906 results about "Robotic hand" patented technology

The invention discloses a manipulator grabbing method based on deep learning target detection and image segmentation, and relates to the technical field of artificial intelligence and robotics.The manipulator grabbing method comprises the following steps that a scene image to be processed is collected, the image quality is improved through the multi-light-source fusion image enhancement technology, and recognition errors caused by uneven illumination are reduced; and inputting the enhanced image to a pre-trained deep learning model, executing a target detection task, and outputting an initial bounding box and a category label of the target object. According to the method, through multi-light-source image enhancement and high-precision image segmentation, the accuracy of target recognition and contour extraction is remarkably improved, and the capture failure rate caused by image misjudgment is reduced. And meanwhile, geometric consistency verification and a multi-factor grabbing scoring mechanism are introduced, dynamic screening and collision pre-detection are conducted on the paths, the grabbing stability and safety of the mechanical arm in the complex environment are effectively guaranteed, and the intelligence and robustness of the whole system are remarkably improved.

The invention discloses an operation and maintenance manipulator intelligent control method and system based on visual identification, and relates to the technical field of intelligent manipulator control, and the method comprises the steps: collecting RGB image data and depth image data of an operation and maintenance operation area, and obtaining a standardized image matrix and a mapping relation matrix; inputting the standardized image matrix into an improved ResNet residual network model, generating a comprehensive feature descriptor, and calculating a spatial position coordinate and an attitude angle of the target equipment based on the mapping relation matrix; based on the current joint angle state of the manipulator, an improved Jacobian matrix inverse kinematics algorithm is used for solving a target angle sequence of each joint, a preset operation mode library is matched based on the comprehensive feature descriptor, and a grabbing force parameter and a motion speed parameter are determined; and converting the target angle sequence into a control instruction, and sending the control instruction to each joint driver of the manipulator to drive the manipulator to complete action planning. According to the invention, full-process automation from environment perception to task execution is realized.

The invention provides a full-closed-loop control method and system for a multi-station servo feeding manipulator, and the method comprises the steps: firstly, detecting the thickness of a plate in real time, querying a preset thickness-pressure relation, and determining a target pressure value when the multi-station servo feeding manipulator carries out the grabbing of the plate; a real-time pressure value is then generated through a dynamic pressure regulation mechanism. And the system compares the real-time pressure with the target pressure to obtain a deviation value, dynamically adjusts proportional, integral and differential parameters in combination with a fuzzy rule base, and calculates and generates a dynamic pressure compensation amount by using the adjusted parameters. And the compensation amount and the target pressure are superposed to form a comprehensive control instruction, and the comprehensive control instruction is fed back to the execution mechanism to realize closed-loop adjustment, so that the pressure stability and the position fixation of the plate in the whole grabbing and transferring process are finally kept, and the slippage phenomenon is effectively eliminated. The pressure control precision and the anti-sliding stability in the plate transferring process are improved.

The invention relates to the technical field of badge production, in particular to a badge production machine. Comprising a machine table, a linear circulating conveying mechanism, a plurality of accompanying carriers, a backing paper feeding mechanism, a picture feeding mechanism, a backing paper transfer carrier, a picture transfer carrier, a pin assembling device, an elastic piece bending forming station, a pin flattening mechanism, a press fitting device, a lower cover feeding station, a pin feeding mechanism and a multi-head synchronous feeding and discharging mechanical arm. The invention discloses a multi-head synchronous transfer manipulator and a bending driving mechanism. Synchronous assembly of the upper cover assembly and the lower cover assembly is achieved, online assembly of the pin is also achieved, the pin is directly assembled below the bottom face of the lower cover, subsequent production and processing can be conducted without conducting redundant overturning actions on the lower cover and the lower cover assembly, the badge production efficiency is greatly improved, and the production cost is reduced. And all feeding and transferring actions can be completed only by two manipulators, so that the number of the manipulators is reduced, and the production and manufacturing cost is reduced.

The utility model discloses a finger mechanism and a robot, relates to the technical field of robots, and solves the technical problems that in the prior art, a finger mechanism hinders bending of a palm of a mechanical arm, and the palm is difficult to be matched with the finger mechanism to grasp the finger mechanism, the finger mechanism comprises a rigid driving assembly, a connecting rod assembly and a knuckle assembly, the connecting rod assembly is connected with the rigid driving assembly and the knuckle assembly, the rigid driving assembly is driven to drive the knuckle assembly to grab an object to be grabbed through the connecting rod assembly, and the rigid driving assembly, the connecting rod assembly and the knuckle assembly are arranged in the first direction and in the second direction. The connecting position of the connecting rod assembly and the rigid driving assembly is higher than the connecting position of the connecting rod assembly and the knuckle assembly so that the rigid driving assembly can be driven to deviate relative to the knuckle assembly in the second direction, the second direction is the direction away from the knuckle assembly, and an included angle is formed between the first direction and the second direction.

