an abnormal interface locking status signal, and an electrical alarm signal from the local control box. The above records are merged into one multi-source event convergence record within the same preset time window according to the same work unit identifier and then called in the next stage.

[0062] Preferably, the preset time window can be set to 30s to 180s, the gas anomaly signal can adopt a reporting rhythm of 1s to 5s, the temperature anomaly signal can adopt a reporting rhythm of 1s to 10s, the equipment status anomaly signal can be triggered by status change, and the electrical alarm signal can be triggered by event. In a preferred embodiment, on one valve group operating position, the preset time window is set to 60s, the gas concentration recording unit is set to ppm, the temperature recording unit is set to ℃, and a total of 4 gas anomaly signals, 3 temperature anomaly signals, 2 equipment status anomaly signals, and 1 electrical alarm signal are obtained in the same work unit. After alignment, noise reduction, and completion, a multi-source event convergence record is generated, and the overall writing delay is controlled within 500ms. Without changing the technical concept of this step, the source of the gas anomaly signal can be replaced by a fixed probe to a robot-following supplementary probe, and the source of the temperature anomaly signal can be replaced by a thermal imager to a point temperature measuring element. As long as the multi-source event convergence record is still generated according to the unified timing reference and work unit identification, it belongs to the implementable form of this step.

[0063] S2. Extract risk boundaries, order of disposal objects, docking posture constraints, operation posture constraints, action range constraints, and result verification constraints from the multi-source event convergence records to generate an event response constraint package. The specific implementation is as follows:

[0064] After the multi-source event convergence record, which has been formed and sealed in the previous step, is retrieved by the current work session, no action is issued directly. Instead, risk boundaries, order of disposal objects, docking posture constraints, operating posture constraints, action range constraints, and result verification constraints are extracted item by item around the multi-source event convergence record, and an event response constraint package is generated accordingly. The executing entity is the field control side corresponding to the current work session of the composite robot, and the action location is the target work unit corresponding to the multi-source event convergence record. The calling time period is limited to the continuous processing interval from the closing of the preset time window corresponding to the multi-source event convergence record to the calling of the next step. The calling fields include at least the target work unit identifier, the start and end time of the preset time window, the gas abnormal signal set, the temperature abnormal signal set, the equipment status abnormal signal set, the electrical alarm signal set, the rule version identifier, and the evidence trace index. No new record sources outside the current step are introduced.

[0065] This step first extracts the risk boundary, which is fixedly formed by the set of on-site locations corresponding to all valid anomaly sources within the current target work unit. The extraction order is fixed as follows: first locate the abnormal measurement point location corresponding to the gas anomaly signal, the temperature rise location corresponding to the temperature anomaly signal, the abnormal object location corresponding to the equipment status anomaly signal, and the alarm source location corresponding to the electrical alarm signal. Then, verify the correspondence between each location and the target work unit boundary, and subsequently form the accessible zone, restricted access zone, and prohibited stay zone. Among them, the gas anomaly measurement point location is taken from the installation position or accompanying supplementary point location in the original gas record, and the temperature rise location is taken from the measurement point number and on-site location in the original temperature record. Mapping: The location of the abnormal object is taken from the object location field in the status source record, and the location of the alarm source is taken from the pre-association table between the alarm source device and the target work unit. The expansion range is fixed according to the risk boundary expansion caliber corresponding to the current rule version. When there is no equipment entity separation between two abnormal source locations and their expansion ranges intersect, they are merged into the same continuous boundary area. When there is equipment entity separation or the expansion ranges do not intersect, they are retained as independent boundary areas and no cross-area splicing is performed. If an abnormal source record lacks on-site location mapping and cannot be traced by the evidence trace index, the record will stop participating in risk boundary extraction and will be transferred to the supplementary recording, and will not be supplemented with the presumed location.

[0066] Subsequently, the order of disposal objects is extracted. The disposal objects are limited to on-site objects that the composite robot can directly observe or contact with. Only abnormal objects directly associated with the current multi-source event convergence record are selected. Monitoring equipment bodies that only provide abnormal source information but do not accept subsequent robot contact are not considered as disposal objects. The order of disposal objects is determined by the following fixed order: first, they are arranged in descending order according to the number of categories jointly pointed to by valid abnormal signals. If the number of categories is the same, they are arranged in ascending order according to the on-site distance between the disposal object and the center of the risk boundary. If the distance is the same, objects that can form continuous approach on the current docking surface are given priority. If they are still the same, they are arranged according to the order of the objects in the preset number of the target work unit. When the same object is repeatedly pointed to by multiple abnormal records, only one position is retained. The repeated records are only used as evidence to strengthen that position.

[0067] Next, the docking posture constraints are extracted. The docking posture constraints are the posture restrictions that enable the composite robot chassis to form an effective docking within the current risk boundary. During extraction, they are obtained jointly from the on-site access baseline, ground dockable area, work surface orientation record, and current risk boundary corresponding to the target work unit, without regenerating the access path. The on-site access baseline and work surface orientation record are fixedly taken from the effective on-site reference record corresponding to the current target work unit. The posture constraints include at least the chassis orientation range, docking position range, approach direction requirements with respect to the object being handled, and withdrawal direction retention requirements. The docking position range is recorded according to the pre-marked ground position coordinate interval within the target work unit, the chassis orientation range is recorded according to the angle interval relative to the installation front of the object being handled, and the withdrawal direction retention requirements are recorded according to the minimum retention distance with no obstruction in at least one of the chassis retreat direction or lateral retreat direction. The value range of the docking posture constraints must not exceed the accessible area in the risk boundary.

[0068] Next, extract the operational posture constraints. These constraints are the posture limitations that allow the robotic arm's end effector to establish contact or observation in front of the current object being handled. The extraction is based on the object's order, position, docking posture constraints, abnormal equipment status signals, and the object's installation structure record. The object's installation structure record permanently records the object's installation position, orientation, structure type, accessible surfaces, prohibited surfaces, and travel boundaries. If this record is missing and cannot be traced back through evidence traces, no operational posture constraints are formed for the corresponding object, and the object is transferred to the pending verification stage. During extraction, records are made separately according to the object's structure type. For rotating parts, at least the axial approach direction range, grip center height range, and wrist rotation allowable range are recorded. For pressing parts, at least the forward approach direction range, pressing axis direction, and end effector entry depth range are recorded. For plug-in parts, at least the plug-in axis direction range, alignment height range, and swing allowable range are recorded. The structure type of the current object being handled is taken from the object's installation structure record and is not temporarily inferred in this step. Furthermore, the operational posture constraints must not cause the end effector to enter the prohibited area within the risk boundary.

