Robot control system based on artificial intelligence

Abstract

The invention discloses a robot control system based on artificial intelligence, which comprises a data acquisition module, an AI decision module, a control execution module, a safety protection module and a power management module, and is characterized in that the data acquisition module comprises an IMU sensor and an RGB-D camera, and is used for synchronously acquiring pose and environment data; the AI decision module comprises a diffusion Transform model unit used for processing emergency obstacle avoidance type real-time tasks, and further comprises a visual language model unit used for executing complex path planning. Rapid response and dynamic adjustment response of complex work execution planning are achieved through dual-system decision of the AI decision module, meanwhile, load dynamic response and real-time adjustment are achieved through the PID parameter dynamic adjustment module arranged in the joint controller, and therefore the device can more flexibly cope with load changes, and the control precision is improved. The method has the advantages that the operation continuity of the robot is improved, the adaptive capacity is higher, an ethical bottom layer constraint mechanism is achieved, and the capacity of judging the currently executed behavior harm is achieved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence robot technology, and more specifically, to an artificial intelligence-based robot control system. Background Technology

[0002] Artificial intelligence robots are intelligent systems that integrate perception, decision-making, and execution capabilities. Their core lies in real-time acquisition of environmental data through multimodal sensors such as vision and force sensors, combined with deep learning algorithms for dynamic decision-making, ultimately driving mechanical structures to complete complex tasks.12 A key difference between these systems and traditional automated equipment lies in their nonlinear control characteristics—the motion of each joint exhibits coupling effects, requiring decoupling algorithms to coordinate multi-degree-of-freedom motion.34 Modern artificial intelligence robots also possess environmental adaptability; by establishing information databases and reinforcement learning mechanisms, they can automatically optimize control strategies based on working conditions.14 For example, in industrial scenarios, their control systems must simultaneously meet the requirements of sub-millimeter positioning accuracy and millisecond-level response speed.

[0003] However, most current AI robot control systems still rely on preset programs and algorithms for motion control, which makes them poorly adaptable to sudden load changes, affecting work continuity. They also lack ethical underlying constraints, and tend to run pre-programmed operations during execution, resulting in a low ability to judge the harmfulness of the current execution behavior. Summary of the Invention

[0004] In view of the problems existing in the prior art, the purpose of this invention is to provide a robot control system based on artificial intelligence to solve the problems in the background technology.

[0005] To achieve the above objectives, the present invention adopts the following technical solution.

[0006] An artificial intelligence-based robot control system includes a data acquisition module, an AI decision-making module, a control execution module, a safety protection module, and a power management module.

[0007] The data acquisition module includes an IMU sensor and an RGB-D camera, used to simultaneously acquire pose and environmental data;

[0008] The AI ​​decision-making module includes a diffusion Transformer model unit for handling real-time tasks such as emergency obstacle avoidance, and a visual language model unit for performing complex path planning.

[0009] The control execution module includes multiple joint controllers, which receive motion commands via a CAN bus connected to the diffusion Transformer model unit and the visual language model unit.

[0010] The safety protection module includes a balance state monitoring unit for detecting the robot's real-time balance state.

[0011] The power management module is used to control the robot's built-in battery power supply, as well as to control standby voltage regulation and dynamic power consumption allocation strategies.

[0012] As a further description of the above technical solution: the IMU sensor and the RGB-D camera use an extended Kalman filter algorithm to perform multimodal fusion calculations of environmental data to generate sensing data in the same coordinate system.

[0013] As a further description of the above technical solution: the corresponding frequency of the diffusion Transformer model unit is 120Hz.

[0014] As a further description of the above technical solution: the joint controller adopts an improved Jacobian matrix inverse solution algorithm to realize six-degree-of-freedom coordinated control, and the joint controller has a built-in PID parameter dynamic adjustment module.

[0015] As a further description of the above technical solution: the joint controller is also equipped with an overcurrent detection module, which is used to detect the current of the execution unit of the joint controller and provide real-time feedback to the AI ​​decision module.

[0016] As a further description of the above technical solution: the security protection module also has a built-in ethical constraint data unit, which contains rule data for verifying the compliance of the executed task.

[0017] As a further description of the above technical solution: the data acquisition module also includes a human-computer interaction module, which has a built-in voice acquisition unit for human-computer voice interaction.

[0018] As a further description of the above technical solution: the human-computer interaction module also includes a 3D gesture recognition unit, which is trained and recognized using a 3D convolutional neural network.