The invention discloses a jewelry flaw detection and grading system based on machine vision and a manipulator. The system comprises a multi-mode imaging module; the image processing module is used for carrying out preprocessing operation on the original image; the edge processing module comprises a GPU acceleration processor and is used for deploying a flaw detection model to detect flaws; training data are automatically generated through a semi-supervised learning algorithm; the mechanical execution module comprises a six-axis mechanical arm, a path planning algorithm and a collision detection model for a bounding box and a directed bounding box of the six-axis mechanical arm; the cloud management module is used for storing detection data based on a block chain technology and supporting third-party verification; and dynamically updating the expert knowledge base. According to the jewelry sorting method and system, the types and grades of the jewelry are divided through the DINOv2 defect detection model, meanwhile, the PnP algorithm is adopted to be fused with the depth information, the jewelry is sorted according to grade division, meanwhile, consumers can rapidly and accurately obtain relevant information (such as materials, authenticity and value evaluation) of the jewelry, and the shopping experience is greatly improved.

The invention discloses a flexible finger friction contact collision dynamics simulation method based on an absolute node coordinate method, which comprises the following steps: simplifying a finger of a flexible manipulator into a connecting rod system, taking a tail end knuckle as a flexible beam for consideration, and carrying out unit division on the flexible beam system based on the absolute node coordinate method; a contact collision process is described by using an additional constraint method, meanwhile, switching of a viscous state and a sliding state is considered, a friction-containing flexible finger contact collision dynamic model is established, and a dynamic equation of a system is solved by adopting a double-circulation implicit integral algorithm; and obtaining a curve graph that the flexible finger tail end displacement changes along with time and the finger tail end contact force changes along with time. The invention provides a new frictional contact collision dynamic model for a flexible manipulator system, and provides more perfect data information and image information for researchers in the field.

The invention discloses a hub bearing eddy current testing machine, and relates to the technical field of hub bearing detection. A hub bearing eddy current testing machine comprises a workbench, a frame body is installed on the workbench, a first mechanical arm is installed on the frame body, the first mechanical arm is provided with a clamping and rotating mechanism, and the clamping and rotating mechanism is configured to be capable of clamping parts and driving the parts to rotate; a second mechanical arm is arranged on the workbench, a detection mechanism is installed on the second mechanical arm, and the detection mechanism is configured to conduct inner end face detection and flange tooth detection on rotating parts; the workbench is further provided with a feeding mechanism and a discharging mechanism, wherein the feeding mechanism is used for conveying the parts to the clamping and rotating mechanism, and the discharging mechanism is used for discharging the parts. The detection efficiency can be improved.

Provided is a technique that allows even a person without knowledge on robots to intuitively and efficiently teach a work motion a robot by a work demonstration while maintaining a normal human motion as much as possible, and that can reduce the man-hours for pre-adjustment of teaching, facilitate the introduction, etc. A robot teaching method is a method for performing teaching for generating a motion of a robotic hand mechanism corresponding to a work motion including an operation on an operation object by a hand of a teacher based on measurement of a work motion, and includes, as steps performed by a computer system: a step of acquiring a first measured pose obtained by measuring a pose of the operation object and a second measured pose obtained by measuring a pose of the hand of the teacher during the work motion; a step of detecting the operation on the operation object by the teacher; and a step of generating a teaching pose for generating the robotic motion data based on the first measured pose, the second measured pose, and the detected operation.

The invention discloses a computer case heat dissipation side plate stamping machining device and technology, and relates to the technical field of computer case machining. The machining device comprises an unwinding module used for outputting a metal plate and a stamping module; the stamping module comprises a first stamping part, a second stamping part, a third stamping part and a cutting part which are arranged in sequence; a feeding module is further arranged between the unwinding module and the stamping module, and the feeding module is used for conveying the unwound and output metal plates to the stamping module at equal intervals; according to the metal plate stamping and cutting device, the multiple stamping parts and the cutting part are integrated, the feeding module is arranged to convey unreeled and output metal plates at equal intervals, and after one-time stamping is completed, a mechanical arm does not need to be adopted for transferring the metal plates, so that the stamping efficiency can be improved, the stamping precision of all the stamping parts is guaranteed, and the production cost is reduced. And the deflection phenomenon in the transferring process is avoided, and the production quality is further improved.

