TWI555524B - Walking assist system of robot - Google Patents

Description

Robotic mobility aid system

The present invention relates to a robotic system, and more particularly to a mobility aid system for a robot having at least two robotic arms.

European Approved Patent Publication No. 2361734 proposes a wheelchair for use as a walking assist type robot, the transmission portion of which is provided with a drive motor and a frame mounted on the wheel. The stent can be moved up and down along the guide rails, combined with the user's exoskeleton to lower the body, increase muscle strength, or help walk or recover. However, this design is only a single action aid, but no other integration features.

Chinese Patent Publication No. 202397747 proposes an assisting elderly and assisting disabled robot, which can walk in any road condition by using the proposed triangular interlocking wheel, and its functions include visitor identification, item grabbing, assist walking, item handling and the like. However, this patent does not mention how to assist the user. If the robot is guided autonomously, it loses the meaning of the user's flexible direction change.

Chinese Patent Publication No. 202015325 proposes a multifunctional assisted walking robot with a touch-slip sensor that judges and assists the behavior or state of the user according to the signal detected by the touch-slip sensor. Although this patent can achieve the tendency of falling, Time detection, however, this design is only designed for a single mobile accessibility feature, and no other features are added to the integration.

In the past document ("Dual-arm Service Robots for Mobile Operation in Indoor Environment," in Proceeding of the 2012 IEEE International Conference on Mechatronics and Automation, Chengdu, China, 2012, pp. 1898-1903), Qi et al. The robot, which utilizes the stereo vision system and its own dual robotic arm, achieves the task of visual servoing both hands. In addition, the robot can independently learn the unknown complex home environment, establish a hierarchical map, and combine dynamic object tracking strategy algorithms to achieve dynamic navigation. However, this robot does not have the function of assisting the user to walk.

In another past document ("Development of smart mobile walker for elderly and disabled," in Proceeding of the 2013 IEEE The 22nd IEEE International Symposium on Robot and Human Interactive Communication Gyeongju, Korea, August 26-29, 2013, pp. 300- In 301), Yuk et al. proposed a motion-assisted robot design that uses a force sensor to measure the user's external force to follow the user's intention. Although this system additionally has a sitting-to-stand support mechanism, this design is designed for only one mobile assistant function, and no other functions are added to the integration.

The invention provides a motion assisting system for a robot, wherein the robot has a double mechanical arm for grasping articles, and the double mechanical arms can also be combined into a support The hand transforms the robot body into a walking aid to provide assistance for the user to walk. In this way, the same robot can have both the functions of grabbing and walking assistance.

The invention provides a motion assistance system for a robot, wherein the robot has a first robot arm and a second robot arm. The mobility assistance system includes a receiving unit, a grab mode module, a walking assist mode module, and a control unit. The receiving unit is configured to receive a command. The capture mode module is used to execute the capture mode. The walking assist mode module is used to execute the walking assist mode. The control unit is coupled to the receiving unit, and controls the grab mode module to execute the grab mode or the travel assist mode module to execute the walking assist mode according to the command. Wherein, when the capture mode module executes the capture mode, the capture mode module releases the double-loop buckle mechanism between the first robot arm and the second robot arm to release the buckle. When the walking assist mode module executes the walking assist mode, the walking assist mode module locks the two-armed buckle mechanism between the first robot arm and the second robot arm.

Based on the above, in the action assisting system of the robot of the present invention, the action assisting system can switch the operation mode in accordance with the user's demand, and execute the grasping mode or the walking assist mode. When the motion assisting system performs the grab mode, the robot can perform the task of picking objects through the first robot arm and the second robot arm. On the other hand, when the user needs the walking assistance, the mobility assistance system will switch to the execution of the walking assistance mode, at which time the buckle mechanism on the first robot arm and the second robot arm will buckle to provide user assistance. In this way, the invention can realize the functions of assisting the object delivery and the walking assistance on the same robot at the same time.

In order to make the above features and advantages of the present invention more apparent, the following is a special The embodiments are described in detail below in conjunction with the drawings.

10‧‧‧ Robot

102‧‧‧ about the arm

10a‧‧‧Local

Blocks 12, 14, 16, 18‧‧

12a, 12b‧‧ Path

20‧‧ ‧ arms buckle mechanism

220‧‧‧Right arm

222‧‧‧Right arm ring mechanism

224a‧‧‧Small cylindrical mechanism

224b‧‧‧ Large cylindrical mechanism

240‧‧‧ left arm

242‧‧‧ Left mechanical arm buckle mechanism

244a‧‧‧ small groove

244b‧‧‧ large groove

250‧‧‧ torque sensing unit

400‧‧‧Aids assisted system

405‧‧‧Image capture unit

410‧‧‧ Receiving unit

420‧‧‧Control unit

430‧‧‧Grab mode module

440‧‧‧Travel assist mode module

450‧‧‧Displacement Control Unit

460‧‧‧ odometer

503‧‧‧Feature point database

504‧‧‧Location and map building unit

505‧‧‧Extended Kalman Filter

506‧‧‧Path Tracking Control Unit

508‧‧‧ object detection unit

510‧‧‧Grab path planning unit

602‧‧‧ force sensing unit

603‧‧‧Pre-treatment unit

6032‧‧‧Low Pass Filter Unit

6034‧‧‧ force compensation unit

6036‧‧‧Coordinate conversion unit

604‧‧‧ compliant motion control unit

606‧‧‧Light ranging unit

608‧‧‧ obstacle dodge control unit

610‧‧‧Speed Fusion Control Unit

M1~M5‧‧‧Assist mode

O‧‧ Center

P‧‧‧ users

R1~R9‧‧‧ area

S1‧‧‧Image Information

S2‧‧‧In-depth information

S302~S310‧‧‧ Each step of the buckle locking process of the double-arm buckle mechanism

