Control device, robot system, and robot control method

Abstract

A robot control device includes a flexible portion and a locking mechanism for securing the flexible portion. The control device comprises a locking control unit for controlling the locking and unlocking of the flexible portion; and a motion control unit for controlling the robot's motion using different control strategies depending on whether the flexible portion is locked or unlocked. This control device effectively controls a robot capable of switching between locking and unlocking the flexible portion.

Description

Technical Field

[0001] The present invention relates to controlling a robot having a flexible portion. Background Art

[0002] When using a robot composed solely of high-rigidity mechanisms to perform actions involving contact with an object, careful attention is required. After determining and modeling the object and the operating environment, the movement speed must be further reduced for control. Furthermore, if precision operations are to be performed using only high-rigidity mechanisms, high-precision state measurement is required, but in reality, measurement errors are present. Therefore, using a robot composed solely of high-rigidity mechanisms results in slow movement speeds and an inability to achieve the high success rate required on-site.

[0003] On the other hand, robots with flexible, deformable mechanisms can perform operations involving contact with objects at high speeds and efficiency. However, flexible mechanisms can cause instability due to vibrations during movement, making them more stable.

[0004] In order to achieve both the advantages of a rigid and flexible mechanism, robots having a flexible element with a locking mechanism are proposed in Patent Documents 1 to 3. These robots use a locking mechanism that can fix the flexible part (compliant element) at an origin or at any position.

[0005] However, Patent Documents 1 to 3 do not examine how to control when the soft portion is locked and when the soft portion is unlocked.

[0006] Prior art literature

[0007] Patent Literature

[0008] Patent Document 1: Japanese Patent Application Laid-Open No. 05-192892

[0009] Patent Document 2: Japanese Patent Application Laid-Open No. 2005-177918

[0010] Patent Document 3: Japanese Patent Application Laid-Open No. 08-118281 Summary of the Invention

[0011] Technical problem to be solved by the invention

[0012] The present invention has been made in view of the above-mentioned actual situation, and an object of the present invention is to provide a technology capable of effectively controlling a robot capable of switching between locking and unlocking of a flexible portion.

[0013] Technical solutions to technical problems

[0014] In order to achieve the above-mentioned object, the present invention adopts the following configuration.

[0015] A first aspect of the present invention is a control device for a robot, the robot having a flexible portion and a locking mechanism for securing the flexible portion, the control device comprising:

[0016] a locking control unit for controlling locking and unlocking of the soft portion; and

[0017] The motion control unit controls the motion of the robot using different control strategies according to whether the soft part is in a locked state or an unlocked state.

[0018] The term "flexible component" refers to physically flexible elements, that is, passive elements that generate a restoring force in response to displacement. Physically flexible elements typically include elastic bodies such as springs and rubber, dampers, and pneumatic or hydraulic cylinders. A flexible component can also be a mechanism that achieves flexibility through force control or solely through force control.

[0019] A control policy is an element (mapping) that determines the next action u that the robot should take when it is in a certain state x. A control policy is typically expressed in the form of a probability density function π(u|x) or a function u=π(x).

[0020] Types of control strategies include, for example, control strategies based on machine learning, control strategies using state space models based on modern control theory, and control strategies of classical control. Control strategies based on machine learning are control strategies derived from data-driven methods using machine learning algorithms, including, for example, control strategies derived through reinforcement learning, control strategies derived through deep learning, and control strategies derived through model-based reinforcement learning. Control strategies based on state space models are control strategies derived based on known information related to the controlled object, namely, system models (state space models), including control strategies based on optimal control or model predictive control. Control strategies of classical control are control strategies based on classical controls such as PID control. Control strategies based on state space models and classical control strategies can also be understood as analytical control strategies.

[0021] In this aspect, the motion control unit can be configured to control the robot's motion using a machine learning-based control strategy when the soft portion is in an unlocked (unsecured) state. When the soft portion is unsecured, the robot's tip position becomes uncertain, making it difficult to model and control using a state-space model, etc. Therefore, a machine learning-based control strategy is preferably used.

[0022] In this aspect, the motion control unit can be configured to control the robot using a classical control strategy or a state-space model-based control strategy when the soft portion is in a locked (fixed) state. When the soft portion is fixed, the robot's tip position can be accurately determined, enabling high-precision control using an analytical control strategy.

[0023] The control device of this aspect can also control the locking and unlocking of the flexible portion based on input operation instructions, and the motion control unit can switch the control strategy used to control the robot's motion. Because locking / unlocking and switching of control strategies are performed based on a single input, namely the operation instruction, the process is simple. Furthermore, depending on the operation instruction, for example, any one of multiple machine learning-based control strategies with different model parameters can be used.