[0019] Compared with the prior art, the advantages of this invention are:

[0020] (1) This solution achieves fast response and dynamic adjustment response of complex work execution planning through the dual system decision-making of the AI ​​decision module. At the same time, it utilizes the PID parameter dynamic adjustment module built into the joint controller to achieve dynamic load response and real-time adjustment, thereby enabling the device to cope with load changes more flexibly, improve the continuity of robot operation, and have stronger adaptability.

[0021] (2) This solution uses the built-in rule data of the ethical constraint data unit to verify the compliance of the task execution. It uses the data from the data acquisition module of the AI ​​decision module to perform fusion perception calculation and verify it with the rule data inside the ethical constraint data unit to make a judgment on the harm of the action. Thus, the device has the advantage of having an ethical underlying constraint mechanism when the robot executes the preset program and has the ability to judge the harm of the current behavior. Attached Figure Description

[0022] Figure 1 This is a schematic diagram illustrating the operating principle of the present invention;

[0023] Figure 2 This is a schematic diagram of the data acquisition module of the present invention.

[0024] Figure 3 This is a schematic diagram of the joint controller principle structure of the present invention;

[0025] Figure 4 This is a schematic diagram of the principle structure of the safety protection module of the present invention.

[0026] Explanation of the labels in the diagram:

[0027] 1. Data Acquisition Module; 10. IMU Sensor; 11. RGB-D Camera; 2. AI Decision Module; 21. Diffusion Transformer Model Unit; 22. Visual Language Model Unit; 3. Control Execution Module; 31. Joint Controller; 311. Overcurrent Detection Module; 32. PID Parameter Dynamic Adjustment Module; 4. Safety Protection Module; 41. Balance State Monitoring Unit; 42. Ethical Constraint Data Unit; 5. Power Management Module; 6. Human-Computer Interaction Module; 61. Built-in Voice Acquisition Unit; 62. 3D Gesture Recognition Unit. Detailed Implementation

[0028] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0029] To address the issue that most current AI robot control systems still rely on preset programs and algorithms for motion control, resulting in poor adaptability to sudden load changes and affecting work continuity, Example 1 is proposed:

[0030] Please see Figures 1-3In this invention, an artificial intelligence-based robot control system includes a data acquisition module 1, an AI decision-making module 2, a control execution module 3, a safety protection module 4, and a power management module 5. The data acquisition module 1 also includes a human-computer interaction module 6, which has a built-in voice acquisition unit 61 for human-computer voice interaction. The human-computer interaction module 6 also includes a 3D gesture recognition unit 62, which is trained and recognized using a 3D convolutional neural network.

[0031] The data acquisition module 1 includes an IMU sensor 10 and an RGB-D camera 11, which are used to simultaneously acquire pose and environmental data. The IMU sensor 10 and the RGB-D camera 11 use an extended Kalman filter algorithm to perform multimodal fusion calculation of environmental data and generate perception data in the same coordinate system.

[0032] The AI ​​decision-making module 2 includes a diffusion Transformer model unit 21 for handling real-time tasks such as emergency obstacle avoidance, and a visual language model unit 22 for performing complex path planning.

[0033] The control execution module 3 includes multiple joint controllers 31. The joint controllers 31 are connected to the diffusion Transformer model unit 21 and the visual language model unit 22 via a CAN bus to receive motion commands. The corresponding frequency of the diffusion Transformer model unit 21 is 120Hz. The joint controllers 31 use an improved Jacobian matrix inverse algorithm to achieve six-degree-of-freedom coordinated control. The joint controllers 31 have a built-in PID parameter dynamic adjustment module 32. The joint controllers 31 also have an overcurrent detection module 311 inside. The overcurrent detection module 311 is used to detect the current of the execution unit of the joint controllers 31 and provide real-time feedback to the AI ​​decision module 2.

[0034] The safety protection module 4 includes a balance state monitoring unit 41 for detecting the robot's real-time balance state.

[0035] The power management module 5 is used to control the power supply of the robot's built-in battery, as well as to control the standby voltage adjustment and dynamic power consumption allocation strategy.