S312~S322‧‧‧Steps of lifting the buckle mechanism of the double-arm buckle mechanism

S5042, S5044, S5046, S512, S514, S516, S518, S520, S522‧‧

S702~S716‧‧‧Steps performed by the mobile assistant system of this experimental example

1A is a schematic diagram of a mobile assistance system of a robot in accordance with an embodiment of the present invention.

1B and 1C are schematic views illustrating changes in the shape of the robot body.

2A is a partial enlarged view of the robot of FIGS. 1B and 1C.

2B to 2D are schematic views of the buckle fastening in sequence.

3A is a schematic view of a robot release/looping double-arm buckle mechanism in accordance with an embodiment of the present invention.

3B to 3C are flowcharts showing the action of the double arm buckle mechanism according to an embodiment of the present invention.

4 is a block diagram of a mobile assistance system in accordance with an embodiment of the present invention.

5A is a block diagram of a capture mode module in accordance with an embodiment of the present invention.

FIG. 5B is a schematic diagram illustrating omnidirectional mobile simultaneous positioning and map establishment according to an embodiment of the present invention. FIG.

Figure 5C is a schematic diagram showing visual servo capture in accordance with an embodiment of the present invention.

6A is a block diagram of a walking assist mode module in accordance with an embodiment of the present invention.

6B is a schematic diagram of the operation of the walking assist mode module in accordance with an embodiment of the present invention.

Fig. 6C is a schematic view showing a walking intention estimating method according to an embodiment of the present invention.

Figure 6D is a schematic illustration of a light sensing range in accordance with an embodiment of the present invention.

Figure 7 is a flow chart showing the steps performed by the action assistance system of the experimental example.

8(a) to 11(i) are scene images executed by the motion assistance system of the experimental example.

Fig. 12 is a schematic view showing the walking path of the robot of the experimental example.

The robot of this embodiment has a first robot arm and a second robot arm. For convenience of explanation, the robot of the embodiment has a right arm and a left arm. Among them, the right arm and the left arm can be buckled and released, so as to perform different operations of the robot. Furthermore, the design of the right and left robots in this embodiment is illustrated as an example, and in other embodiments, the design of the right and left arms may also be interchanged. The above described features and advantages of the invention will be apparent from the following description.

1A is a schematic diagram of a mobile assistance system of a robot in accordance with an embodiment of the present invention. Referring to FIG. 1A, the mobile assistant system of the present embodiment has two operation modes, namely a grab mode and a walking assist mode. Wherein, the robot, the right arm and the left arm respectively have a double-arm buckle mechanism, and the robot can change the shape of the robot body by deformation of the left and right robot arms according to the needs of the user (such as the block 12). (eg block 14), to convert the operating mode of the robot, for example For example, a grab mode (such as block 16) or a travel assist mode (such as block 18), to achieve individual tasks.

For example, FIG. 1B and FIG. 1C are schematic diagrams illustrating changes in the shape of the robot body. When the user instructs the robot to take the item or deliver the item, the robot will now switch to the grab mode to achieve, wherein the left and right robot arms 102 of the robot 10 are separated, as shown in FIG. 1B. When the user needs the robot to assist, the robot changes to the walking assist mode to assist the user to walk. The left and right robot arms 102 of the robot 10 are buckled for the user to assist, as shown in FIG. 1C. In addition, the left and right robot arms 102 of the robot 10 respectively include an upper arm and a lower arm, and the degrees of freedom of the left and right robot arms 102 are respectively designed to be 6 degrees of freedom, so that the left and right robot arms 102 can flexibly rotate and grasp articles, respectively. In addition, the base of the robot 10 is provided with a four-wheel steer-and-drive platform that provides omnidirectional rotation and walking of the robot 10.

The robot's dual-arm buckle mechanism will be introduced below, and then the robot's mobility aid system will be described in detail how to adapt the robot's operation mode with the double-arm buckle mechanism.

FIG. 2A is an enlarged view of a portion 10a of the robot 10 of FIGS. 1B and 1C. Referring to FIG. 2A, the two-arm buckle mechanism 20 of the robot has a right arm ring mechanism 222 (slanted line) and a left arm ring mechanism 242 (slanted line), wherein the right arm ring mechanism 222 is disposed on On the right arm 220, the left arm ring mechanism 242 is disposed on the left arm 240.