[0024] Furthermore, the control device according to the present invention may also control the locking and unlocking of the flexible portion based on the distance between the robot and an object, or based on whether the robot is in contact with the object, and the motion control unit may switch the control strategy for controlling the robot's motion. When the robot approaches or contacts an object, it is preferable to unlock the flexible portion for safety reasons, and the control strategy may be switched accordingly.

[0025] Furthermore, the control device according to the present invention can determine not only the control strategy to be used but also the sensor information to be used, depending on whether the flexible portion is in a locked or unlocked state. Determining the sensor information to be used includes, for example, determining or changing the type of sensor information acquired from the robot, and determining or changing the type of sensor information input into the control strategy without changing the sensor information acquired from the robot. For example, if the flexible portion is in an unlocked state, it may be considered to determine the use of sensor information indicating the state of the flexible portion.

[0026] The soft portion in this aspect may be provided at least at any one of the midway position of a gripper of a robot, between the gripper and an arm, and midway position of the arm, wherein the robot includes a gripper for gripping an object and an arm for moving the gripper.

[0027] A second aspect of the present invention is a robotic system, comprising:

[0028] A robot comprising a gripper for gripping an object and an arm for moving the gripper, wherein the robot has a soft portion at least in one of a midpoint of the gripper, between the gripper and the arm, and a midpoint of the arm; and

[0029] The first aspect relates to a control device.

[0030] A third aspect of the present invention is a method for controlling a robot, the method being performed by a control device, wherein the robot comprises a flexible portion and a locking mechanism for securing the flexible portion, the method comprising:

[0031] a locking control step for controlling the locking and unlocking of the soft portion;

[0032] a strategy selection step of selecting a control strategy for controlling the movement of the robot according to whether the soft portion is in a locked state or an unlocked state; and

[0033] The action control step uses the selected strategy to control the action of the robot.

[0034] The present invention can also be understood as a program for implementing the above-mentioned control method or a recording medium that non-transitorily records the program. In addition, the above-mentioned units and processes can be combined with each other as much as possible to constitute the present invention.

[0035] Effects of the Invention

[0036] According to the present invention, a robot capable of switching between locking and unlocking of a flexible portion can be effectively controlled. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 This is a diagram illustrating an overview of a robot system to which the present invention is applied.

[0038] Figure 2 (A) Figure 2 (B) is a diagram showing a schematic structure of a robot.

[0039] Figure 3 This is a block diagram showing the hardware configuration of a control device of a robot system.

[0040] Figure 4 This is a diagram illustrating an example of a table showing the correspondence between work commands, the locking states of flexible parts, and control strategies.

[0041] Figure 5 It is a diagram showing a control flow of the control device. DETAILED DESCRIPTION

[0042] Application Examples

[0043] Reference Figure 1 The following describes an application example of the robot system 1 according to the present invention. The robot system 1 comprises a robot 10 having a flexible portion 13 and a control device 20 for controlling the robot 10. The flexible portion 13 of the robot 10 can be locked or unlocked. The control device 20 controls the movement of the robot 10 using different control strategies depending on whether the flexible portion 13 of the robot 10 is locked or unlocked.

[0044] Specifically, when the flexible portion 13 is unlocked, the control device 20 controls the robot 10's movements using a machine learning-based control strategy. Since the position of the tip of the robot's gripper 12 is uncertain when the flexible portion 13 is unlocked, the use of a machine learning-based control strategy allows for appropriate execution of tasks. On the other hand, when the flexible portion 13 is locked, the control device 20 controls the robot 10 using a classical control strategy or a state-space model-based control strategy. Since the position of the tip of the robot's gripper 12 can be accurately determined when the flexible portion 13 is locked, highly accurate control can be achieved using an analytical control strategy.

[0045] In this way, the control device 20 can achieve effective control when the soft part 13 is in the locked state and the unlocked state by using an appropriate control strategy according to the state of the soft part 13.

[0046] <First embodiment>

[0047] exist Figure 1 In the embodiment of the present invention, a robot system 1 includes a robot 10 and a control device 20 .

[0048] (robot)

[0049] Figure 2 (A) Figure 2 (B) is a diagram schematically illustrating the structure of a robot 10. The robot 10 in this embodiment is a six-axis vertical articulated robot, with a gripper (claw) 12 provided at the distal end of an arm 11 via a flexible portion 13. For example, the robot 10 performs insertion operations in which the gripper 12 grasps a component (e.g., a shaft, peg) and inserts it into a hole.