[0069] Then, the action amplitude constraints are extracted. These constraints limit the allowable displacement, rotation, pressing, or pulling amplitude for the current object being handled. They are obtained from the current status value of the equipment status anomaly signal, the object's installation structure record, and the field operation reference record. The field operation reference record is fixedly taken from the normal operation status reference record that has taken effect under the current rule version for the current target work unit. The normal operation status reference record is fixedly taken from the effective operation reference file of the corresponding object of the current target work unit, and at least includes the target stable state, allowable state range, installation orientation, and work surface orientation of the object being handled. Temporary manual input values ​​are not used as field operation reference records. The action amplitude constraints must simultaneously record the action direction, the upper limit of a single action amplitude, and the cumulative action amplitude boundary. The action direction is recorded according to the direction in which the object being handled changes from the current state to the target stable state. The upper limit of a single action amplitude must not exceed the cumulative action amplitude boundary, and the cumulative action amplitude boundary must not cross the termination position corresponding to the target stable state. If the current status value is missing and cannot be recovered from the evidence trace index within the same preset time window, then no action amplitude constraint is formed for that object, and the current object is transferred to pending verification.

[0070] Finally, the result verification constraints are extracted. These constraints are used to confirm whether the current handling has reached the specified state after the next step has completed the action. The extraction is based solely on the current multi-source event convergence record and the corresponding on-site operation benchmark record, without performing action confirmation in advance. The result verification constraints include at least object state change verification items and abnormal signal fallback verification items. The object state change verification items are recorded according to the target state value, allowable deviation range, and confirmation period. The abnormal signal fallback verification items are recorded according to the corresponding abnormal signal's release judgment condition and confirmation period. Each handling object corresponds to at least one set of object state change verification items. Each type of abnormal signal appearing in the current multi-source event convergence record corresponds to at least one set of abnormal signal fallback verification items. Furthermore, the action amplitude constraint must not cause the handling object's state to exceed the target stable state corresponding to the result verification constraint.

[0071] After the above six constraints are extracted, they are written into the event response constraint package in a fixed order: risk boundary, order of disposal objects, docking posture constraint, operating posture constraint, motion range constraint, and result verification constraint. The writing order is not changed, and constraint content from other target work units is not referenced across packages. The event response constraint package must contain at least the target work unit identifier, multi-source event convergence record number, rule version identifier, risk boundary record, order of disposal object record, docking posture constraint record, operating posture constraint record, motion range constraint record, result verification constraint record, generation time, and evidence trace index. Each constraint record is associated with a corresponding evidence trace index, and constraint records that cannot be traced back to the original basis are not allowed. When extraction is performed repeatedly under the same multi-source event convergence record, the same rule version identifier, and the same extraction basis content, a new event response constraint package is not generated; only the currently valid event response constraint package number is returned. An overriding version is only allowed to be generated when the extraction basis content changes and the original package has not yet been registered and called by the next stage.

[0072] On-site, when a valve group operating position simultaneously experiences valve position abnormality, valve body temperature rise, and adjacent control box alarm within the same preset time window, a continuous boundary area can be delineated within the corresponding target operating unit of the valve group. Then, the valve body is placed first in the order of handling objects. Next, the chassis is restricted to dock from the side opposite the valve handwheel, the robotic arm end is restricted to approach from the side avoiding the temperature rise area, and a single rotation angle is restricted not to exceed the remaining rotation angle range from the current half-open position to the closed position. The result verification constraint is written that the valve position should return to the closed position and the alarm source device should no longer maintain the current alarm state. This forms an event response constraint package that can be directly called by the next step.

[0073] Preferably, the risk boundary can be extended outward by 0.3m to 1.5m from the location of the abnormal source within the current target work unit to form a restricted access zone. The allowable deviation of the chassis orientation can be set to ±15° to ±45°. The end-point approach height range can be written within 0.05m to 0.3m above and below the center height of the object to be dealt with. The upper limit of the single action amplitude can be recorded as 0.2 to 1.0 times the difference between the current state and the target stable state. The confirmation period in the result verification constraint can be set to 10s to 120s. In a preferred embodiment, at one filling interface position, the multi-source event convergence record includes 3 gas abnormality signals, 2 temperature abnormality signals, 2 equipment status abnormality signals, and 1 electrical alarm signal, based on which... Extract one continuous boundary zone, two disposal object positions, one set of chassis docking orientation ranges, one set of end-axis approach intervals, one set of action direction and amplitude boundaries, and two sets of result verification constraints. Finally, generate one event response constraint package and write it to the field control side record library within 300ms. Without changing the technical concept of this step, the disposal object position can be obtained from the fixed installation position record or from the evidence trace index in the current multi-source event convergence record. As long as the risk boundary, disposal object sequence, docking posture constraints, operation posture constraints, action amplitude constraints, and result verification constraints are extracted item by item according to the multi-source event convergence record and the event response constraint package is generated, it belongs to the implementable form of this step.

[0074] S3. Based on the event response constraint package, filter the target docking position of the composite robot, and combine the target docking position, the position of the object to be handled, and the operation posture constraints to generate the corresponding operation anchor record. The specific implementation is as follows:

[0075] After the event response constraint package is generated and registered as the current valid version, the current operation session of the composite robot does not immediately enter object contact. Instead, it first filters the target docking position based on the event response constraint package, and then generates the corresponding operation anchor record by combining the target docking position, the position of the object to be handled, and the operation posture constraints. The executing entity is the on-site control side corresponding to the current operation session of the composite robot, the position of action is limited to the target operation unit corresponding to the event response constraint package, and the calling time period is limited to the continuous sorting interval from the time the event response constraint package takes effect to the time before the next step is registered for calling. The calling fields include at least the target operation unit identifier, the order record of the object to be handled, the risk boundary record, the docking posture constraint record, the operation posture constraint record, the rule version identifier, and the evidence trace index. No new action instruction sources outside the current step are introduced.

[0076] This step first filters the target docking positions for the composite robot. The target docking positions are limited to candidate docking positions within the current target work unit that meet the docking posture constraints and can uniquely generate the operation anchor record corresponding to the current disposal object. The current disposal object is fixedly taken from the first valid object in the disposal object sequence record in the event response constraint package, and the object position is not changed in this step.

[0077] During the screening process, the parking position range, chassis orientation range, approach direction requirements, and evacuation direction retention requirements are first retrieved from the parking posture constraint record. Then, they are compared item by item with the accessible area, restricted access area, and prohibited stay area in the risk boundary record. Only candidate positions that simultaneously meet the following requirements are retained: located within the accessible area, chassis orientation within the allowable angle range, approach direction consistent with the forward approach requirement of the current object being handled, and evacuation direction with a continuous unobstructed retention distance. The candidate parking position set here is fixed to the discrete parking position set pre-calibrated by the target work unit. Each candidate position corresponds to a unique position number and position coordinate range. No new ground positions are generated temporarily in this step.

[0078] When there are more than two candidate positions, the target parking position is selected according to the following fixed order: first, it is sorted in ascending order by the horizontal distance between the center of the target parking position and the center of the object to be dealt with; then, it is sorted in ascending order by the absolute value of the angle between the chassis orientation and the preferred approach direction in the operating posture constraint; then, it is sorted in ascending order by the chassis orientation adjustment range; if they are still the same, they are sorted in order by the preset number of the candidate positions in the target work unit. Finally, only the first candidate position is retained as the target parking position. If all candidate positions cannot simultaneously meet the parking posture constraint and risk boundary record, the selection will stop and the current target work unit will be transferred to the pending verification. The target parking position will not be estimated based on manual experience, and no other position will be selected outside the current target work unit.