In the walking assist mode, the robot's arm buckle mechanism 20 is deformed into A configuration similar to the grip to provide support for the user while walking. Further, the right arm ring mechanism 222 at this time is used as a buckle, and its shape is composed of two cylindrical mechanisms, that is, a small cylindrical mechanism 224a and a large cylindrical mechanism 224b. The function of the small cylindrical mechanism 224a having a small radius is to fix the right arm 220 and the left arm 240 in the X-axis direction (as shown in FIG. 2B) when combined with the left arm ring mechanism 242, so that the left and right machines are The arm (ie, the right arm 220 and the left arm 240) does not move back and forth, and the large cylindrical mechanism 224b having a larger radius is fixed for the Z-axis direction (as shown in FIG. 2B) so that the left and right arms do not cause up and down. mobile.

On the other hand, the left mechanical arm buckle mechanism 242 is provided with a groove 244 of the buckle, which is mainly designed as two semi-circular grooves, and the large groove 244b with a larger radius above is used for buffering, so that the right mechanical arm buckle The small cylindrical mechanism 224a on the mechanism 222 is nested in a relatively smooth manner into the small groove 244a of the lower radius, so that the left and right robot arms are locked in the X-axis direction. It is worth mentioning that the small groove 244a is also designed with a circular inner groove (not shown) having the same radius as the large cylindrical mechanism 224b on the right arm ring mechanism 222, that is, the inner groove. It can conform to the size of the large cylindrical mechanism 224b. The main purpose of this design is to enable the left and right robot arms to fix the large cylindrical mechanism 224b on the right mechanical arm buckle mechanism 222 with the circular inner groove by using a displacement of the lower arm parallel to the left and right mechanical arms. At this time, since the radius of the small groove 244a is smaller than the radius of the large cylindrical mechanism 224b of the right arm ring mechanism 222, the left and right arms can be fixed in the Z-axis direction to complete the concept of two-stage locking, so that the left and right machinery The arm becomes a solid auxiliary grip. 2B to 2D are a schematic view of the above-mentioned ring buckle locking, which is used to indicate that the left and right mechanical arms perform a double-arm ring fastening machine. The structure is locked, and FIG. 2D to FIG. 2B are respectively schematic diagrams of releasing the buckle to indicate that the left and right mechanical arms are restored to the state in which the arms are independently operated.

3A is a schematic view of a robot release/looping double-arm buckle mechanism in accordance with an embodiment of the present invention. FIG. 3B and FIG. 3C are flowcharts showing the operation of the double-armed buckle mechanism according to an embodiment of the present invention, wherein steps S302-S310 represent a loop fastening process of the double-arm buckle mechanism, and steps S312-S320 represent a double-armed ring. The buckle mechanism of the buckle mechanism.

Referring to FIG. 2A, FIG. 3A and FIG. 3B, in step S302, the right robot arm 220 and the left robot arm 240 are separated from each other. In step S304, the mobility assistance system plans the movement path of the right robot arm 220 and the left robot arm 240 according to a preset buckle buffer point. In detail, the present embodiment is based on the right robot arm 220, wherein the preset buffer point is the lower edge of the cylindrical mechanism disposed at the right arm ring mechanism 222. Here, since the force and torque sensing unit 250 (described later in detail) of the present embodiment is designed above (but not limited to) the right robot arm 220, the buffer point is set in consideration of simplifying the calculation of the biasing force. This position is set in a positive L-shaped manner in which the right arm 220 is 90 degrees above and below the arm. The left arm 240 is parallel to the right arm 220, so that the large groove 244b of the left arm ring mechanism 242 is aligned with the small cylindrical mechanism 224a of the right arm, and the movement mode is a way of setting the card coordinate point as a path point. The reverse kinematics of the left and right robot arms are moved to reach the buckle buffer point when the left and right robot arms do not touch each other.

Then, as shown in step S306, the mobility assistance system performs the adjustment of the buckle lock. In detail, the mobility assistance system is based on the right mechanical arm buckle mechanism 222 and The mutual position of the left mechanical arm buckle mechanism 242 defines the relative position of the buckle locking points, wherein the mobility assistance system utilizes the respective degrees of angle to control the fine-tuning action. This action can be divided into two steps. First, the right mechanical arm buckle mechanism 222 is directly above the left mechanical arm buckle mechanism 242, so the mobility assistance system first adjusts the left robot arm 240 so that the left and right robot arms are The X-axis direction is fixed (step S307), and the right arm ring mechanism 222 and the left arm ring mechanism 242 are fine-tuned so that the left and right arms are fixed in the Z-axis direction (step S308). Of course, this embodiment does not limit the sequence of steps S307 and S308. At this time, the action assisting system completes the buckle locking of the left and right robot arms (step S310)

As for the process of releasing the buckle, please refer to FIG. 3A and FIG. 3C. The manner in which the motion assisting system releases the buckle is opposite to the flow of the locking system of the motion assisting system, that is, first, the left and right arms are locked by the buckle. (Step S312). Next, the mobility assistance system performs adjustment of the buckle release lock (step S314), including unlocking the left and right robot arms in the Z-axis direction (step S315) and unlocking the left and right robot arms in the X-axis direction (step S316). ). Similarly, the present embodiment does not limit the order of steps S315 and S316. After that, the mobility assistance system will plan the road to separate the left and right robot arms (step S318), so that the left and right robot arms return from the buckle lock state to the release buckle state, that is, the left and right armes are separated back. Status (step S320).

