42 results about "Robot planning" patented technology

A robotic controller for controlling a robot according to a sequence of robotic actions. comprises an input interface configured to receive a plurality of multimodal inputs each specifying instructions for performing a task in a different modality including audio, video, and a text modality. The controller also comprises a multimodal large language model, an action sequence decoder, and a controller. The multimodal LLM includes a multimodal LLM encoder and an LLM decoder. The multimodal LLM encoder is trained with machine learning to transform the multimodal instructions into encodings and the LLM decoder is configured to decode the encodings into a sequence of robotic instructions. The action sequence decoder is trained with machine learning to transform the sequence of robotic instructions into a sequence of actions using a library of robotic skills. The controller is configured to control a robot according to the sequence of actions.

Techniques for controlling a robot include receiving sensor data and one or more goal specifications, processing the sensor data, a robot size, and the one or more goal specifications using one or more trained encoders to generate a plurality context tokens, processing the plurality of context tokens using one or more trained decoders to generate a robot plan, and controlling a robot based on the robot plan.

The invention provides a hierarchical intention-driven humanoid robot man-machine interaction system and a method thereof, relates to the field of man-machine interaction, effectively solves the defects of a deep learning method in long-time task planning, overcomes the limitation of a large-scale pre-training model in the aspect of fine-grained control, and improves the working efficiency. It is ensured that the robot can freely adapt to and stably execute tasks in a dynamic and complex environment, and the intelligent level of man-machine interaction and the robustness of the system are improved. The system extracts key object point information in a scene through a scene understanding module; predicting a future short-term attitude sequence of the mechanical arm through an attitude prediction module; a robot behavior decision module realized through a multi-modal large model completes robot high-order behavior prediction according to key object point information and a future short-term attitude sequence; and a robot planning code is generated through a robot planning code generation module according to the key object point information, the future short-term attitude sequence and robot high-order behavior prediction. The method is suitable for the man-machine interaction application field.

The invention relates to an artificial intelligence technology, can be applied to business system platforms of medical health, financial science and technology and the like, and discloses a robot control method, device and equipment based on artificial intelligence and a medium. Generating a planning path instruction of the robot according to the decomposition task; executing the path planning instruction in the simulation environment, and outputting operation track information and environment state change data of the robot; the simulation execution result is verified, and if verification succeeds, the natural language task instruction, the decomposition task, the operation track information and the success label serve as multi-modal information to be stored; training the initial diffusion strategy model based on the multi-modal information to generate a robot decision model; and controlling the robot based on the robot decision model. Through the automatic and large-scale data generation process, the robot learning efficiency and robustness are remarkably improved, and a natural language instruction can be responded to execute multiple tasks.

The invention discloses a hierarchical reinforcement learning control system based on target guidance, and belongs to the field of robot control. The system comprises a visual pose prediction module which detects and tracks the pose of a target object in the environment in real time through an improved YOLOv8 algorithm to obtain real-time environment data; the upper-layer decision-making subsystem can evaluate the task progress according to the upper-layer decision-making target and the current state of the real-time environment data and select a corresponding middle-layer operation target; the middle-layer planning subsystem can generate robot planning actions transmitted to the lower-layer control subsystem according to the current state and a middle-layer operation target; and the lower-layer control subsystem can control the actual motion track of the robot in combination with impedance control after robot motion track interpolation in combination with a depth kinematics model and a depth dynamics model corresponding to the robot so as to accurately execute operation actions. According to the system, through hierarchical reinforcement learning, all hierarchies can effectively cooperate, and the decision-making efficiency and the execution effect of the robot in complex tasks are improved.

The invention discloses a quadruped robot supporting leg counter-acting force planning control method, which belongs to the technical field of robot control, and comprises the following steps: firstly, acquiring current gait information of a quadruped robot, and identifying a current supporting leg; secondly, constructing a ground support polygon model; then setting stable support constraint conditions based on a ground support polygon model, and calculating a current expected state of the mass center of the robot; constructing a virtual model of the quadruped robot, and calculating the reactive force of the current supporting legs of the quadruped robot; and finally, the output torque of each motion joint of the quadruped robot is calculated to generate a robot planning control instruction. A stable supporting constraint condition is formed by recognizing supporting legs of the quadruped robot and establishing a ground supporting polygon model, and the current reactive force of the supporting legs is calculated based on the stable supporting constraint condition, so that a robot planning control instruction is generated, planning control over the reactive force is achieved, and the gait stability of the robot is improved.