[0079] After the target docking position is selected, the corresponding operation anchor record is generated by combining the target docking position, the position of the object to be disposed of, and the operation posture constraints. The position of the object to be disposed of is fixedly taken from the current position record of the object to be disposed of. The current position record of the object to be disposed of reflects the actual position of the object to be disposed of in the current state and includes at least the object position number, installation orientation, center height, and accessible surface identification. When the current state causes the object's contact surface to change relative to the initial installation position, the position record corresponding to the current state shall prevail and shall not be replaced by the initial installation position.

[0080] The operation anchor record is the on-site positioning record of the robotic arm end effector establishing a subsequent approach relationship with the target docking position, with the chassis facing downwards and towards the current object being handled. It is not the same as the motion trajectory, nor does it provide subsequent contact actions in advance. Among them, the approach reference position is recorded according to the reference position range of the end effector relative to the center of the object being handled under the target docking position, and the posture reference position is recorded jointly according to the end effector approach direction, the target height range, and the wrist allowable posture range.

[0081] During generation, the target docking position is first used as the chassis fixed reference position. Then, the installation front, center height and accessible surface position of the current object are determined according to the position of the object to be disposed of. Subsequently, the approach direction range, height range, wrist posture range, entry depth range or yaw allowable range corresponding to the current object structure type are retrieved from the operation posture constraint record and compared in turn: when the chassis orientation corresponding to the target docking position can make the end approach direction fall into the allowable range, the end target height fall into the allowable height range, and the wrist posture does not enter the restriction range corresponding to the prohibited contact surface, the operation anchor record corresponding to the object to be disposed of is generated.

[0082] Regardless of the structural type of the object being handled, the operation anchor record must include at least the target docking position number, the object's position number, and the approach direction information of the end relative to the object. The remaining fields are supplemented according to the structural type. For rotating parts, at least the target docking position number, the object's position number, the axial approach direction, the grip center height, and the wrist's permissible posture range must be written. For pressing parts, at least the target docking position number, the object's position number, the forward approach direction, the pressing axis direction, and the upper limit of the end's entry depth must be written. For plug-in parts, at least the target docking position number, the object's position number, the plug-in axis direction, the alignment height, and the permissible sway range must be written.

[0083] The structural type of the object being processed is fixed and taken from the object's installation structure record, and is not temporarily presumed in this step; when the object's installation structure record is incomplete or missing, or when any of its necessary fields such as installation orientation, structural type, accessible surface, or travel boundary are missing and cannot be traced back by evidence traceability, the corresponding object's operation anchor record will not be generated and will be transferred to pending verification.

[0084] The target height range and the upper limit of the entry depth are recorded in length units, while the allowable wrist posture range, the end-effector approach direction, and the allowable sway range are recorded in angle ranges. All of these maintain the same field measurement standards corresponding to the current version of the rules. The operation anchor record must not cause the end-effector reference position to fall into the prohibited area in the risk boundary record, and must not exceed the allowable range corresponding to the operation posture constraint record. Moreover, the operation anchor record is only valid when the chassis orientation corresponding to the target docking position is down. When the target docking position changes, the original operation anchor record will automatically become invalid and will not be reused.

[0085] To ensure the uniqueness of records, the current step generates only one currently valid operation anchor record for the same target docking position, the same disposal object location, and the same rule version identifier. When records are generated repeatedly for the same target docking position, the same disposal object location, the same rule version identifier, and the same operation posture constraint content, the same operation anchor record number is returned, and a new number is not reassigned. An overriding version is only allowed to be generated when the target docking position changes, the disposal object location changes, or the operation posture constraint version changes, and the original record has not yet been registered and called by the next step.

[0086] The operation anchor record must include at least the target work unit identifier, event response constraint package number, target docking position number, object location number, structure type identifier, end-effector approach direction, target height range, wrist permissible posture range, upper limit of entry depth or permissible sway range, rule version identifier, generation time, and evidence trace index. The evidence trace index must at least point to the event response constraint package, object installation structure record, target work unit ground docking position record, and this generation log that participated in the current generation. After generation, it is written to the local operation record area of ​​the composite robot and simultaneously written to the field control side record library for direct retrieval in the next step. No end-effector action command is issued in this step. This step only generates the target docking position and operation anchor record, and does not generate the robotic arm joint sequence, end-effector trajectory sequence, or object contact command.

[0087] When an abnormal position of the handcart occurs at a control panel on site, accompanied by an adjacent alarm source, a target parking position slightly to the left can be selected within the allowable area directly in front of the control panel based on the event response constraints. Then, by combining the position of the object to be handled corresponding to the handcart's operating port and the operating posture constraints corresponding to the pressing parts and plug-in parts, an operation anchor record containing the forward approach direction, alignment height, and upper limit of the entry depth can be generated. This allows the next step to align the current object with the same target parking position without having to reselect the position.

[0088] Preferably, the horizontal distance between the target docking position and the location of the object to be disposed of can be set to 0.4m to 1.5m, the allowable range of the angle between the chassis orientation and the installation front of the object to be disposed of can be set to ±10° to ±45°, the target height range can be recorded as 0.03m to 0.25m above and below the center height of the object to be disposed of, and the upper limit of the end-entry depth can be written as 0.3 to 1.0 times the accessible depth in the object installation structure record; in a preferred embodiment, on one valve group operating position, the event response constraint package corresponds to two candidate docking positions. After sorting by proximity distance, the degree of consistency of the preferred proximity direction, and the orientation adjustment range, one target docking position is selected, and then combined with the valve handle center position and the corresponding operating posture of the rotating part. One operation anchor record is generated by the constraints. The record contains one target docking position number, one object location number, one axial approach direction, one grip center height range, and one set of wrist allowable posture ranges. The overall write latency is controlled within 200ms to 500ms. Without changing the technical concept of this step, the candidate set of target docking positions can be obtained from fixed ground docking position records or from the set of discrete points of the docking area pre-marked by the target work unit. As long as the target docking positions of the composite robot are still screened according to the event response constraint package, and the corresponding operation anchor record is generated by combining the target docking position, object location, and operation posture constraints, it is an implementable form of this step.

[0089] S4. Control the composite robot to reach the target docking position, drive the robotic arm to sequentially perform observation and confirmation, tentative contact, and restricted actions on the object to be handled, and form a corresponding object operation record. The specific implementation is as follows:

[0090] After the previous step has established a record of the current valid target docking position and operation anchor position, the on-site control side controls the composite robot to reach the target docking position and drives the robotic arm to perform observation confirmation, trial contact, and restricted actions around the current object to be handled in sequence, thereby forming the corresponding object operation record. The executing entity is the on-site control side corresponding to the current operation session of the composite robot, the position of action is limited to the target operation unit where the target docking position is located, and the source of the call is limited to the current valid event response constraint package, the current valid target docking position, the current valid operation anchor position record, the current object position record, and the action range constraint record.