The invention discloses an integrated tile-shaped magnet outer circle grinding device and method based on artificial intelligence, and belongs to the technical field of grinding machining. The integrated tile-shaped magnet outer circle grinding device comprises a feeding device, a grinding device and a visual detection device; the feeding device comprises a feeding conveying belt, a guide groove, a spring bin, a material blocking barrel, a feeding barrel, a magnetic suction cup, an electromagnetic push rod, a cooperative mechanical arm, a reset spring and a discharging conveying belt. The visual detection device comprises a magnetic shoe counting camera, an annular light source, a counting camera bracket, a measuring camera bracket, a backlight plate, an outer contour measuring camera, a positioning camera and a positioning camera bracket; integrated automatic operation of feeding, grinding, detection and discharging is achieved, the production efficiency is greatly improved, automatic counting, precision detection and discharging are completed through cooperation of visual detection and the conveying belt, the manual operation intensity is effectively reduced, production interruption caused by personal errors is avoided, and the productivity in unit time is remarkably improved.

The invention discloses an under-actuated metamorphic manipulator integrating grasping and pressing operations, which comprises a palm bracket, at least two under-actuated grasping fingers, a rigid pressing single finger and a driving mechanism, and is characterized in that the under-actuated grasping fingers and the rigid pressing single finger are arranged on the palm bracket. The main functions of the manipulator include self-adaptive enveloping and grabbing of objects in different shapes and execution of operations such as switch and button pressing, the enveloping and grabbing function is mainly completed by the under-actuated grabbing finger, and the rigid pressing single finger can also play a role in auxiliary clamping; in the pressing function, force is output by a mechanical arm connected with the manipulator, and pressure is applied to a target through a rigid pressing single finger. The under-actuated metamorphic manipulator integrating grabbing and pressing operation is used for fine operation in the flammable and explosive environment, the grabbing adaptability and the structural rigidity are both considered, multiple kinds of grabbing and pressing operation can be conducted, and the under-actuated metamorphic manipulator has the advantages of being light in weight, high in integration level and large in output force.

The invention discloses a smart type library-based teleoperation method, system and device, and a storage medium. The method comprises the following steps of: constructing a smart operation type library; analyzing the task instruction through a preset multi-modal large language model MLLM to obtain a corresponding operation step, and matching a dexterous operation type corresponding to the current step from a dexterous operation type library; intuitive accurate control between the hand and the manipulator is realized through an interpolation mapping strategy. The interpolation mapping strategy specifically comprises the following steps: capturing a fingertip position of a human hand in real time; calculating the normalized projection proportion of the current posture of the human hand relative to the stretching state and the contraction state; and according to the projection proportion, linear interpolation is carried out on the extension and contraction joint angles of the manipulator, and a manipulator joint angle instruction is generated. According to the method, the adaptability and stability of teleoperation are provided by establishing a systematized smart type library and combining a type automatic retrieval module assisted by a multi-modal large language model and an interpolation mapping control strategy.

The utility model discloses a high-flexibility flexible mechanical finger supporting forward and reverse holding and a manipulator. The mechanical finger comprises a middle finger section, a base finger section and an end finger section, wherein the base finger section and the end finger section are rotationally connected with the two ends of the middle finger section respectively. Driving units are arranged in the end finger section, the middle finger section and the base finger section; bevel gear transmission units are arranged at the tail end of the end finger section, the tail end of the middle finger section and the tail end of the base finger section; the driving units are in transmission connection with the corresponding bevel gear transmission units; the driving unit drives the end finger section and the base finger section to rotate by taking the corresponding bevel gear transmission unit as a central shaft; through the structure and position design of the micro driving motor and the bevel gear transmission unit, each finger and each knuckle of the manipulator can be flexibly controlled, so that the manipulator has higher bionic flexibility and more precise grabbing capacity, and can be suitable for different scene requirements; by means of the special inserting mode between knuckles, the left-right hand grabbing function of the single mechanical arm is achieved.