The robot's mobility aid system will be described in detail below to match the armrest mechanism to convert the robot's operating mode.

4 is a mobile assisted system block in accordance with an embodiment of the present invention. Figure. Referring to FIG. 4, the mobile assistant system 400 includes a receiving unit 410, a control unit 420, a grab mode module 430, and a walking assist mode module 440. The functions of these units and modules are described in detail below.

The receiving unit 410 can receive a command issued by the user, and cause the control unit 420 to execute the grab mode module 430 or the walking assist mode module 440. In the present embodiment, the command issued by the user is, for example, a command for causing the robot to execute the grab mode, or a command for the robot to execute the walking assist mode. The command can be presented in various forms. For example, the user can call the robot to capture the item or provide an auxiliary function by voice, and the robot can transmit the voice of the user to the receiving unit 410 through the voice recognition module, so that the control unit 420 performs Corresponding operation. Alternatively, the user can send a signal to the receiving unit 410 through an electronic device (for example, a tablet computer, a mobile phone, a computer, etc.), thereby causing the control unit 420 to perform a corresponding operation.

The control unit 420 is coupled to the receiving unit 410. The control unit 420 can be a functional module implemented by hardware and software. The hardware may include a hardware device having a computing function such as a central processing unit, a chipset, a microprocessor, or a combination of the foregoing hardware devices, and the software may be an operating system, a driver, or the like. Here, the control unit 420 can receive the command received by the receiving unit 410 to control the capture mode module 430 or the walking assistance mode module 440 to perform a corresponding operation. That is, when the receiving unit 410 receives the command issued by the user to execute the capture mode, the control unit 420 can start executing according to the command capture mode module 430; on the other hand, when the receiving unit 410 receives the user The command to execute the walking assist mode is issued, and the control unit 420 can start executing the walking assist mode module 440 according to the command.

For example, when the signal issued by the user is a command to execute the capture mode, wherein the command may include information such as the position of the object and the object, the capture mode module 430 starts to execute, so that the robot can walk according to the command. Go to the position of the target object to grab the target object. On the other hand, when the signal issued by the user is a command to execute the walking assist mode, the walking assist mode module 440 starts to execute, so that the robot buckles the arms of the robot according to the command to provide the user's support function. Moreover, the robot can simultaneously adjust the moving speed of the robot according to the speed at which the user walks and the surrounding environment, thereby assisting the user to walk.

The operation of the grab mode module 430 and the walking assist mode module 440 will be described in detail below with reference to FIGS. 5A-6B.

When the command issued by the user is a command to execute the capture mode, the capture mode module 430 starts to execute, and at this time, the left and right robot arms of the robot are switched to release the buckle state.

FIG. 5A is a block diagram of a capture mode module 430 in accordance with an embodiment of the present invention. Referring to FIG. 5A first, the capture mode module 430 includes a positioning and map establishing unit 504, a path tracking control unit 506, an object detecting unit 508, and a grab path planning unit 510. In addition, the capture mode module 430 can also be coupled to the image capture unit 405, the displacement control unit 450, and the odometer 460. For convenience of description, the operation of the capture mode module 430 can be roughly divided into two parts, namely, omnidirectional movement simultaneous positioning and map establishment and visual capture, but the embodiment is not limited to these two parts. In this embodiment, the capture mode module 430 is based on the image capturing unit 405, the positioning and map establishing unit 504, the path tracking control unit 506, and the object detecting unit. 508 and the grab path planning unit 510, the displacement control unit 450, and the odometer 460, establish map information of the robot and the position of the robot in the map information, then move to the target object, and detect the angle of the target object. In order to determine the grabbing path of the target object, and finally the left and right robot arms of the robot perform the grabbing action, as described in detail below.

The operation of omnidirectional mobile simultaneous positioning and map establishment: FIG. 5B is a schematic diagram illustrating omnidirectional mobile simultaneous positioning and map establishment according to an embodiment of the present invention. 5A and 5B, the image capturing unit 405 is configured to capture the image information S1 and the depth information S2, wherein the image capturing unit 405 is, for example, a Kinect RGB-D camera or other electronic device having an image capturing function. And for example, it is placed at the head of the robot. Then, the positioning and map establishing unit 504 receives the image information S1 and the depth information S2 captured by the image capturing unit 405, so as to achieve omnidirectional movement simultaneous positioning and map establishment according to the image information S1 and the depth information S2 (Simultaneous) Localization and Mapping, SLAM). In detail, the positioning and map establishing unit 504 can use the Speeded Up Robust Features (SURF) algorithm to extract multiple feature points of the environment from the RGB-D information of the image information S1 and the depth information S2 (steps) S5042), and comparing the feature points according to the data in the feature point database 503 (step S5044), to calculate the three-dimensional feature position, and transforming the feature point with the robot state through the conversion of the coordinate system, thereby estimating the robot The posture (step S5046). Since the motion model and the vision system of the robot are nonlinear, the positioning and map establishing unit 504 also utilizes an extended Kalman filter (Extended Kalman Filter, EKF) 505 combines image information and predicts the status of the system. Further, the extended Kalman filter 505 corrects the robot state vector by using the state vector and the feature point state vector estimated by the odometer 460 and the observed feature point state vector as inputs at the previous moment. The feature point database 503 is updated, and the robot gesture is output to the path tracking control unit 506. Thereafter, the path tracking control unit 506 outputs a motion command to the displacement control unit 450 according to the robot attitude, and the displacement control unit 450 will control the movement of the robot according to the motion command.