[0091] First, control the composite robot to reach the target docking position. The on-site control side retrieves the target docking position number, position coordinate range, target chassis orientation, and withdrawal direction retention requirements. Drive the composite robot chassis into the corresponding position coordinate range and adjust the real-time chassis orientation to within the allowable angle range of the target orientation. Arrival confirmation is obtained by comparing the current position record and real-time orientation record of the composite robot chassis with the target docking position record item by item. When the current position falls into the corresponding position coordinate range of the target docking position and the real-time orientation falls into the allowable angle range, it is determined that the target docking position has been reached. If the number of position adjustments reaches the upper limit set by the current rule version or the number of orientation adjustments reaches the upper limit set by the current rule version, and the current position still does not fall into the corresponding position coordinate range of the target docking position or the real-time orientation still does not fall into the allowable angle range, continue adjusting and enter the pending verification stage. Before the chassis completes the target docking position confirmation, the robotic arm shall not enter for observation and confirmation.

[0092] After arrival confirmation, the robotic arm is driven to observe and confirm the current object being handled. This observation and confirmation is performed with the chassis facing downwards from the target docking position. The on-site control side prioritizes retrieving the on-site status readback record corresponding to the current object being handled. If the on-site status readback record is missing, the visual status record is retrieved. Simultaneously, the object's position number, end-effector approach direction, target height range, and wrist permissible posture range are retrieved from the operation anchor record. The robotic arm's end-effector is then controlled to enter the pre-approach position range corresponding to the approach reference position without contacting the object being handled. Within this pre-approach position range, the installation front of the current object is observed... The contactable surface, current state, and movement direction corresponding to the movement range constraints are checked item by item. When the current object position record is inconsistent with the object installation structure record, the current object position record shall prevail. The object installation structure record shall only be used as the basis for interpreting the structure type, contactable surface, and travel boundary. When the current object position, the contactable surface direction corresponding to the current object position record, and the approach direction are consistent with the operation anchor record, and the end target height falls within the target height range and the wrist posture falls within the allowable range, the observation confirmation is deemed to have passed. Otherwise, the observation confirmation is deemed to have failed and the attempt to make contact is stopped.

[0093] After observation and confirmation, the robotic arm is driven to make trial contact with the object being handled. Trial contact refers to the control of the end effector to establish initial contact with the accessible surface of the object along the approach direction corresponding to the operation anchor record, provided that the single movement amplitude does not exceed the upper limit of the single movement amplitude constraint. The length of the trial contact displacement or the angle of the trial contact must not exceed the upper limit of the single movement amplitude. During the advancement process, the end effector contact status record, joint load readback record, and local state change record of the object being handled are continuously collected. The local state change record of the object being handled is taken from the current state readback record or visual state record corresponding to the object being handled. The local state change record of the object being handled includes at least the collection time, the corresponding object identification, the local state change position, and the local state change result. When the end effector continuously maintains contact within the allowable contact surface... If the contact state reaches the minimum holding time set in the current rule version, and within this holding time the contact position does not leave the allowed contact surface, the contact direction does not change in the opposite direction, and there are no slippage, jamming, reverse force, or contact position deviation records, the trial contact is deemed successful. Among them, the slippage record is limited to the state record where the end contact position leaves the corresponding position range of the allowed contact surface; the jamming record is limited to the state record where the state of the object being handled remains unchanged for the holding time set in the current rule version while the movement direction remains unchanged; the reverse force record is formed based on any valid record from the contact state record of the robotic arm end and the joint load readback record. If the above conditions are not met, the trial contact is deemed unsuccessful, and the restricted action is not initiated.

[0094] After successful initial contact, the robotic arm is driven to perform restricted actions on the object being handled. These restricted actions are fixed within the range of motion amplitude constraints. The on-site control side retrieves the motion direction, the upper limit of a single motion amplitude, and the cumulative motion amplitude boundary. The robotic arm's end effector is then controlled to perform rotation, pressing, insertion, extraction, or locking actions on the current object being handled according to the motion direction. Throughout the entire process, the upper limit of a single motion amplitude and the cumulative motion amplitude boundary in the motion amplitude constraints must not be exceeded, and the motion direction must not be changed during execution. A restricted action is considered complete only if the current object being handled remains within the endpoint of the target direction or the allowable endpoint range at the end of the action. If the endpoint range was reached during the action but the object has retreated at the end, the restricted action is considered incomplete.

[0095] After observation and confirmation, probing contact, and execution of restricted actions are completed, the on-site control side generates a corresponding object operation record. This record is fixed and corresponds to the current target work unit, current target docking position, current operation anchor record, and current object being handled. It does not merge records across objects and must include at least the target work unit identifier, event response constraint package number, target docking position number, operation anchor record number, current object being handled identifier, object being handled location number, observation and confirmation result, observation and confirmation start and end times, probing contact result, probing contact start and end times, restricted action direction, single restricted action amplitude, and cumulative restricted action amplitude. The system includes the start and end times of restricted actions, the object operation result identifier, the rule version identifier, and the evidence trace index. The object operation result identifier is generated according to a fixed priority of pending verification, incomplete, and completed. When a key record is missing and cannot be traced back, it is prioritized as pending verification. The key record includes at least the target docking position record, the operation anchor position record, the current position of the object being handled, and the action range constraint record. Provided that the key record is complete, it is written as incomplete if any one of the following is true: observation confirmation failed, trial contact failed, or restricted action not completed. It is written as completed only if observation confirmation passed, trial contact passed, and restricted action completed.

[0096] To ensure record uniqueness, only one currently valid object operation record is generated under the same target docking position, the same operation anchor position record, the same object location number, and the same rule version identifier. If the observation confirmation result, the trial contact result, the restricted action direction, the restricted action amplitude, and the object operation result identifier remain unchanged after repeated execution, the same object operation record number will be returned. Only when any one of the restricted action direction, restricted action amplitude, or object operation result identifier changes and the original record has not been registered and called by the next stage, is an overriding version allowed to be generated.

[0097] After the object operation record is generated, it is written to the local operation record area of ​​the composite robot and simultaneously written to the record library on the field control side for direct retrieval in the next stage.

[0098] Preferably, the observation and confirmation interval can be set to 1s to 10s, and the trial contact displacement length or trial contact angle can be set to 0.1 to 0.3 times the upper limit of the single action amplitude, with the cumulative amplitude of the restricted action not exceeding the cumulative action amplitude boundary. In a preferred embodiment, at one valve group operating position, after the composite robot reaches the target docking position, it completes the observation and confirmation within 3s, and then performs a trial contact at 0.2 times the remaining target angle. After the trial contact is passed, the shut-off rotation is completed according to the action amplitude constraint, and an object operation record is formed, which includes "observation and confirmation passed," "trial contact passed," and "restricted." The direction of the action is the off direction, the amplitude of a single action is 0.8 times the remaining allowable angle, the object operation result is marked as completed, and the overall writing delay is controlled within 300ms to 800ms. Without changing the technical concept of this step, observation and confirmation can be completed either by on-site status rereading or by visual status recording when on-site status rereading is missing. As long as the composite robot is still controlled to reach the target docking position, the robotic arm is driven to perform observation and confirmation, tentative contact, and restricted actions on the object to be handled in sequence, and corresponding object operation records are formed, all of which are feasible forms of this step.