[0050] like Figure 2 As shown in (A), the robot 10 has an arm 11 with six degrees of freedom having joints J1 to J6. Each joint J1 to J6 is connected to each other by a motor (not shown) so that the connecting rods can rotate in the directions of arrows C1 to C6. Here, a vertical multi-joint robot is used as an example, but a horizontal multi-joint robot (SCARA robot) can also be used. In addition, a six-axis robot is used as an example, but a multi-joint robot with other degrees of freedom such as five or seven axes can also be used, or a parallel link robot can also be used.

[0051] The gripper 12 has a set of clamping portions 12a, and controls the clamping portions 12a to clamp the parts. The gripper 12 is connected to the front end 11a of the arm 11 via the soft portion 13 and moves with the movement of the arm 11. In this embodiment, the soft portion 13 is composed of three springs 13a to 13c, which are arranged so that the base of each spring forms the vertices of an equilateral triangle. However, the number of springs can be any number. In addition, the soft portion 13 can be any other mechanism as long as it generates a restoring force relative to changes in position to achieve flexibility. For example, the soft portion 13 can also be an elastic body such as a spring or rubber, a damper, a pneumatic cylinder, or a hydraulic cylinder. The soft portion 13 can also be a mechanism that uses force control or only force control to achieve flexibility. The soft portion 13 enables the front end 11a of the arm and the gripper 12 to move relative to each other in the horizontal and vertical directions by more than 5 mm, preferably more than 1 cm, and more preferably more than 2 cm.

[0052] The robot 10 includes a locking mechanism that switches between a flexible (relatively movable) state and a fixed state for the gripper 12 relative to the arm 11. The fixed positional relationship between the gripper 12 and the arm 11 can also be referred to as a state in which the flexible portion 13 is fixed or locked. Furthermore, the flexible state of the gripper 12 and the arm 11 can also be referred to as a state in which the flexible portion is not fixed or unlocked.

[0053] In this embodiment, arm 11 has movable members 11b and 11c that can move from the front end of arm 11 toward soft portion 13. These movable members 11b and 11c press soft portion 13. When soft portion 13 is pressed, the locking portion 12b of claw 12 engages with the locking portion 11d of arm 11, thereby fixing the positional relationship between claw 12 and arm 11.

[0054] The locking mechanism described above is merely a specific example; any locking mechanism may be used as long as it can secure the gripper 12 relative to the arm 11. For example, negative pressure may be applied between the gripper 12 and the arm 11, or the gripper 12 may be pulled toward the arm 11 to secure the gripper 12 relative to the arm 11. Alternatively, the gripper 12 and the arm 11 may be secured by increasing the elastic modulus of the flexible portion 13.

[0055] In addition, although the example here illustrates a configuration in which the soft portion 13 is provided between the distal end 11a of the arm 11 and the gripper 12, the soft portion 13 may also be provided midway along the gripper 12 (e.g., at a knuckle or midway along the columnar portion of a finger) or midway along the arm (e.g., at any of the joints J1 to J6 or midway along the columnar portion of the arm). Furthermore, the soft portion 13 may be provided at multiple locations among these locations.

[0056] The soft parts of the robot 10 are locked and unlocked, and the joints are operated based on commands from the control device 20. The robot 10 also has various sensors that acquire its state, and transmits the acquired sensor information to the control device 20.

[0057] (Control device)

[0058] Next, the control device 20 will be described.

[0059] Figure 3 : is a block diagram showing the hardware structure of the control device involved in this embodiment. Figure 3 As shown, the control device 20 has the same structure as a general computer (information processing device) and includes a CPU (Central Processing Unit) 31, ROM (Read Only Memory) 32, RAM (Random Access Memory) 33, storage 34, keyboard 35, mouse 36, monitor 37, and communication interface 38. These components are connected via a bus 39 so that they can communicate with each other.

[0060] In this embodiment, ROM 32 or memory 34 stores programs and data for controlling robot 10. CPU 31 is a central processing unit (CPU) that executes various programs or controls various components. Specifically, CPU 31 reads programs from ROM 32 or memory 34 and executes them using RAM 33 as a workspace. CPU 31 controls the aforementioned components and performs various computations according to the programs stored in ROM 32 or memory 34. ROM 32 stores various programs and data. RAM 33 temporarily stores programs and data as a workspace. Memory 34 is comprised of an HDD (Hard Disk Drive), SSD (Solid State Drive), or flash memory and stores various programs and data, including the operating system. Keyboard 35 and mouse 36 are examples of input devices used for various inputs. Monitor 37 is, for example, a liquid crystal display (LCD) that displays a user interface. Monitor 37 can also employ a touch panel to function as an input unit. Communication interface 38 is an interface for communicating with other devices, using standards such as Ethernet (registered trademark), FDDI, or Wi-Fi (registered trademark).