Operation of Visual Grab: FIG. 5C is a schematic diagram illustrating visual servo capture in accordance with an embodiment of the present invention. Referring to FIG. 5A and FIG. 5C, the object detecting unit 508 uses the image information S1 captured by the image capturing unit 405 as an input, cuts the image of the environment, and cuts the image image of the image information into XY. The plane and the YZ plane (step S512). The XY plane and the YZ plane after the cutting can not only describe the objects in the environment, but also the images obtained by the image capturing unit 405 to find and compare the feature points in the whole picture, and reduce the XY plane and the YZ plane after the comparison. The feature points inside are therefore more likely to speed up the comparison of the target object. In addition, the object detecting unit 508 can also define a safety area around the position where the target object is located according to the feature points in the XY plane and the YZ plane after cutting (step S514), in other words, the robot can appropriately move in the safe area to grasp Take the target object without touching other non-target objects. On the other hand, the object detecting unit 508 also compares the image observed by the robot with the feature points of the target object in the feature point database 503 to detect the target object (steps) S516), thereby finding a reference point that can represent the target object and detecting the posture of the target object. In this way, the object detecting unit 508 can detect the grab angle of the target object according to the posture of the target object (step S518). Finally, the grab path planning unit 510 can plan the path of the grab target object according to the output of the object detecting unit 508 (step S520), thereby generating a control command for controlling the left and right robot arm to grab the target object (step S522). . In addition, in the above steps S512-S522, the capture mode module 430 continuously corrects the relative position of the image capturing unit 405 and the target object, so that the claw portions of the left and right robot arms finally reach the predetermined posture and position and grab. The target object, in order to complete the task of autonomous capture.

When the command issued by the user is a command to execute the walking assist mode, the walking assist mode module 440 starts to execute, and the left and right robot arms of the robot are changed from the state in which the buckle is released to the state in which the buckle is locked.

FIG. 6A is a block diagram of a walking assist mode module 440 in accordance with an embodiment of the present invention. FIG. 6B is a schematic diagram of the operation of the walking assist mode module 440 in accordance with an embodiment of the present invention. Referring to FIG. 6A and FIG. 6B, the walking assistance mode module 440 may include a force sensing unit 602, a compliance motion control unit 604, a light ranging unit 606, an obstacle dodge control unit 608, and a speed fusion control unit 610, wherein walking The auxiliary mode module 440 can also be coupled to the displacement control unit 450 and the odometer 460.

The force sensing unit 602 is configured to sense force information applied to the left and right mechanical arms on the robot, such as the sensed force and moment, wherein the force sensing unit 602 is disposed at the intersection of the right arm of the robot and the elbow thereof. At the office. In this embodiment, the robot The movement is based on the user's urging force and the torque sensed by the user as the input of the judgment data. When the force sensing unit 602 senses the force and the moment to obtain the force information, in order to consider the stability and accuracy of the judgment data. The walking assistance mode module 440 can also perform pre-processing of the force information through a force pre-processing unit 603. The force pre-processing unit 603 includes a low-pass filtering unit 6032, a force compensation unit 6034, and a coordinate conversion unit 6036.

In detail, the low-pass filtering unit 6032 performs a filtering operation on the force data sensed by the force sensing unit 602. Since the force sensing unit 602 is disposed at the intersection of the upper and lower arms of the right arm of the robot, the force information sensed by the force sensing unit 602 is affected by the weight of the lower arm, so the force pre-processing unit 603 also transmits the force. The compensation unit 6034 causes the lower arm weight to cause the initial gravity deduction of the force sensing unit 602 to facilitate clear division of the user's force determination.

Since the robot is locked in the buckle state, the left and right robot arms of the robot are nearly 90 degrees flat with the robot itself, and the upper arm and the lower arm are 90 degrees to interact with the chest of the robot, so the force pre-processing unit The 603 also derives the force conversion matrix through the coordinate conversion unit 6036 to convert the force information from the coordinate system sensed by the force sensing unit 602 into the coordinate system of the robot itself, and then applies the converted force information to the compliance. The motion control unit 604 serves as a judgment of the user's intention.

Since the force sensing unit 602 senses the force applied by the user on the right arm, and when the user pushes back and forth on the left and right arms of the robot, the torque is generated, and the users Force or torque is easy to affect the use The walking or turning intention is determined. Therefore, the compliance motion control unit 604 of the walking assist mode module 440 also performs the walking intention estimation, which is, for example, the difference of the three force information (ie, the coordinate system of the robot itself). The amount of change in force in the X-axis direction and the Y-axis direction and the amount of change in torque are used as inputs to calculate and judge the intention of the user. For convenience of description, the embodiment divides the user's intention into five kinds of auxiliary modes, namely, straight forward, horizontal movement, right oblique 45 degree angular movement, left oblique 45 degree angular movement, and in situ rotation. It should be noted that the present embodiment does not limit the five auxiliary modes, and in other embodiments, it may be divided into more or less auxiliary modes.