The invention, which relates to the technical field of device testing, discloses a visual self-calibration system for a microneedle automatic patch device and a calibration method thereof, comprising: a data acquisition module for measuring a spatial relationship among a composite calibration target, a calibration visual camera, a laser contourgraph and a manipulator coordinate system; the compensation model building module is used for building a corresponding relation model of the grabbing condition and the pose compensation amount based on patch deformation data collected by the laser contourgraph; the drifting prediction module is used for monitoring the motor temperature and the movement mileage and establishing a drifting prediction model; the deformation pre-compensation module is used for acquiring deformation characteristics after the patch is picked up and processing and acquiring a pose pre-compensation amount by utilizing a compensation model; the deformation compensation correction module fuses the pose pre-compensation amount and the drift pre-compensation amount, generates a final compensation control instruction and drives the manipulator to complete compensation motion; under complex operation conditions, the deformation compensation model shows great adaptability and robustness.

Apparatuses, systems, and techniques to perform a grasp of on object using an articulated robotic hand equipped with one or more tactile sensors. In at least one embodiment, a machine-learned model trained in simulation to grasp a cuboid using signals received from tactile sensors is applied to grasping objects of various shapes in a real-world environment.

The invention discloses an intelligent learning method for grabbing of a three-finger dexterous hand for multiple types of workpieces, and belongs to the technical field of intelligent grabbing application of manipulators. A pre-defined rule and simulation expert double-layer mode is provided, a high-quality demonstration data set is automatically generated, a simulation grabbing scene of a mechanical arm and a three-fingered dexterous hand is constructed, recognition and classification of multiple types of workpieces are achieved by analyzing workpiece contour features, grabbing hand shapes are matched according to the recognition and classification, strategy pre-training is carried out through simulation learning, and the recognition and classification efficiency is improved. And optimizing the strategy based on a PPO algorithm and a composite reward function. Aiming at the problem of lack of a standardized three-fingered dexterous hand parameter model, a modeling method based on parameterized URDF and automatic script generation is used to realize efficient configuration and optimization of the three-fingered dexterous hand in simulation. Through performance evaluation, stable and high-quality grabbing of multiple types of workpieces can be achieved.

The invention discloses a control method and device of a multi-mode intelligent manipulator, and relates to the technical field of manipulators and artificial intelligence. Capturing left and right images of the target object by using a binocular vision technology; outputting the category and boundary coordinates of the target object through a YOLO model; extracting center coordinates of a target object in the left and right images; matching objects consistent with the target object in category and similar to the target object in coordinate in the left and right images; calculating parallax and depth information between the left image and the right image to combine three-dimensional space coordinates of the target object; smoothing the historical coordinates of each target object to output a target detection result of the YOLO model; inputting the target detection result into a task decision model and converting the target detection result into a structured instruction; the grabbing force is executed, calculated and adjusted according to the structured instruction; according to the control method and device for the multi-mode intelligent manipulator, through deep learning and an intelligent control algorithm, the manipulator can autonomously recognize a target and optimize an execution path.

The invention is suitable for the field of manipulators, and provides a frame butt-joint assembly mechanical arm which comprises a supporting base body, and two position adjusting tables are arranged on the supporting base body and driven by an adjusting unit to be close to or away from each other. A rotary table is arranged on each position adjusting table and drives a first arm rod to rotate, the first arm rod is connected with a second arm rod and the assembly bearing plate through a hinge structure, and the angle of the first arm rod is controlled by a first push-pull cylinder and a second push-pull cylinder. An adjustable rotary clamping unit is installed on the assembly bearing plate and comprises a third push-pull cylinder and a second motor, and the motor drives a sleeve to rotate and drives a guide plate and a grabbing arm rod to rotate. The grabbing arm rod is connected with the clamping jaw through a quick release assembly, and a shaft sleeve of the third push-pull cylinder is in linkage with the grabbing arm rod through a transmission connecting arm. The rotary welding device has the advantages of being flexible in adjustment, high in adaptability, stable in clamping and convenient to quickly adjust and perform rotary welding.

The invention relates to a microswitch assembling device, which comprises an assembling frame, a plurality of first manipulators, a plurality of first vibration discs, a first conveying mechanism, a dispensing assembly, a second conveying mechanism and a drying assembly, and is characterized in that the first manipulators are connected with the first vibration discs; the first mechanical arms are arranged towards the first conveying mechanism, the first conveying mechanism is connected with the second conveying mechanism, the dispensing assembly is arranged towards the second conveying mechanism, and the second conveying mechanism is arranged towards the drying assembly. The invention provides the assembling device capable of automatically assembling the microswitch.