The invention provides a warehouse scheduling method and device, a scheduling server, a storage medium and a program product. The method comprises the following steps: calculating key parameters required for allocating first robots according to the total number of warehouse-out tasks of a warehousing system, the total number of the first robots in the warehousing system, the number and state of warehouse-out tasks of each physical roadway in the warehousing system and the number of bound first robots; according to the key parameters, determining a first robot planned allocation number corresponding to each physical roadway, and realizing reasonable planning of the first robot planned allocation number corresponding to each physical roadway; the first robots bound to the physical roadways are correspondingly adjusted according to the planned allocation number of the first robots corresponding to the physical roadways, and it can be ensured that the first robots are evenly distributed in the physical roadways according to the ex-warehouse task requirements; the problem that local resources are insufficient and other physical roadway resources are idle due to the fact that the first robot is excessively concentrated in some physical roadways is solved, and the overall efficiency of the warehousing system is improved.

The invention relates to the technical field of path planning, in particular to an automobile part cutting robot operation path planning method, which comprises the following steps: determining segmentation point pairs between adjacent contours; in response to the fact that the direction mutation factor between the segmentation point pairs is larger than or equal to an abnormal threshold value, the deceleration duration between the segmentation point pairs is calculated; obtaining the total path length between the segmentation point pairs according to the product of the direction transformation angle between the segmentation point pairs and the actual turning radius; based on the total path length and the deceleration duration between the segmentation point pairs, obtaining the constant speed duration between the segmentation point pairs, and calculating a time cost factor; continuously acquiring time cost factors of segmentation point pairs between other adjacent contours; in the multiple cutting routes of the automobile part contour generated by the heuristic algorithm, the working path of the cutting robot is optimized, and the efficiency of planning the working path by the automobile part cutting robot is effectively improved.

The invention provides a marking operation calibration method for a bulkhead marking robot, which comprises the following steps: mounting a laser tracker on a ship wall to be marked, arranging the robot on the ship wall to be marked, and mounting a gyroscope on the robot; calibrating a conversion relation from a laser tracker coordinate system s to a cabin wall coordinate system b; obtaining a ZYX Euler angle change relation from a cabin wall coordinate system b to a mechanical arm coordinate system r, and calibrating a zero point of a gyroscope; measuring the position of the origin of the robot in the cabin wall coordinate system b in real time, and guiding the robot to reach a target workstation; and in combination with an original point position measured in real time and a zero point of a gyroscope, a motion control track of the robot in the working station under a mechanical arm coordinate system r is solved, and a marking operation calibration task of a single station is completed. According to the method, the conversion process of robot planning motion data from drawing stipulation data to site specific application data is combed, and accurate lineation station guide data and lineation point guide data are provided for wall surface operation of the LNG bulkhead robot.

A controller is provided for manipulating an object using a robotic arm with a gripper. The controller includes: a Large Language Model (LLM) planner configured to acquire the state and the task description, and generate a sequence of motion commands to operate the robotic arm with the gripper based on the task description, the current observations, and historical information including historical actions and historical observations from previous steps. The controller also includes: a robot interface configured to generate and send commands to operate the robotic arm in response to the sequence of motions from the LLM planner; and an LLM evaluator configured to acquire the state and the current observations, analyze the historical information including historical actions and historical observations from previous steps, evaluate missing information required to complete the task, evaluate new information acquired by the sensors, estimate the task execution state after the robotic arm operates the generated sequence of motions, and output updated information to the LLM planner.

The present invention provides a method and system for collaborative motion control of a quadruped single-arm robot, relating to the field of quadruped single-arm robot planning and control. The method comprises: establishing a multi-rigid body dynamics model and a single rigid body reduced-order model as model constraints; screening feasible landing areas; generating a safe landing area; formulating a landing point cost function and determining the optimal landing point; discretizing the state transition equation of the quadruped single-arm robot; introducing the traversable steps as an inequality constraint and the optimal landing point as an elevation map equality constraint into a model predictive controller; optimizing the multi-task planning method based on the multi-rigid body dynamics model, rationally allocating control strategies, dynamically coordinating conflicting tasks between quadruped motion and manipulator operation, and collaboratively controlling multiple tasks; and integrating target data from a depth camera at the end of the manipulator arm to autonomously plan the posture adjustment path of the robot body. The present invention significantly improves the stability and accuracy of collaborative grasping by the robot.