[0061] Next, the functional configuration of the control device 20 will be described.

[0062] Figure 1 FIG. 2 shows an example of the functional structure of the control device 20. Figure 1As shown in the figure, the control device 20 includes a work instruction input unit 21, an operation mode switching unit 22, a lock control unit 25, an action control unit 26, and a sensor information acquisition unit 27. The operation mode switching unit 22 includes a lock state switching unit 23 and a control strategy switching unit 24. Each functional unit is implemented by the CPU 31 reading a learning program stored in the ROM 32 or memory 34, loading it into the RAM 33, and executing it. Furthermore, some or all of the functions may be implemented using dedicated hardware devices.

[0063] The work instruction input unit 21 receives work instructions from an external device or user, indicating the work to be performed by the robot 10. The level of abstraction or granularity of work instructions can be defined arbitrarily. For example, for the shaft insertion operation, work instructions might include "grip the object with the gripper 12," "move the gripper 12 near the hole," "probe the hole and bring the shaft into contact with the hole," "make the shaft parallel to the hole," "insert the shaft into the hole," and "secure the shaft in a specified position." Work instructions can be defined at a low or high level of abstraction.

[0064] The motion mode switching unit 22 switches between the control of locking and unlocking the flexible portion 13 and the control strategy for controlling the motion of the robot 10. In this embodiment, the motion mode switching unit 22 performs these two controls based on the work command input to the work command input unit 21. Therefore, the motion mode switching unit 22 pre-stores a setting table 40 ( ) that associates the work command with the locking or unlocking of the flexible portion 13 and the type of control strategy to be used. Figure 4 ), and based on the input operation instruction and the setting table 40, the locking / unlocking of the soft part 13 and the type of control strategy to be used are determined.

[0065] Figure 4 This diagram illustrates an example of a settings table 40 stored in the action mode switching unit 22. In this example, a control strategy based on a state-space model is defined for the operation instructions for moving the gripper (Move) and for fixing the gripper or shaft (Fix). A control strategy based on machine learning is defined for the operation instructions for exploring a hole (Explore), for aligning a shaft parallel to a hole (Align), and for inserting a shaft into a hole (Insert). In settings table 40, it is also possible to define which operation strategy to use, rather than the type of operation strategy. For example, for exploration operations, a control strategy based on machine learning can be defined for exploration operations. The same applies to other operations.

[0066] The lock state switching unit 23 determines whether to set the flexible portion 13 to the locked state or the unlocked state based on the work command input to the work command input unit 21 and the setting table 40, and outputs the result to the lock control unit 25. The control strategy switching unit 24 determines the type of control strategy or the control strategy to be used by the motion control unit 26 based on the work command input to the work command input unit 21 and the setting table 40, and outputs the result to the motion control unit 26.

[0067] The lock control unit 25 outputs a command to the robot 10 to turn on or off the lock of the soft portion 13 based on the input from the lock state switching unit 23 .

[0068] The motion control unit 26 generates a motion command for instructing the robot 10 to move, and outputs the motion command to the robot 10. The motion control unit 26 generates the motion command using the sensor information (state observation data; state variables) of the robot 10 acquired from the sensor information acquisition unit 27 and the control strategy designated by the control strategy switching unit 24.

[0069] Furthermore, the motion control unit 26 does not need to use all the sensor information acquired from the sensor information acquisition unit 27 to generate motion commands. It is also possible to determine which sensor information to use based on the control strategy being used. For example, it is possible to utilize sensor information representing the state of the flexible portion 13 when a control strategy based on machine learning is used, while omitting sensor information representing the state of the flexible portion 13 when a control strategy based on a state-space model is used.

[0070] The sensor information acquisition unit 27 acquires sensor information indicating the state of the robot 10 from the robot 10. Examples of the sensor information indicating the state of the robot 10 include information acquired by encoders of each joint, visual sensors (cameras), force-related sensors (force sensors, torque sensors, tactile sensors), and displacement sensors.

[0071] (deal with)

[0072] Figure 5 This is a flowchart showing the flow of processing of a method for controlling the robot 10 performed by the control device 20 according to the present embodiment.