[0099] S5. Retrieve the current status feedback based on the object operation record, execute actions according to the result verification constraints to complete the verification and risk convergence verification, and generate an event response judgment record. The specific implementation is as follows:

[0100] After the composite robot completes the actual contact action of the current object based on the operation anchor record, the on-site control side does not directly proceed to the next object. Instead, it first retrieves the current status feedback based on the object operation record, and then executes the action completion verification and risk convergence verification in sequence according to the result verification constraints, and generates an event response judgment record under the same rule version. The executing entity of this step is the on-site control side corresponding to the current operation session of the composite robot. The position of action is limited to the target operation unit corresponding to the object operation record. The time period of action is limited to the continuous confirmation interval from the completion of the object operation record writing to the reading of the judgment result in the next stage. The source of the call is limited to the currently valid object operation record, the result verification constraints corresponding to the object operation record, the status feedback record of the current target operation unit, and the release judgment caliber corresponding to the current rule version. No new control command sources outside the current step are introduced.

[0101] First, retrieve the current status feedback based on the object operation record. When multiple object operation records exist for the same currently handled object, only the most recently written valid object operation record is taken as the sole upstream record for this step. Records marked as pending verification are not used as the basis for action completion verification. The current status feedback is fixedly taken from the feedback set formed within the confirmation period after the object operation record is written. The confirmation period is fixed with the end time of the object operation record as the starting point and the end time calculated from the corresponding confirmation period length in the result verification constraint as the ending point. The object status feedback prioritizes the device's built-in status collection record corresponding to the handled object; if the device's built-in status collection record is missing, then it is taken from... For the corresponding on-site status readback records, abnormal signal feedback will prioritize on-site abnormal feedback records of the same source category as the current multi-source event convergence record; the same source category is fixed into four categories: gas abnormal signals, temperature abnormal signals, equipment status abnormal signals, and electrical alarm signals; during retrieval, the writing time of the object operation record and the collection time of each feedback record are aligned according to a unified time synchronization benchmark, and only feedback records within the confirmation period are retained. Historical records outside the confirmation period are not retrieved, and the old state before the object operation is not used to replace the current state feedback; both object status feedback and abnormal signal feedback will maintain the original measurement caliber and original state caliber of their source records for comparison, and cross-caliber conversion will not be performed in this step before judgment.

[0102] If multiple similar feedback records are obtained within the same confirmation period, they are sorted in ascending order of collection time, and the last valid feedback is taken as the current feedback value. When the last valid feedback within the confirmation period conflicts with the previous valid feedback of the same type in terms of status value or de-status, the two most recent consecutive feedbacks in the same direction at the end of the confirmation period are taken as the current feedback value. If there are no two consecutive feedbacks in the same direction within the confirmation period, the verification is stopped and the process is transferred to pending verification. The key feedback conflict here is limited to two valid feedback records corresponding to the same source category, the same disposal object, or the same abnormal signal within the same confirmation period that give mutually exclusive status conclusions.

[0103] If no valid feedback is obtained within the confirmation period, or if the current status feedback exists but its collection time does not fall within the same confirmation period as the end time of the object operation record, or if the source object cannot be matched one-to-one with the current object being processed, then the verification will be stopped and the case will be transferred to pending verification, without supplementing with manual estimates.

[0104] After retrieving the current status feedback, the action completion verification is first performed according to the result verification constraints. The action completion verification is fixed in the corresponding object status change verification item in the result verification constraints, which at least revolves around the target status value, allowable deviation range, and confirmation period of the current object being handled. The on-site control side first retrieves the target status value and allowable deviation range that correspond one-to-one with the current object being handled from the result verification constraints, and then compares the object status feedback in the current status feedback with the target status value item by item. Only when the object status feedback remains within the allowable deviation range corresponding to the target status value before the end of the confirmation period is the action completion verification deemed to have passed. When the object status feedback exceeds the allowable deviation range, the feedback direction is opposite to the operation direction in the object operation record, the object status remains in the same state as before the operation, the object status enters a non-target intermediate state and does not return to the target status range within the confirmation period, or the object status has entered the target status range and then jumped out, the action completion verification is deemed to have failed.

[0105] For example, if the target state of the valve is the shut-off position, the valve position feedback is retrieved during the confirmation period. When the valve position feedback continuously enters the allowable range of the shut-off position, the action is considered to have been completed and the verification is passed. If the valve position remains in the half-open position, exceeds the corresponding termination position of the shut-off position, or briefly enters the shut-off position and then returns to the half-open position, the action is considered to have failed the verification.

[0106] After the action verification is completed, risk convergence verification is performed according to the result verification constraints. The risk convergence verification fixes the abnormal signal fallback verification item in the corresponding result verification constraints. The release judgment condition is fixed based on the abnormal signal fallback verification item already written in the current result verification constraints. The original rule configuration is not reread or rewritten in this step. The field control side compares the abnormal signal feedback in the current status feedback with the release judgment condition item by item. When each type of corresponding abnormal signal in the current multi-source event convergence record meets the release judgment condition within the confirmation period, the risk convergence verification is deemed to have passed. When there are multiple feedback sources for the same type of abnormal signal in the current target work unit, the fallback verification of the abnormal signal is deemed to have passed only when all valid feedback sources corresponding to the abnormal signal of that type meet the release judgment condition. When any type of corresponding abnormal signal remains in an abnormal state, although it has fallen back, it has not met the release judgment condition, new abnormal records are added during the confirmation period, or the original single abnormality is expanded into a new abnormal type, the risk convergence verification is deemed to have failed.

[0107] For example, after the filling interface is processed, if the interface locking state has been restored but the abnormal gas signal around the interface continues to be added during the confirmation period, the action completion verification can pass but the risk convergence verification will fail; after the panel is processed, if the switch position has reached the target position but the relevant alarm source equipment still maintains the current alarm state, the risk convergence verification will fail.

[0108] After both the action completion verification and risk convergence verification are completed, the on-site control side generates an event response judgment record. This record is fixed and corresponds to the current target work unit, the current object being handled, and the current object's operation record. It is not merged across objects, and each event response judgment record corresponds to only one object operation record. Verification results for multiple object operation records are not written side-by-side in the same record. It must at least include the target work unit identifier, object operation record number, result verification constraint number, action completion verification result, risk convergence verification result, judgment conclusion, confirmation period, rule version identifier, generation time, and evidence trace index. The judgment conclusion is written as "verification passed" or "action incomplete" according to a fixed standard. If a result is selected from "Completed," "Risk Not Converged," or "Pending Verification," only one main conclusion is allowed in the judgment conclusion field. Additional main conclusions cannot be written in parallel. Other verification results that are not selected as main conclusions are only retained in the action completion verification result field or the risk convergence verification result field. When both action completion verification and risk convergence verification pass, the judgment conclusion is written as "Verification Passed." When both action completion verification and risk convergence verification fail, the judgment conclusion is written as "Action Not Completed." When action completion verification passes but risk convergence verification fails, the judgment conclusion is written as "Risk Not Converged." When feedback is missing, key feedback conflicts occur, the judgment conditions for lifting cannot be compared, or the source cannot be matched, the judgment conclusion is written as "Pending Verification."