Fig. 6C is a schematic view showing the above-described walking intention estimating method according to an embodiment of the present invention. Please refer to FIG. 6C, _{ΔF wx} (that is, the amount of change of the force in the X-axis direction of the coordinate system of the robot itself), ΔF _xy (that is, the amount of change in the force in the Y-axis direction of the coordinate system of the robot itself), and Δ τ _wz (ie, The coordinate value of the coordinate of the robot itself is the difference between the current force applied value of the user and the initial value of the force sensing unit that detects the user's unapplied force. It can be judged by three differences of the formula (1). Three inputs.

_{△ F wx = F wx (k} ) -F initial_x △ F wy = F wy (k) -F initial_y △ τ wz = τ wz (k) - τ initial_z formula (1)

Where F _wx(k) , F _wy(k) and τ _wz(k) are the values sensed by the current force sensing unit 602, and F _{initial_x} , F _{initial_y} and τ _{initial_z} are the force sensing unit 602 not yet applied by the user. The initial value sensed by force.

Specifically, when _{ΔF wx} >0, _{ΔF wy} =0, the compliance motion control unit 604 determines that the robot is straight forward (auxiliary mode M1), left oblique 45 degree angular movement (auxiliary mode M3), right oblique 45 Degree of angular movement (auxiliary mode M4) or rotation (auxiliary mode M5). At this time, the compliance motion control unit 604 further uses the magnitude of Δ τ _wz to determine and standardize the possible actions to further determine whether the robot moves or rotates the straight forward, left oblique/right oblique 45 degrees. When _{ΔF wx} =0 and _{ΔF wy} >0, the compliance motion control unit 604 is defined as horizontal movement (auxiliary mode M2).

In addition, the compliant motion control unit 604 also converts the applied force into a corresponding speed command. When the user applies a thrust to the auxiliary robot in a certain direction, the robot can adjust the motion according to the magnitude and direction of the thrust. To comply with the user's actions. In this embodiment, the compliance motion control unit 604 can calculate the auxiliary speeds V _wx , V _wy , ω _wz that the robot needs to conform to through the formula (2), where M _robot is the mass coefficient and I _robot is the moment of inertia coefficient. , and B _robot is the damping coefficient.

On the other hand, in order to allow the robot to evade obstacles without collision during the movement, the optical ranging unit 606, for example, a laser range finder, can scan and detect the surroundings of the laser with laser light. The object obtains at least one distance sensing value between at least one adjacent object and the robot. Next, the obstacle dodge control unit 608 outputs the rotational speed change amount of the robot based on the distance sensed value.

In particular, Figure 6D is a schematic illustration of a light sensing range in accordance with an embodiment of the present invention. Referring to FIG. 6D, the area scanned by the optical ranging unit 606 is, for example, a range in which the laser light is scanned from 0 degrees to 180 degrees, for example, every 20 degrees is divided into one area, so the scanned area is divided into 9 areas. R1~R9, where the center O indicates the position of the robot. The distance information used by the designer when evading the obstacle is from the region R3 to R7, and the distance information directly in the regions R4, R5, and R6 is the output value for determining the rotational speed of the robot. Since the obstacle is closer to the obstacle when the obstacle is closer to the robot, the distance information in the right front region R3 and the left front region R7 can be used to know which side is needed on the right side and the left side. A larger dodge space is used to determine the direction of advancement of the robot.

In addition, in order to avoid the robot encountering complex environments, such as obstacles on the left and right sides, and the distance is very close, the right side is detected at the last moment due to obstacles on the left side, but the right side is detected at the next moment. There is an obstacle, which causes the robot to swing left and right. Therefore, the obstacle dodge control unit 608 sets the robot when the distance on the left side is greater than the distance on the right side by a predetermined distance (for example, 30 cm). Choose to turn left, otherwise let the robot choose to turn right. In other words, this preset distance is used to increase the stability of the performance of the dodging obstacle.

In order to take into account the relationship of the rotation angle of the robot itself, the obstacle dodge control unit 608 defines a minimum collision distance L _b and a collision warning distance L _d . In detail, when the area in front of the robot R4, R5 and R6 are the distance to the obstacle is less than L _min the smallest miss distance L _b, then the angular velocity ω _obs robot shall be the maximum rotational angular velocity _ω max_obs dodge, and when the front of the robot is between L _b and L _d and L _min is between the distance to the obstacle, places of formula formula (3) to set the desired angular velocity of the robot ω _obs. Further, as for the dodging in the left and right direction, the obstacle dodging control unit 608 can determine by the equation (4), wherein L _R3 and L _R7 respectively indicate the distances in the right front region R3 and the left front region R7. Equations (3) and (4) define a counterclockwise rotation to be a positive value and a clockwise rotation to a negative value.

When the walking assist mode module 440 performs the walking assist mode, the speed of the robot includes the values output by the compliance motion control unit 604 and the obstacle dodge control unit 608 in order to assist the robot walking assist behavior required by the time environment and the user's movement. .