[0073] In step S51 , the work command input unit 21 acquires a work command from an external device or a user to instruct the robot 10 to perform a work. In step S52 , the operation mode switching unit 22 selects whether to enable or disable locking of the flexible portion and the control strategy to be used based on the work command and the setting table 40 .

[0074] In step S53, it is determined whether the lock was enabled in step S52. If the lock was enabled, the process proceeds to step S54; if the lock was disabled, the process proceeds to step S56. In step S54, the lock control unit 25 outputs an instruction to the robot 10 to disable (unlock) the lock of the flexible portion 13. In step S55, the motion control unit 26 reads the control strategy used for unlocking, namely, the machine learning-based control strategy. In step S56, the lock control unit 25 outputs an instruction to the robot 10 to enable (lock) the lock of the flexible portion. In step S57, the motion control unit 26 reads the control strategy used for locking, namely, the classical control strategy or the state-space model-based control strategy.

[0075] In step S57 , the motion control unit 26 generates a motion instruction based on the control strategy selected in step S54 or step S56 and the sensor information obtained from the sensor information acquisition unit 27 , and sends the motion instruction to the robot 10 , thereby controlling the robot 10 .

[0076] In step S58, it is determined whether the job is completed. If the job is not completed, the process returns to step S51. If the job is completed, the process ends.

[0077] (Advantageous Effects of the Present Embodiment)

[0078] Depending on the operation to be performed by the robot, there are cases where it is preferred to make the robot flexible and cases where it is preferred not to make the robot flexible. Moreover, when making the robot flexible, it is preferred to use a control strategy based on machine learning for control due to the uncertainty of the gripper position. On the other hand, when making the robot non-flexible, it is preferred to use an analytical control strategy for control due to the certainty of the gripper position. According to this embodiment, since the locking / unlocking of the soft part of the robot is controlled according to the operation to be performed by the robot, and the control strategy used for motion control is switched, the robot can be made to perform various operations efficiently. In addition, since the locking / unlocking of the soft part and the control strategy to be used are determined according to the operation instruction, the user can create a control program for the robot even if he does not explicitly specify the locking state of the soft part or the control strategy to be used.

[0079] Modifications

[0080] The above-mentioned embodiment is merely an illustrative example of the configuration of the present invention. The present invention is not limited to the above-mentioned specific embodiment, and various modifications can be made within the scope of the technical concept.

[0081] In the above embodiment, the locking / unlocking of the soft part 13 and the switching of the control strategy are performed according to the operation instruction, but as long as different types of control strategies are used depending on whether the soft part 13 is in the locked state or the unlocked state, any specific control method can be used.

[0082] For example, the locking state of the flexible portion 13 and the switching of the control strategy employed can be controlled based on the distance between the gripper 12 of the robot 10 and the object. Specifically, if the distance between the gripper 12 and the object is less than a threshold, the control device 20 controls the flexible portion 13 to an unlocked state and switches the control strategy employed by the motion control unit 26 to a machine learning-based control strategy. On the other hand, if the distance between the gripper 12 and the object is greater than the threshold, the control device 20 controls the flexible portion 13 to a locked state and switches the control strategy employed by the motion control unit 26 to a classical control strategy or a state-space model-based control strategy.

[0083] In other examples, the locking state of the soft portion 13 and the switching of the control strategy employed can also be controlled based on whether the gripper 12 of the robot 10 is in contact with an object. More specifically, if the gripper 12 is in contact with an object, the control device 20 controls the soft portion 13 to an unlocked state and switches the control strategy employed by the motion control unit 26 to a machine learning-based control strategy. On the other hand, if the gripper 12 is not in contact with an object, the control device 20 controls the soft portion 13 to a locked state and switches the control strategy employed by the motion control unit 26 to a classical control strategy or a state-space model-based control strategy.

[0084] The distance between the gripper 12 and the object or the contact detection between the gripper 12 and the object may be determined based on sensor information obtained from a sensor included in the robot 10 or a sensor observing the robot 10 .

[0085] Furthermore, the control device 20 may also determine the sensor information used by the motion control unit 26 to generate motion commands based on the control strategy being utilized. For example, the control device 20 may change the type of sensor information that the sensor information acquisition unit 27 acquires from the robot 10. Alternatively, the control device 20 may not change the sensor information acquired from the robot 10 but instead change the sensor information provided by the sensor information acquisition unit 27 to the motion control unit 26, or may change the sensor information utilized by the motion control unit 26. For example, it may be possible to use sensor information indicating the relative positional relationship between the robot 10's arm 11 and gripper 12, or force and tactile sensor information in the gripper 12, to generate motion commands when the flexible portion 13 is in the unlocked state (i.e., when utilizing a machine learning-based control strategy), while omitting this sensor information when the flexible portion 13 is in the locked state.