[0109] To ensure record uniqueness, only one currently valid event response judgment record is generated under the same object operation record, the same result verification constraint number, and the same rule version identifier. When the verification is performed repeatedly and the content remains unchanged, the same event response judgment record number is returned, and a new number is not reassigned. A covered version is only allowed to be generated when a new valid feedback is added within the confirmation period and the original judgment record has not yet been registered and called by the next stage. A covered version is only generated when the new valid feedback changes the action completion verification result, risk convergence verification result, or judgment conclusion. If a new valid feedback is added within the confirmation period but does not change the action completion verification result, risk convergence verification result, or judgment conclusion, a covered version is not generated, and only the evidence trace index is updated.

[0110] After the event response determination record is generated, it is written to the local operation record area of ​​the composite robot and simultaneously written to the field control side record library for direct retrieval in the next stage. The event response constraint package is not directly rewritten in this step.

[0111] Preferably, the confirmation period can be set to 5s to 120s. In a preferred embodiment, after the valve shut-off operation is completed in one valve group operation position, the field control side retrieves one valve position feedback, two temperature feedbacks, and one alarm feedback within a 30s confirmation period. When the valve position continuously enters the allowable range of the shut-off position, the temperature feedback stops rising, and the alarm feedback is cleared, an event response judgment record is generated. The record contains the action completion verification passed, the risk convergence verification passed, and the verification pass conclusion. The overall judgment delay is controlled within 300ms to 800ms. Without changing the technical concept of this step, the current status feedback can be obtained by a fixed status acquisition device or by the field readback record corresponding to the current object being handled. As long as the current status feedback is retrieved based on the object operation record, and the action completion verification, risk convergence verification, and event response judgment record are executed according to the result verification constraints, they all belong to the implementable form of this step.

[0112] S6. When the judgment conclusion of the event response judgment record is any one of the following: action not completed, risk not converged, or pending verification, adjust the target docking position, operation anchor position, and disposal object sequence according to the event response constraint package, and execute the next round of response control. The specific implementation is as follows:

[0113] When the event response determination record concludes that the action is incomplete, the risk has not converged, or the action is pending verification, the current operation session of the composite robot does not regenerate the multi-source event convergence record and the event response constraint package. Instead, under the same target operation unit, the same rule version, and the same event response constraint package number, the target docking position, operation anchor position, and the order of the objects to be handled are readjusted according to the event response constraint package, and the next round of response control is executed accordingly. The executing entity is the on-site control side corresponding to the current operation session of the composite robot, and the effective period is limited to the continuous processing interval from the generation of the event response determination record to the registration and effective period of the next round of response control. The sources of invocation are limited to the currently valid event response constraint package, the currently valid event response determination record, the object operation record corresponding to the current object to be handled, and the current status feedback corresponding to the failure conclusion. No new constraint sources outside the current step are introduced, and readjustment is only performed according to the event response constraint package when the current event response constraint package is still the current valid version and the rule version identifier has not changed. When the event response constraint package fails or the rule version identifier changes, the readjustment link in this step is not entered.

[0114] First, the target docking position is readjusted according to the event response constraint package. When both the action completion verification and risk convergence verification fail in the event response judgment record, the main conclusion recorded in the judgment conclusion field is used as the sole basis for this step of readjustment. When the judgment conclusion is that the action is not completed, one or more of the target docking position, operation anchor position, and disposal object order can be adjusted. The on-site control side will only exclude the target docking position used in the previous round if the reason for failure can be attributed to insufficient approach side, orientation, or evacuation direction related to the target docking position. Then, from the candidate docking position set corresponding to the original event response constraint package, candidate positions that are still located in the accessible area, meet the docking posture constraints, and are consistent with the forward approach requirements of the current disposal object are retained. The next target docking position is re-selected according to the established approach distance, the degree of consistency of the preferred approach direction, and the chassis orientation adjustment range in the original event response constraint package. When the reason for failure only corresponds to the improper selection of operation anchor position and not to the target docking position itself, the target docking position of the previous round is retained and only the operation anchor position is readjusted.

[0115] When the judgment conclusion is that the risk has not converged, at least one item in the order of operation anchor or disposal object should be adjusted; if the target docking position in the previous round is still within the allowable range but its corresponding proximity side is adjacent to the source of the unresolved anomaly, the docking position corresponding to the proximity side should be excluded first, and then the remaining candidate positions should be re-selected according to the same sorting criteria; when the judgment conclusion is pending verification, adjustments are only allowed if the necessary records corresponding to the target docking position, disposal object position and operation posture constraint are complete and can form the current valid operation anchoring position record. If any one of the necessary records is missing or cannot be traced back, adjustments should be stopped and the pending verification conclusion should be maintained. The target docking position should not be supplemented by manual experience.

[0116] After the target docking position is readjusted, the operation anchor position is readjusted again according to the event response constraint package. The operation anchor position is still generated within the operation posture constraint range corresponding to the current event response constraint package, without relaxing the original approach direction range, target height range, wrist allowable posture range, entry depth upper limit, or yaw allowable range. The field control side first matches the approach reference position and posture reference position corresponding to the previous round of operation anchor position records with the reasons for failure item by item. When the direct reason for the failure of the action is manifested as end-point approach direction deviation, target height mismatch, wrist posture entering the restricted area of ​​the prohibited contact surface, or insufficient entry depth, a new operation anchor position is selected within the original operation posture constraint record allowable range based on the readjusted target docking position. When readjusting the approach and attitude reference positions of the pre-processed object, the target height range from the previous round should be kept unchanged first. Then, the approach direction of the end point should be adjusted first, followed by the wrist's permissible attitude range, and finally the upper limit of the entry depth or the permissible sway range. The adjustment should only proceed to the next layer if the previous layer cannot form a record of the current valid operation anchor position. When the risk has not converged but the action completion verification has passed, the approach side and entry depth combination that caused the abnormal signal to remain in the previous round should not be used. Instead, without breaking the original operation attitude constraints, the approach reference position on the opposite side of the source of the unresolved abnormality or the smaller upper limit of the entry depth should be selected first. When the target docking position changes, the previous operation anchor position automatically becomes invalid and must be readjusted.