Techniques for controlling a robot includes receiving a task and motion planning (TAMP)-domain language input comprising one or more symbolic elements and one or more continuous elements, generating a TAMP-domain representation based on the TAMP-domain language input, TAMP-domain language data, and one or more subgraph classification rules, generating a robot plan based on the TAMP-domain representation, a state, and a goal, and causing the robot to perform at least part of the robot plan.

The present invention discloses a puncture robot planning and control method, which belongs to the field of medical robot technology, including: establishing the connecting rod coordinate system of the puncture robot by improving the D-H method; calculating the transformation matrix between the coordinate systems of the puncture robot according to the D-H coordinate parameters, and multiplying the transformation matrix between the coordinate systems to obtain the conversion matrix from the base coordinate system of the puncture robot to the puncture needle coordinate system; calculating the forward kinematics equation of the puncture robot, defining the puncture angle and puncture length, and determining the direct relationship between the position of the puncture needle tip and the displacement of each motor sensor; using the geometric analysis method to solve the inverse kinematics for the displacement of each motor sensor. The present invention adopts the above-mentioned puncture robot planning and control method, which can improve the accuracy and efficiency of the biopsy process, reduce complications and patient discomfort, reduce the fatigue of doctors' operation, and improve the safety and reliability of biopsy surgery.

The invention provides a robot scheduling method and device and a storage medium, and the method comprises the steps: determining that a target robot and other robots are in a target elevator region, and triggering the scheduling; wherein the target robot and the other robots comprise at least one elevator exiting robot and at least one robot planned to take an elevator; determining a scheduling strategy of the target robot according to the terrain information of the target elevator area and the operation information of each robot for triggering scheduling; and controlling the target robot to move according to the scheduling strategy. By means of the scheme, the situation that in the prior art, when the robot goes out of the elevator but other robots exist near the elevator, the robot is stuck can be effectively solved, and therefore normal execution of robot tasks can be guaranteed.

The invention belongs to the field of robot planning control, and particularly relates to a method for a cleaning robot to memorize obstacles under the condition of sensing limitation, which comprises the following steps of: acquiring multi-source point cloud data of a laser radar and a depth camera, and uniformly converting the multi-source point cloud data into a robot main body coordinate system through external parameter calibration; effective point clouds are screened based on a sensor visual field model, and sensor weights and distance weights are fused and then mapped to a local grid map; updating the grid occupancy probability by adopting a Bayesian updating rule, and executing time-based linear attenuation on an area which is not continuously observed; and finally, generating a global cost map according to the updated local cost map for path planning and obstacle avoidance control of the robot. According to the invention, the problem of unstable obstacle memory under sensing limited conditions such as a sensor blind area, shielding and illumination change is effectively solved, and the navigation safety and planning efficiency of the cleaning robot in a dynamic environment are improved.

The invention relates to the technical field of photovoltaic power generation accessories, in particular to a photovoltaic cleaning robot arrangement planning method, and the method comprises the steps: carrying out the block processing of each support in an original power station CAD arrangement drawing; a cleaning robot is distributed to the bracket subjected to block processing, and numbering processing is carried out; communication addresses are allocated to the cleaning robot according to areas, so that monitoring management of the cleaning robot during subsequent operation and maintenance is facilitated; and exporting a robot planning and designing list, wherein the list comprises the serial numbers of the cleaning robots, the walking distance of each cleaning robot, the corresponding communication addresses and the number of the cross-row machines. The workload of the arrangement scheme design of the cleaning robot is reduced, and the working efficiency is improved.

The invention relates to an information processing method for robot planning and an intelligent mowing robot. The information processing method for robot planning is used for an execution unit of the mowing robot. The information processing method for robot planning comprises the steps that S1, an execution unit obtains a grassland modeling graph; s2, obtaining grassland standardization information from the grassland modeling graph; and S3, performing path planning based on the grassland standardization information, and generating a mowing path scheme. In the prior art, under the condition that a mowing robot faces a large-range grassland, the workload for processing a to-be-mowed area is very large, time consumption is long, and the mowing robot is limited by the limited electric quantity of a battery, so that mowing cannot be completed at a time; on the basis of the scheme, the negative influence of the electric quantity on the mowing robot can be reduced under the condition of facing a large-range grassland, the time required for mowing is shortened, and then the mowing efficiency for processing the target large-range grassland is remarkably improved.

Techniques for training a vision-based robot control model include generating, based on scene data, a plurality of scenes, generating, based on the plurality of scenes, one or more goal specifications, determining, based on the one or more goal specifications and a robot model, one or more robot plans, generating, based on the one or more robot plans and the plurality of scenes, simulated sensor data, and performing one or more training operations to generate a trained vision-based robot control model based on the one or more goal specifications, the one or more robot plans, and the simulated sensor data.