In detail, the speed fusion control unit 610 fuses the angular velocity ω _wz output by the compliance motion control unit 604 and the angular velocity ω _obs output by the obstacle dodge control unit 608 to form the rotational angular velocity of the robot, and the speed fusion control unit 610. The method of fusion is to multiply a weight value by an individual angular velocity, as shown in equation (5).

ω _f =( W _obs × ω _obs )+((1- Wobs )× ω _wz ) (5)

Where W _obs is the weight value of the obstacle dodge part, (1-W _obs ) is the weight value of the compliant motion part, and ω _f is the combined rotation angular velocity.

Further, the weight value W _obs is determined according to the minimum collision distance L _b and the collision alert distance L _d . Specifically, when the distance between the front of the robot and the obstacle L _{min is} less than the minimum collision distance L _b , the representative robot is too close to the obstacle at this time, so the robot must avoid the obstacle with the maximum avoidance angle, that is, the weight value W _obs Is set to 1. When the distance between the front of the robot and the obstacle L _{min is} between the minimum collision distance L _b and the collision warning distance L _d , the weight value W _obs is adjusted by the distance between the robot and the obstacle. Further, when the robot with an obstacle front collision warning distance L _min is greater than the distance L _d, then the robot will be intended to command a user to do a control body motion commands, i.e. the weight W _obs value is set to zero. The definition formula of this weight value is as shown in equation (6).

In addition, the linear velocity output by the speed fusion control unit 610 is the output of the compliance motion control unit 604 (ie, the auxiliary speeds V _wx and V _wy ), so the fusion speed V _m after the fusion of the speed fusion control unit 610 is as shown in equation (7). Said.

In this way, when the walking assist mode module 440 executes the walking assist mode, the speed of the robot will travel according to the fusion speed V _m merged by the speed fusion control unit 610, so that the robot not only provides the function of assisting the user to walk, At the same time, it also has the function of obstacle avoidance to prevent users from falling due to touching obstacles.

The grasping mode and the walking assist mode included in the mobility assisting system of the robot of the present embodiment are described below with a home situation as an experimental example.

Figure 7 is a flow chart showing the steps performed by the action assistance system of the experimental example. 8(a) to 11(i) are scene images executed by the motion assistance system of the experimental example. Referring to FIG. 7, first, the double arm buckle mechanism of the robot is in a state of releasing the buckle (step S702), wherein the left and right mechanical arms are normally placed, for example, returning to the preset posture. After the robot receives the command issued by the user (for example, the order to pour the beverage from the beverage can and deliver the beverage), as shown in Figures 8(a) to 8(b), the robot utilizes the establishment of the room map and self-positioning, and The navigation of the target object is started autonomously (step S704). In this experimental example, the target object is, for example, a beverage can and a cup on a table, so that the robot is, for example, moved to the front of the table. It should be noted that the robot will also detect obstacles during the movement, and when there are obstacles during the movement, as shown in Fig. 8(c) to Fig. 8(d), the robot depends on the laser ranging. The instrument performs the dodge of the obstacle.

When the robot moves to the front of the table, as shown in Fig. 8(e) to Fig. 8(g), the robot detects the target object (i.e., the beverage can and the cup) (step S706), for example, using a Kinect sensor. Take environmental image information, determine the position and posture of the beverage can, use the left robotic arm to grab the beverage can in the environment, and then determine the position of the cup. And the posture, using the right robot arm to grab the cup (step S708), and then, as shown in FIG. 8(h), the double robot arm pours the beverage in the left hand beverage can into the right hand through the previously planned pouring beverage path. Inside the cup. When the beverage is poured, as shown in Fig. 8(i), the robot puts the beverage can back on the table.

Next, as shown in FIGS. 9(a) to 9(f), the robot navigates to the position of the user (step S710), and delivers the cup to the user (step S712). At this time, the user wants to assist the robot to walk and gives a command corresponding to the walking assist mode, and the robot switches to the walking assist mode. At this time, as shown in FIG. 10( a ) to FIG. 10( c ), the robot combines the left and right mechanical arms by using the loop path planning strategy, so that the two-armed buckle mechanism of the left and right mechanical arms is in the buckle locking mode (step S714). Therefore, the user can use the left and right robot arms of the robot as the armrests of the mobility aid.

Next, as shown in FIGS. 11(a) to 11(i), the robot can obtain the user's urging force through the force/torque sensor, estimate the walking intention, and assist in walking. In the assisting process, if the user does not notice that there is an obstacle on the walking path, the robot can assist the user to safely walk out of the room by performing a dodging action on the obstacle by the laser range finder (step S714). After that, when the user does not walk through the left and right robot arms of the robot, the double arm buckle mechanism on the left and right robot arms of the robot can automatically return to the release buckle mode (step S702). Of course, in other experimental examples, The double-arm buckle mechanism on the left and right arms of the robot can be maintained in the buckle lock mode.