[0117] Subsequently, the order of the disposal objects is readjusted according to the event response constraint package. The order of disposal objects is only adjusted within the set of disposal objects recorded in the current event response constraint package, and no new objects outside the set are added. When the judgment conclusion is that the action is not completed, if the current disposal object still has a substitute target docking position or substitute operation anchor position that can form a current valid record within the constraint range of the current event response constraint package, the current disposal object is kept in the first position, and only its target docking position and operation anchor position are adjusted before it enters the next round of response control first. When the current disposal object has no substitute target docking position or operation anchor position, it is only allowed to move one position to the right, and it will be directly associated with the unresolved abnormal signal. Furthermore, the next object with the current valid target docking position or the current valid operation anchor position is moved to the first position, while the remaining objects maintain their original relative order. When the judgment conclusion is that the risk has not converged, the object directly associated with the unresolved abnormal signal is moved to the first position first. Objects that have completed actions in the previous round but failed to eliminate the risk are moved one position back or remain in their original positions, but the original object dependency relationship is not disrupted after the move. When the order of the objects changes, the target docking position and the operation anchor position are readjusted according to the new first object. Only when the order of the target docking position and the order of the objects remains unchanged is it allowed to readjust the operation anchor position separately.

[0118] After completing the above three readjustments, the next round of response control is executed. Here, the next round of response control refers to generating a new response invocation basis based on the readjusted target docking position, operation anchor position, and disposal object order, and submitting this basis to the subsequent object response execution steps. This step only completes the adjustment and registration, and does not complete new object contact actions within this step. The field control side generates a next round of response control registration record based on this, which includes at least the target work unit identifier, event response constraint package number, previous round event response judgment record number, readjusted target docking position number, readjusted operation anchor position record number, readjusted disposal object order record, rule version identifier, registration time, and evidence trace index. Each next round of response control registration record corresponds to only one readjustment result. Multiple target docking positions, multiple operation anchor positions, or multiple first disposal objects are not written in parallel in the same registration record. The evidence trace index points at least to the event response judgment record, the event response constraint package involved in this readjustment, the excluded or retained target docking position record, the newly generated operation anchor position record, and this adjustment log.

[0119] To ensure record uniqueness, only one currently valid next-round response control registration record is allowed to be generated under the same event response judgment record, the same event response constraint package number, and the same rule version. If adjustments are repeatedly executed and the target docking position, operation anchor position, and disposal object order remain unchanged, no new next-round response control registration record will be generated; only the currently valid registration record number will be returned. An overriding version is only allowed to be generated when any one of the target docking position, operation anchor position, or disposal object order changes and the original registration record has not yet been invoked by the next-round response control registration.

[0120] When the number of times the target dock position is readjusted reaches the upper limit set in the current rule version and the event response judgment record still fails, the adjustment will stop. When the number of times the operation anchor position is reselected under the same target dock position reaches the upper limit and a current valid operation anchor position record still cannot be formed, the adjustment will also stop and the failure conclusion corresponding to the current event response judgment record will be maintained.

[0121] On-site, after the control panel completes one tripping operation, if the event response judgment record indicates that the risk has not converged, and the current status feedback shows that the tripping status has been achieved but the relevant alarm has not been cleared, the original event response constraint package is retained first, the previous round target docking position adjacent to the alarm source is excluded, and a new target docking position is selected from the remaining candidate positions. The operation anchor position that avoids the alarm side is reselected within the original operation posture constraint allowable range. At the same time, the handling object directly corresponding to the alarm source is moved to the first position, thus forming the next round of response control.

[0122] Preferably, the readjustment of the next round of response control can be initiated within 1 second after the event response judgment record is generated. The number of times the target docking position is adjusted can be set to 1 to 3 times, and the number of times the operation anchor is reselected under the same target docking position can be set to 1 to 3 times. In a preferred embodiment, on a valve group operation position, if the first round of event response judgment record shows that the action is not completed, after the field control side excludes the target docking positions used in the first round, a new target docking position is selected from the remaining 2 candidate positions. A new operation anchor record is formed within the same operation posture constraint range, and the current valve handling object is kept in the first position. A next round of response control registration record is generated, and the overall adjustment delay is controlled within 300ms to 900ms. Without changing the technical concept of this step, the readjustment of the target docking position can be implemented based on a discrete docking position set or based on a pre-numbered equivalent candidate point set within the docking area. As long as the order of the target docking position, operation anchor, and handling object is readjusted according to the event response constraint package, and the next round of response control is executed, it belongs to the implementable form of this step.

[0123] In the operating scenario shown in this embodiment: one valve group operation position in the chemical plant area is pre-coded as one target operation unit. This target operation unit corresponds to one feed branch isolation valve, one set of fixed combustible gas probes, one valve body surface temperature measurement point, one valve position status acquisition point, and one nearby control box alarm acquisition point. When the composite robot travels along the preset inspection path to the vicinity of this area, the on-site control side receives the abnormal acquisition push of the gas monitoring point associated with the target operation unit. The current operation session of the composite robot does not directly enter the disposal arrangement, but first starts the multi-source event convergence process with the target operation unit as the object.

[0124] Within a preset time window of 60 seconds, the fixed gas probe continuously reports four gas abnormality signals, the valve body surface temperature measurement point generates two temperature abnormality signals, the valve position status acquisition point generates one equipment status abnormality signal indicating that the valve position is not in place, and the adjacent control box generates one position contact abnormality alarm signal. The field control side converts the above records to a unified time reference, organizes them in the order of alignment, noise reduction, and completion, and retains only the valid records that can uniquely correspond to the target work unit. Finally, under the same rule version, a multi-source event convergence record is generated and written to the local operation record area of ​​the composite robot and the field control side record library.

[0125] Subsequently, the on-site control side extracts risk boundaries, order of disposal objects, docking posture constraints, operating posture constraints, action range constraints, and result verification constraints based on the convergence record of the multi-source event. Among these, the valve body is determined as the first disposal object in the order of disposal objects. The temperature rise area around the valve body and the area near the abnormal measuring point are formed into a restricted access zone. The ground dockable area on the opposite side of the valve handwheel is determined as the allowable docking position range. The axial approach direction of the robotic arm end to the valve, the grip center height range, and the wrist allowable posture range are written into the operating posture constraints. The upper limit of the single action range and the cumulative action range boundary are formed based on the difference between the current valve position and the target shut-off position. At the same time, the result verification constraints of the valve position returning to the shut-off position and the relevant alarm being cleared are written into the constraint, thereby generating one event response constraint package.

[0126] Subsequently, based on the event response constraint package, the on-site control side eliminates one candidate position that falls into the edge of the restricted access zone from the three candidate docking positions pre-marked for the target work unit. Then, it selects one target docking position in the order of horizontal distance between the center of the target docking position and the center of the object to be dealt with, consistency of the preferred approach direction, and chassis orientation adjustment range. Combined with the current object position record and operation posture constraints, an operation anchor record is generated. The operation anchor record specifies the target docking position number, valve position number, axial approach direction, grip center height range, and wrist allowable posture range.

[0127] The composite robot then arrived at the target docking position under the drive of the on-site control side. After the chassis position and orientation were confirmed to be in place through item-by-item comparison, the robotic arm first performed observation and confirmation in the pre-approach position range. After verifying that the valve installation front, contact surface, current valve position status, and the closing direction corresponding to the movement amplitude constraint were consistent with the operation anchor position record, it entered the trial contact stage. The end effector established the first contact relationship along the axial approach direction. Stable contact was formed under the condition that it did not exceed the upper limit of the single movement amplitude, and no slippage, jamming, or reverse force state was recorded. Therefore, the trial contact was judged to be successful.