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for robot planning. One of the methods includes maintaining a library of pre-generated candidate actions that represent respective robot capabilities; receiving an initial task plan for performing a particular task, wherein the initial task plan includes a sequence of two or more actions that comprises a first action followed by a second action; processing the initial task plan to determine that the first effect of the first action does not result in satisfaction of the second precondition of the second action; selecting one or more selected pre-generated candidate actions from the library of pre-generated candidate actions; and modifying the initial task plan to include the one or more selected pre-generated candidate actions as the intermediate actions to generate a final task plan to be executed by the one or more particular robots when performing the particular task.

The application discloses a kind of magnetic micro-robot parallel collaborative field navigation system, it is related to micro-nano operation technical field, and the micro resource conflict resolution scheduler MCDO in this system is based on the smooth single-machine path of dynamic evolution navigation planner DEN, the safety path constraint of conflict-aware evolutionary planner CAEP and the coil current with anti-interference control of double-field disturbance suppressor DFDS, to each magnetic micro-robot planning obstacle avoidance path and outputting multi-machine conflict-free instruction, to drive coil array to generate distributed gradient magnetic field, realize multi-robot independent drive;Micro-vision anchor MVAM is used to track the motion trajectory of magnetic micro-robot in real time, and provides robot pose data for MCDO and DFDS.The above-mentioned module realizes the whole process logic closed-loop management and control from path planning to physical drive, interference suppression to state monitoring, and is improved in path planning efficiency, anti-interference control accuracy, multi-machine cooperation capability, environment adaptation and the like.

This disclosure provides a motion optimization obstacle avoidance control method and system for vehicle-like robots, relating to the field of robot path planning and obstacle avoidance technology. The method includes: defining the state space of the vehicle-like robot; planning the global path of the vehicle-like robot; fitting the sparse state sequence using Hermite curves to obtain a time-space continuous trajectory; establishing a collision model between the vehicle-like robot and obstacles; transforming the geometric separation conditions between the robot and obstacles into a set of continuously differentiable nonlinear algebraic constraints based on duality theory; using the time-space continuous trajectory as a reference trajectory; and constructing a trajectory optimization objective function. This function can solve for the motion control quantities of the vehicle-like robot while satisfying its dynamic constraints, input constraints, and acceleration smoothing constraints, thus achieving obstacle avoidance planning and control of polygonal obstacles. This disclosure enables real-time, full-size robot planning and obstacle avoidance control.

To enable a mobile robot to arrive at a destination by required time.SOLUTION: A processing circuit of a robot movement system is configured to acquire priorities of a first mobile robot and a second mobile robot, determine whether or not the first mobile robot and the second mobile robot satisfy determination requirement including simultaneous use requirement that the first mobile robot and the second mobile robot will simultaneously use an elevator, determine which of the priority of the first mobile robot and the priority of the second mobile robot is higher, and execute priority processing of determining, when the determination requirement is satisfied, a call floor and a destination floor of an elevator car so that the priority robot of the first mobile robot and the second mobile robot which has higher priority than the non-priority mobile robot of the first mobile robot and the second mobile robot which has lower priority is prioritized.SELECTED DRAWING: Figure 6

The invention discloses a mobile robot planning method and system for a dynamic position environment, namely, dynamic environment sensing and re-planning are introduced into a robot planning method based on sampling, specifically, in the moving process along a robot planning path, the surrounding environment is sensed synchronously, and the obstacle change type is recognized; the method comprises the following steps: constructing a blocking node set consisting of nodes blocked by a newly detected obstacle and child nodes thereof, constructing a new node set by nodes in a blank area left after the obstacle disappears, performing dynamic processing on an edge structure of a tree, deleting edges of blocking nodes for blocking a current path for an unknown obstacle, and then performing dynamic processing on the edge structure of the tree; transferring other nodes to an unaccessed node set, transferring nodes in a new node set to the unaccessed node set for disappearing obstacles, participating in subsequent planning path generation, and dynamically adjusting the tree; in addition, inert rewiring and convex hull structures and directional activation are provided, the exploration direction is dynamically adjusted, and the replanning efficiency is improved.