The invention discloses a consecutive fruit picking control system and method based on Simulink and ROS. The system comprises forward and inverse kinematics models of a mechanical arm based on an MATLAB / Simulink software platform, and forward and inverse solution control system modules of the mechanical arm are established; a network connection and communication mechanism of a Linux operating system and an ROS system is established, data is issued to Simulink according to an ROS message format, and a fruit detection visual system module is constructed; opening and closing of fingers can be controlled according to the state of a picking hand controller, two message formats under an ROS are established, and construction of a picking hand system control module is completed. And according to target positioning and forward and inverse solution models, planning a motion track of the mechanical arm, controlling the tail end of the mechanical arm to operate, and completing the whole process from an approaching point to a target fruit, and picking and collecting the fruit by the mechanical arm. A visual identification system, a picking hand grabbing system and a mechanical arm motion control system are integrated in the whole system, and the problems that subsystems of a traditional picking robot are difficult to integrate, and the development period is long are solved.

The utility model discloses an annular workpiece precise clamping device which comprises a charging tray, a mobile robot, a temporary storage mechanism and a mobile mechanism, the charging tray is provided with material grooves which are arranged vertically and horizontally, and the material grooves are used for placing annular workpieces; the feeding conveying device is used for conveying unmachined annular workpieces, the discharging conveying device is used for conveying machined annular workpieces, and the temporary storage mechanism is provided with a first temporary storage conveying device and a second temporary storage conveying device which are arranged in a spaced mode. The moving mechanism comprises a ground rail device, a moving seat arranged on the ground rail device, a connecting mechanism arranged on the moving seat, a platform arranged on the moving seat and a multi-axis manipulator arranged on the platform, the platform is provided with a connecting jig, and the connecting jig is used for placing and positioning the annular workpiece. Through cooperation of the mobile robot, the temporary storage mechanism, the transfer mechanism, the ground rail device, the multi-axis mechanical arm and the platform, automatic conveying and clamping of the annular workpieces are achieved, and unmanned work is achieved.

The invention relates to the technical field of injection molding production and manufacturing, and discloses an injection molding raw material feeding mistake-proofing verification system for a safety system airbag, which comprises a central verification server, an edge computing gateway, an intelligent hopper unit provided with an electromagnetic locking mechanism and a handheld intelligent terminal. The handheld terminal is used for collecting information, and the server verifies order matching and generates an unlocking instruction when the physical position identifiers are consistent and are in a preset time window. The system establishes a virtual session to lock the injection molding machine in a single-batch mode, and the product taking-out manipulator and the label printer are linked in real time through the interlocking control module according to the session state. The algorithm engine module calculates single-mode consumption based on screw metering parameters and melt PVT characteristics so as to invert real-time inventory, and pressure and speed characteristics in the glue injection process are extracted in combination with the rheological analysis module to generate rheological fingerprints for attribute verification. According to the invention, the space-time physical interlocking of the feeding authority, the whole-process batch anchoring and the online closed-loop monitoring of the physical attributes of the raw materials are realized.

The invention discloses a vehicle-mounted liquid crystal display screen testing device, and relates to the technical field of display screen testing, and the device comprises a detection table, an automatic plugging assembly, and a pressing positioning assembly. According to the invention, through the automatic plugging assembly, the butt joint of the display power strip interface and the starting board interface can be rapidly completed, the operation is rapid and convenient, only the display screen needs to be placed on the placing board, and compared with the plugging by adopting a high-precision manipulator, various sensors and powerful algorithms are not needed, and the plugging can be realized through mechanical transmission; only one driving source of the electric push rod is needed, so that the cost is lower, the weight reduction of the equipment is facilitated, and other energy sources are not needed for driving and controlling; through the cleaning module, air injection dust removal is carried out on the flat cable interface, poor interface connection contact caused by some dust particles existing at the flat cable interface is avoided, the yield is improved, dust removal is carried out in the butt joint process, the independent dust removal time is shortened, and the detection efficiency is improved.

The utility model relates to the technical field of film bright spot product detection, in particular to a bright spot detection device of a polaroid, which comprises a working table, a feeding unit, a filter disc inspection unit, an optical positioning point identification unit, a bright spot detection unit and a discharging unit are sequentially arranged on the working table, the feeding unit comprises a moving module and a stock bin module, the moving module sucks the polaroid from the stock bin module and moves the polaroid to the filter disc inspection unit; a jig moving module of the filter disc inspection unit receives and bears the polaroid on the moving module, and a filter disc switching module of the filter disc inspection unit is used for selecting a filter disc corresponding to the polaroid; the bright spot detection unit comprises a manipulator, an adsorption jig, an adsorption jig moving module, a re-judgment camera and a warping detection camera, the manipulator sucks the polaroid on the jig moving module and carries the polaroid to the adsorption jig, and the adsorption jig moving module drives the adsorption jig to sequentially move to the re-judgment camera and the warping detection camera for shooting; and the full-automatic detection function of the bright spot film product is realized.