[0036] In this embodiment, an IMU sensor 10 and an RGB-D camera 11 are combined to synchronously acquire pose and environmental data. After multimodal data fusion calculation by the AI ​​decision module 2, perception data in a unified spatiotemporal coordinate system is generated. The diffusion Transformer model unit 21 handles real-time tasks such as emergency obstacle avoidance, while the visual language model unit 22 performs complex path planning, enabling the robot to perform long-distance movement and complex actions such as climbing. Both utilize an attention mechanism to dynamically allocate weights, reducing decision latency. Then, running commands are sent to each joint controller 31. Each joint controller 31 adopts a modified... The advanced Jacobian matrix inverse algorithm realizes six-degree-of-freedom coordinated control to achieve micron-level end-effector positioning. The PID parameter dynamic adjustment module 32 dynamically adjusts the parameters, automatically updating the control parameters every ten seconds according to load changes and feeding them back to the AI ​​decision module 2. The overcurrent detection module 311 performs real-time filtering and detection. When the current exceeds the set value, the AI ​​decision module 2 triggers the overload hardware protection mechanism, including shutdown and stopping the current action to protect the robot hardware. This enables the device to respond more flexibly to load changes, improve the continuity of robot operation, and enhance adaptability.

[0037] Furthermore, to address the issues of a lack of ethical underlying constraints, excessive pre-installed procedures during execution, and a low ability to assess the harmfulness of the current execution behavior, Implementation Example Two is proposed:

[0038] Please see Figure 1 and Figure 4 The security protection module 4 also has a built-in ethical constraint data unit 42, which contains rule data for verifying the compliance of the task execution.

[0039] In this embodiment, the built-in rule data of the ethical constraint data unit 42 is used to verify the compliance of the executed task. When the robot needs to execute a preset instruction action, the AI ​​decision module 2 uses the data from the data acquisition module 1 to perform fusion perception calculation. Based on the current state reflected by the data, it verifies the rule data inside the ethical constraint data unit 42. If the verification passes, the robot executes normally. If the verification fails, a warning is issued and the execution stops. This achieves the advantage that the device has an ethical underlying constraint mechanism when the robot executes a preset program and has the ability to judge the harm of the currently executed behavior.

[0040] The above description is merely a preferred embodiment of the present invention; however, the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and its improved concepts, should be covered within the scope of protection of the present invention.

Claims

1. A robot control system based on artificial intelligence, characterized in that: It includes a data acquisition module (1), an AI decision-making module (2), a control execution module (3), a security protection module (4), and a power management module (5). The data acquisition module (1) includes an IMU sensor (10) and an RGB-D camera (11) for synchronously acquiring pose and environmental data; The AI ​​decision module (2) includes a diffusion Transformer model unit (21) for handling real-time tasks such as emergency obstacle avoidance, and a visual language model unit (22) for performing complex path planning; The control execution module (3) includes multiple joint controllers (31), which are connected to the diffusion Transformer model unit (21) and the visual language model unit (22) via a CAN bus to receive motion commands. The safety protection module (4) includes a balance state monitoring unit (41) for detecting the robot's real-time balance state; The power management module (5) is used to control the power supply of the robot's built-in battery, as well as to control the standby voltage adjustment and dynamic power consumption allocation strategy.

2. The robot control system based on artificial intelligence according to claim 1, characterized in that: The IMU sensor (10) and RGB-D camera (11) use an extended Kalman filter algorithm to perform multimodal fusion calculations of environmental data and generate sensing data in the same coordinate system.

3. The robot control system based on artificial intelligence according to claim 1, characterized in that: The corresponding frequency of the diffusion Transformer model unit (21) is 120Hz.

4. The robot control system based on artificial intelligence according to claim 1, characterized in that: The joint controller (31) uses an improved Jacobian matrix inverse algorithm to achieve six-degree-of-freedom coordinated control, and the joint controller (31) has a built-in PID parameter dynamic adjustment module (32).

5. The robot control system based on artificial intelligence according to claim 1, characterized in that: The joint controller (31) is also equipped with an overcurrent detection module (311), which is used to detect the current of the execution unit of the joint controller (31) and provide real-time feedback to the AI ​​decision module (2).

6. The robot control system based on artificial intelligence according to claim 1, characterized in that: The security protection module (4) also has a built-in ethical constraint data unit (42), which contains built-in rule data for verifying the compliance of the task execution.

7. The robot control system based on artificial intelligence according to claim 1, characterized in that: The data acquisition module (1) also includes a human-computer interaction module (6), which has a built-in voice acquisition unit (61) for human-computer voice interaction.

8. The robot control system based on artificial intelligence according to claim 7, characterized in that: The human-computer interaction module (6) further includes a 3D gesture recognition unit (62), which is trained and recognized using a 3D convolutional neural network.

Images