Approaches presented herein provide for a framework to integrate human provided feedback in natural language to update a robot planning cost or value. The natural language feedback may be modeled as a cost or value associated with completing a task assigned to the robot. This cost or value may then be added to an initial task cost or value to update one or more actions to be performed by the robot. The framework can be applied to both real work and simulated environments where the robot may receive instructions, in natural language, that either provide a goal, modify an existing goal, or provide constraints to actions to achieve an existing goal.

Techniques for controlling a robot include receiving a task and motion planning (TAMP)-domain language input, a goal, and sensor data, generating a plan skeleton based on the TAMP-domain language input, the goal, and the sensor data, generating, based on the plan skeleton, one or more feasible particles, generating, based on the plan skeleton and at least one of the one or more feasible particles, a robot plan, and causing the robot to perform at least part of the robot plan.

The present disclosure relates to the technical field of robot control, and proposes a robot task planning method and system based on graph representation learning and an electronic device, including the following steps: constructing an a priori knowledge reasoning model; based on the constructed a priori knowledge reasoning model, extracting robot planning elements to be processed as graph nodes according to obtained natural language information of robot service execution, taking the relationship between the robot planning elements as the edges of the graph, and constructing a directed graph; performing vectorization on the information of the constructed directed graph and the natural language information of robot service execution to obtain a vectorized representation of task planning; and based on a deep learning model, using a progressive method of template generation and template filling to generate a phased task planning of robot actions. The robot task planning is effectively graphically represented considering service execution requirements, and the service-related natural language is autonomously converted into a selective task planning under the guidance of reinforcement learning.

A robot location planning method and apparatus based on reinforcement learning. The location planning method comprises: performing, by means of a preset positioning pen module, digital description on the location, shape and posture information of an object to be manipulated by means of a robot, so as to obtain digital object data; by using the digital object data as an input to a deep neural network of a simulation platform corresponding to the robot, and using, as an output, a manipulation location at which the robot controls the object, performing training to obtain an initial location planning model of the robot; on the basis of an application scenario of the object, selecting a corresponding observation space and task return logic to perform scenario optimization on the initial location planning model, so as to obtain a robot location planning model; and on the basis of the robot location planning model, predicting the digital object data of the object to be manipulated, so as to obtain a target location at which the robot manipulates the object to be manipulated. Therefore, on the basis of a trained robot location planning model, the robot planning accuracy meets the requirements of various industrial scenarios, and the workload involved in manual operations is greatly reduced.

A computer-implemented system generates an optimal trajectory for completion of a task by a robot that merges robot-oriented behavior with human expectation. The system determines, based on task information and by iterative modification of one or more policy parameters of a policy descriptive of one or more task-oriented actions, one or more updated policy parameters of an updated policy descriptive of an updated set of task-oriented actions for completion of the task by the computer-implemented agent that: (a) satisfies a task constraint associated with the task and a safety constraint on a robot-oriented expected return value associated with a robot-oriented reward for the updated set of task-oriented actions; and (b) maximizes a human-oriented expected return value associated with a human-oriented reward for the updated set of task-oriented actions.

The present application relates to the technical field of robot planning, in particular to an advertisement display robot cruise path intelligent planning method for commercial scenarios, comprising: collecting a commercial scenario static environment map, a dynamic people flow density heat map, real-time state information of an advertisement display robot and an advertisement delivery task list. Identifying the passable area of the scene, the people flow gathering hotspot area and the fixed obstacle area to be avoided, generating recommended stopping points according to the task priority, the target crowd portrait and the expected exposure time. Combined with multiple constraint conditions, the ant colony algorithm optimized by the joint evaluation model of delivery value and energy consumption is used to plan the globally optimal cruise path. Real-time monitoring of environmental changes and dynamic adjustment of the route during the journey, recording running and exposure data to update people flow information. The method continuously iterates the path planning logic, adapts to the commercial dynamic complex environment, and realizes the intelligent and reasonable layout of the advertisement display robot cruise path.

A robotic controller for controlling a robot according to a sequence of robotic actions, comprises an input interface configured to receive a plurality of multimodal inputs each specifying instructions for performing a task in a different modality including audio, video, and a text modality. The controller also comprises a multimodal large language model, an action sequence decoder, and a controller. The multimodal LLM includes a multimodal LLM encoder and an LLM decoder. The multimodal LLM encoder is trained with machine learning to transform the multimodal instructions into encodings and the LLM decoder is configured to decode the encodings into a sequence of robotic instructions. The action sequence decoder is trained with machine learning to transform the sequence of robotic instructions into a sequence of actions using a library of robotic skills. The controller is configured to control a robot according to the sequence of actions.