The invention provides a motor casing preorder processing flow method, which is applied to the technical field of motor casing processing, and comprises the following steps of: obtaining a casing, a connector and a PCB (Printed Circuit Board), and independently processing the casing, the connector and the PCB; the manipulator collaboration comprises but is not limited to a cylinder, a motor and a belt for conveying materials; processing flow positions of the casing, the connector and the PCB are determined through code scanning; the detected defective products are conveyed out of the outside from the defective belt; the shell, the connector and the PCB are independently processed and synchronously circulate; a two-dimensional code / visual system is used for tracking the position of an assembly in real time, processing paths are dynamically distributed, poor contact caused by reverse assembly is prevented through connector positive and negative recognition and marking detection, and accurate alignment of pins and connector jacks is guaranteed through PCB pin detection and position calibration. According to plasma cleaning and welding point position detection, the cleaning effect and welding point integrity are verified through a camera, the cold solder joint and offset risks are eliminated, defective products are rapidly moved out through an independent belt line, qualified products are prevented from being mixed in, and the product consistency is ensured.

The invention provides a chip feeding and discharging device and a chip testing machine. The chip feeding and discharging device comprises a mounting frame, a plurality of material boxes, a plurality of lifting mechanisms, a box changing mechanism and a manipulator. And a plurality of bins are transversely arranged on the mounting frame. The multiple material boxes can be arranged in the material bin in the vertical direction. The multiple lifting mechanisms are arranged on the multiple bins in a one-to-one correspondence mode, and each lifting mechanism comprises a lifting disc and a lifting driver. The box changing mechanism comprises a material box clamping and placing assembly and a transverse driving part. The material box clamping and placing assembly comprises a material box arm, a vertical driving part and a longitudinal driving part. The mechanical arm is arranged on the mounting frame and used for taking and placing the chips at the top of the stock bin and enabling the chips to be transferred between the material box and the detection device. The chip testing machine comprises the chip loading and unloading device and a detection device. According to the feeding and discharging device and the chip testing machine, the to-be-tested chips can be automatically taken away one by one for testing, the tested chips are automatically classified and placed, and the production efficiency is improved.

The invention belongs to the technical field of rehabilitation auxiliary equipment, and discloses a wearable exoskeleton rehabilitation manipulator which comprises a machine shell, a controller, a buffer structure, a fixing belt, a detachable movable shaft, a touch display screen, keys, a processor, a force sensor, a lithium battery, a power device, a language recognizer and a wireless communicator. By means of the detachable movable shaft design and the power device, the manipulator can simulate flexible actions of human fingers, and the grabbing function is effectively enhanced. The fixing band capable of being rapidly detached is adopted, so that the patient can easily and rapidly complete wearing and taking off of the manipulator, and the use convenience is improved. Through the connection function of the wireless communicator and external equipment, the mechanical arm can execute various auxiliary operations, and the patient is helped to improve the life self-care ability and the ability of participating in social activities.

A method for learning finger gaiting skills for multi-fingered robot hands may decompose a finger-gaiting task into shorter tasks by contact groups. The method may augment a reference trajectory for each shorter task. The method may use representation pretraining and exploration for learning.

The manipulator comprises a joint assembly, a first driving part and a transmission assembly, and the joint assembly comprises a first joint part and a second joint part rotationally connected with the first joint part; the first driving piece is arranged on the first joint piece; the transmission assembly is in transmission connection with the first driving piece and the second joint piece, the first driving piece can drive the second joint piece to rotate in the grabbing direction through the transmission assembly, and the transmission assembly is configured to be capable of cutting off power transmission with the first driving piece when the second joint piece is subjected to impact force and rotates in the grabbing direction, or cutting off power transmission with the second driving piece when the second joint piece rotates in the grabbing direction. When the second joint part is subjected to impact force and rotates in the direction opposite to the grabbing direction, the transmission assembly can cut off power transmission between the transmission assembly and the first driving part. According to the manipulator, the risk that the driving piece of the manipulator is damaged can be reduced, and therefore the use cost of the robot is reduced.