It should be noted that, in the above experimental example, after the robot delivers the cup to the user, if the user does not issue a command corresponding to the walking assist mode, then The double arm buckle mechanism on the left and right robot arms of the robot can be maintained in the release buckle mode (that is, returning from step S712 to step S702). In addition, in the above experimental example, the walking path of the robot during the whole experiment can be as shown in FIG. 12, wherein the point distribution is the feature point of the object in the environment, and the path 12a corresponds to the walking of the robot in steps S704 to S714. The path 12b is a path corresponding to the walking of the robot in steps S712 to S716. As can be seen from Fig. 12, the robot can clearly locate the location of the environment, move to the target location, and can also avoid obstacles during the movement.

In summary, in the motion assisting system of the robot of the present invention, the robot has a first robot arm and a second robot arm, and the motion assisting system can switch the operation mode according to the needs of the user, by changing the first robot arm and The configuration of the second robot arm performs a grab mode and a walking assist mode. When the motion assisting system performs the grab mode, the robot can perform the task of picking objects through the first robot arm and the second robot arm having six degrees of freedom. On the other hand, when the user needs walking assistance, the mobility assistance system converts to perform the walking assistance mode, and the left and right mechanical arm buckles are locked by the path planning strategy to form the mobility aid grip to provide user support. In this way, the invention can realize the functions of assisting the object delivery and the walking assistance on the same robot at the same time.

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the present invention, and any one of ordinary skill in the art can make some changes and refinements without departing from the spirit and scope of the present invention. The scope of the invention is defined by the scope of the appended claims.

400‧‧‧Aids assisted system

410‧‧‧ Receiving unit

420‧‧‧Control unit

430‧‧‧Grab mode module

440‧‧‧Travel assist mode module

Claims (10)

A motion assistance system for a robot, wherein the robot has a first robot arm and a second robot arm, the motion assistance system includes: a receiving unit that receives a command; and a grab mode module that performs a grab a walking assist mode module for performing a walking assist mode; and a control unit coupled to the receiving unit, and controlling the grab mode module to execute the grab mode or the walking assist mode according to the command The module executes the walking assist mode, wherein when the grab mode module executes the grab mode, the grab mode module has a two-arm loop mechanism between the first robot arm and the second robot arm And releasing the buckle, wherein when the walking assist mode module executes the walking assist mode, the walking assist mode module performs the double-arm buckle mechanism between the first robot arm and the second robot arm The buckle is locked. The motion assistance system of the robot of claim 1, wherein the first robot arm and the second robot arm have a degree of freedom of 6, respectively. The motion assistance system of the robot of claim 1, wherein the first robot arm or the second robot arm is provided with a force sensing unit. The mobile assistant system of the robot of claim 1, wherein the robot head is provided with an image capturing unit. A motion assisting system for a robot as described in claim 1 of the patent scope, The capture mode module includes: a positioning and map establishing unit, receiving an image information and a depth information captured by an image capturing unit, and performing an omnidirectional movement according to the image information and the depth information; Positioning and map creation. The action assistance system of the robot of claim 5, wherein the capture mode module further comprises a path tracking control unit, wherein the positioning and map establishing unit adopts an accelerated robust feature algorithm, from the image information Extracting a plurality of feature points from the depth information, and comparing the feature points according to a feature point database to estimate the posture of the robot, and using the state vector of the robot estimated by a previous moment Observing a feature point state vector, correcting the state vector of the robot and updating the feature point database, and simultaneously outputting the posture of the robot to the path tracking control unit, wherein the motion command is output according to the posture of the robot The displacement control unit causes the displacement control unit to control the movement of the robot according to the motion command. The action assistance system of the robot of claim 5, wherein the capture mode module further comprises: an object detection unit, according to the image information captured by the image capture unit The image is cut, and a safe area to which the target object belongs is defined according to the plurality of second images after cutting, and a grab angle of the target object is detected according to the target object; and a path planning unit is captured according to the An output of the object detecting unit, planning to capture a path of the target object to generate and control the first robot arm and the second robot arm A control command to grab the target object. The mobility assistance system of the robot of claim 1, wherein the walking assistance mode module comprises: a force sensing unit for sensing the first mechanical arm and the first force applied to the robot a force information of the second robot arm; a compliance motion control unit, based on the force information, calculating an auxiliary speed that the robot needs to conform; and an optical ranging unit detecting at least one of the at least one adjacent object and the robot a distance sensing value; an obstacle dodge control unit that outputs an angular velocity required by the robot according to the distance sensing value; and a speed fusion control unit that outputs the auxiliary speed and the obstacle by the compliance motion control unit The angular velocity output by the dodge control unit is fused to a rotational angular velocity of the robot. The action assistance system of the robot of claim 8, wherein the walking assistance mode module further comprises: a force pre-processing unit having a low-pass filter unit, a force compensation unit, and a standard conversion unit, wherein The low-pass filtering unit performs a filtering action on the force data sensed by the force sensing unit, and the force compensation unit deducts the weight of the lower arm of one of the first robot arm or the second robot arm, and the coordinate The conversion unit converts the coordinate system on the force sensing unit into a coordinate system of the robot itself. The action assistance system of the robot of claim 8, wherein the compliance motion control unit calculates and determines a user's intention based on the force information sensed by the force sensing unit, and the user's The intention is to divide into multiple auxiliary modes.