[0086] Appendix

[0087] 1. A control device (20) for a robot having a soft portion (13) and a locking mechanism for fixing the soft portion, the control device (20) comprising:

[0088] a locking control unit (25) for controlling the locking and unlocking of the soft portion (13); and

[0089] The motion control unit (26) controls the motion of the robot using different types of control strategies depending on whether the soft part (13) is in a locked state or an unlocked state.

[0090] 2. A control method for a robot (10), the control method being performed by a control device (20), wherein the robot (10) has a soft portion (13) and a locking mechanism for fixing the soft portion, the control method comprising:

[0091] a locking control step (S54, S56), controlling the locking and unlocking of the soft portion;

[0092] a strategy selection step (S55, S57), selecting a control strategy for controlling the movement of the robot according to whether the soft portion is in a locked state or an unlocked state; and

[0093] The action control step uses the selected strategy to control the action of the robot.

[0094] Description of Reference Numerals

[0095] 1: Robot system; 10: Robot; 11: Arm; 12: Gripper; 13: Soft part; 20: Control device.

Claims

1. A robotic system comprising: A robot having an arm, a gripper for gripping an object, and a flexible portion, wherein the flexible portion is configured to allow the robot to behave flexibly when in an unlocked state and to allow the robot to behave rigidly when in a locked state; a sensor configured to detect a distance between the robot and the object or a contact between the robot and the object; and a control device that controls the movement of the robot based on sensor information acquired from the sensor, The control device comprises: a lock control unit configured to, when it is determined based on sensor information acquired from the sensor that the distance between the robot and the object is greater than a threshold value or that the robot and the object are not in contact, switch the soft unit to a locked state so that the robot behaves rigidly, and, when it is determined based on sensor information acquired from the sensor that the distance between the robot and the object is less than a threshold value or that the robot is in contact with the object, switch the soft unit to an unlocked state so that the robot behaves softly; as well as A motion control unit, wherein the motion control unit is configured to control the motion of the robot using a control strategy based on classical control or a control strategy based on a state-space model based on sensor information obtained from the sensor when the soft part is in a locked state, and wherein the motion control unit is configured to control the motion of the robot using a control strategy based on machine learning based on sensor information obtained from the sensor when the soft part is in an unlocked state.

2. The robot system according to claim 1, wherein: The motion control unit determines sensor information used to control the motion of the robot from the sensor information acquired from the sensors according to a control strategy used by the motion control unit.

3. The robot system according to claim 2, wherein: The acquired sensor information includes sensor information indicating the state of the soft portion, If the soft part is in the unlocked state, the motion control unit determines sensor information indicating the state of the soft part as sensor information for controlling the motion of the robot.

4. The robot system according to claim 1, wherein: The control device determines sensor information acquired from the sensor of the robot according to a control strategy used by the motion control unit.

5. The robot system according to any one of claims 1 to 4, wherein: The soft portion is provided at least at any one of a midpoint of the gripper, between the gripper and the arm, and a midpoint of the arm.

6. A control method for a robot comprising an arm, a gripper for gripping an object, and a flexible portion, wherein the flexible portion is configured to allow the robot to behave flexibly when in an unlocked state and to cause the robot to behave rigidly when in a locked state, the control method comprising: a step of acquiring sensor information from a sensor, wherein the sensor is configured to detect a distance between the robot and the object or a contact between the robot and the object; a step of switching the flexible portion to a locked state so that the robot behaves rigidly when it is determined based on sensor information acquired from the sensor that the distance between the robot and the object is greater than a threshold value or the robot and the object are not in contact; When it is determined that the distance between the robot and the object is smaller than a threshold value or the robot is in contact with the object, switching the flexible portion to an unlocked state so that the robot behaves flexibly; When the soft portion is in a locked state, controlling the motion of the robot using a control strategy based on classical control or a control strategy based on a state-space model based on sensor information acquired from the sensor; as well as The step of controlling the movement of the robot using a control strategy based on machine learning based on sensor information acquired from the sensor when the soft portion is in the unlocked state.

7. A computer program product comprising a computer program, wherein the computer program is configured to cause a computer to execute the steps of the method according to claim 6.

Images