[0128] Subsequently, the robotic arm performs a restricted action, driving the valve handwheel to rotate along the shut-off direction within the cumulative action range boundary. At the end of the action, the valve position remains within the allowable shut-off range, thus forming one object operation record and writing it into the record library.

[0129] After the operation record of the object is formed, the field control side retrieves the current status feedback within a 30-second confirmation period, performs unified timing alignment on the valve position status feedback, valve body surface temperature feedback, and control box alarm feedback, and first performs action completion verification to confirm that the valve position is continuously maintained within the allowable range of the closed position. Then, it performs risk convergence verification to confirm that the corresponding gas abnormality signal is no longer added, the temperature feedback stops rising, and the alarm feedback is cleared. This generates an event response judgment record with a judgment conclusion of verification passed. This record indicates that the current round of response control has been completed, and the current operation session of the composite robot will not enter the next round of response control, but will switch to the subsequent inspection standby state.

[0130] If, in the same scenario, the valve position has returned to the shut-off position but the control box alarm has not been cleared within the confirmation period, the event response judgment record is written as risk not converged. The field control side retains the original event response constraint package, excludes the previous round target docking position adjacent to the alarm source, re-selects new target docking positions from the remaining candidate positions, and re-forms an operation anchor position record that avoids the alarm side within the original operation posture constraint range. At the same time, the valve object directly associated with the uncleared abnormal signal is kept at the first position in the order of disposal objects. The next round of response control registration record is generated and the subsequent object response execution steps are submitted, thereby completing the entire closed-loop operation from S1 to S6.

[0131] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0132] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0133] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission; wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission includes infrared, microwave, etc. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center containing one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0134] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0135] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0136] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0137] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0138] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0139] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0140] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-source safety event response control method for a composite robot, characterized in that, include: S1. Collect gas abnormality signals, temperature abnormality signals, equipment status abnormality signals, and electrical alarm signals of the target work unit within a preset time window, merge them according to a unified time reference and work unit identifier, and generate a multi-source event convergence record. S2. Extract risk boundaries, order of disposal objects, docking posture constraints, operation posture constraints, action range constraints, and result verification constraints from the multi-source event convergence records, and generate an event response constraint package. S3. Filter the target docking position of the composite robot according to the event response constraint package, and generate the corresponding operation anchor record by combining the target docking position, the position of the object to be dealt with, and the operation posture constraint. S4. Control the composite robot to reach the target docking position, drive the robotic arm to perform observation and confirmation, trial contact and restricted actions on the object to be handled in sequence, and form a corresponding object operation record; S5. Retrieve the current status feedback based on the object operation record, execute actions to complete the verification and risk convergence verification according to the result verification constraints, and generate an event response judgment record; S6. When the judgment conclusion of the event response judgment record is any one of the following: action not completed, risk not converged, or pending verification, adjust the target docking position, operation anchor position, and disposal object sequence according to the event response constraint package, and execute the next round of response control.

2. The multi-source safety event response control method for a composite robot according to claim 1, characterized in that, S1 includes: The gas anomaly signal, temperature anomaly signal, equipment status anomaly signal, and electrical alarm signal are processed in the order of alignment, noise reduction, and completion. When at least two types of valid abnormal signals are obtained under the same target work unit, the same preset time window, and the same rule version identifier, one currently valid multi-source event convergence record is generated.

3. The multi-source safety event response control method for a composite robot according to claim 1, characterized in that, S2 include: Extract risk boundaries and the order of action targets based on the convergence records of multi-source events; The risk boundary is formed by the location of the abnormal measurement point corresponding to the gas abnormal signal, the temperature rise part corresponding to the temperature abnormal signal, the location of the abnormal object corresponding to the equipment status abnormal signal, and the alarm source location corresponding to the electrical alarm signal. The order of objects to be dealt with is determined based on the number of categories that are jointly pointed to by valid abnormal signals, the on-site distance between the object to be dealt with and the center of the risk boundary, the order of objects that can be arranged in a continuous and close manner on the current docking surface, and the order of objects in the preset number of the target work unit.

4. The multi-source safety event response control method for a composite robot according to claim 3, characterized in that, S2 also includes: Extract the operation posture constraints based on the order of the objects to be handled, the location of the objects to be handled, the docking posture constraints, the abnormal signals of the equipment status, and the object installation structure records; Extract motion amplitude constraints based on the current status value of the equipment status anomaly signal, the object installation structure record, and the on-site operation baseline record; The event response constraint package is written in the following order: risk boundary, order of objects to be dealt with, docking posture constraint, operation posture constraint, movement range constraint, and result verification constraint.

5. The multi-source safety event response control method for a composite robot according to claim 1, characterized in that, S3 includes: The parking position range, chassis orientation range, approach direction requirements, and evacuation direction retention requirements are retrieved from the parking posture constraint record and compared item by item with the accessible area, restricted access area, and prohibited stay area in the risk boundary record to screen the target parking position; Then, by combining the target docking position, the location of the object to be disposed of, and the operation posture constraints, an operation anchor record is generated. The operation anchor record includes the target docking position number, the location number of the object to be disposed of, the end approach direction, the target height range, and the wrist allowable posture range.

6. The multi-source safety event response control method for a composite robot according to claim 1, characterized in that, S4 include: After the on-site control side controls the composite robot to reach the target docking position, it drives the robotic arm to perform observation and confirmation, probing contact, and restricted actions in sequence around the object to be handled. After observation and confirmation, the process proceeds to tentative contact; after successful tentative contact, the process proceeds to restricted actions.

7. The multi-source safety event response control method for a composite robot according to claim 6, characterized in that, S4 also includes: After observation and confirmation, tentative contact, and execution of restricted actions are completed, the on-site control side generates an object operation record; The object operation record includes the target docking position number, operation anchor record number, current object identification, object location number, observation confirmation result, trial contact result, restricted action direction, and object operation result identifier.

8. A multi-source safety event response control method for a composite robot according to claim 1, characterized in that, S5 include: When there are multiple object operation records corresponding to the same current object, only the current valid object operation record with the most recent write time is taken as the only upstream record. Records with object operation results marked as pending verification are not used as the basis for action completion verification. Retrieve current status feedback during the confirmation period. Current status feedback includes object status feedback and abnormal signal feedback. First, perform the action to complete the verification based on the result verification constraint, and then perform the risk convergence verification based on the result verification constraint. The judgment conclusion is determined based on the action completion verification results and risk convergence verification results, and an event response judgment record is generated.

9. A multi-source safety event response control method for a composite robot according to claim 1, characterized in that, S6 include: When the judgment conclusion of the event response judgment record is any one of the following: action not completed, risk not converged, or pending verification, under the same target work unit, the same rule version, and the same event response constraint package number, the target docking position, operation anchor position, and order of handling objects shall be readjusted according to the current valid event response constraint package. When the target docking position changes, the original operation anchor position record becomes invalid and needs to be readjusted; Based on this, the next round of response control registration records will be generated.

Images