EP3551394A1

EP3551394A1 - Method for minimizing loads on the articulated joints of a manipulator

Info

Publication number: EP3551394A1
Application number: EP17780368.1A
Authority: EP
Inventors: Fan Dai; Arne WAHRBURG; Björn MATTHIAS; Hao Ding
Original assignee: ABB Schweiz AG
Current assignee: ABB Schweiz AG
Priority date: 2016-12-09
Filing date: 2017-09-29
Publication date: 2019-10-16
Also published as: WO2018103908A1; DE102016123909A1

Abstract

A method for minimizing the load on at least one joint (3) of a manipulator (1), which is intended to exert a force and/or a torque on a work object or its environment by means of a tool (2), is characterized, with regard to the problem of operating a manipulator such that its manipulation capability is as far as possible not impaired by its individual structural design and its individual performance, in that the following steps are carried out: - defining the value and/or the direction of the maximum force and/or maximum torque with which the at least one articulated joint is intended to be maximally loaded, - defining the value and/or the direction of the expected force and/or expected torque which the tool is intended to exert and/or which acts thereon, - defining at least one first spatial orientation and/or spatial position of the tool, said orientation and/or position not being intended to change during the exertion or action of the force and/or torque, and - determining a configuration of the articulated joint that the latter adopts when the tool exerts and/or experiences the defined force and/or defined torque, while maintaining the defined first spatial orientation and/or spatial position, such that the defined value and/or the defined direction of the maximum force and/or maximum torque with which the at least one articulated joint is intended to be maximally loaded is not exceeded. A manipulator (1) is configured according to this method.

Description

Method for minimizing loads on the articulated joints of a

manipulator

The invention relates to a method for minimizing the load of at least one articulated connection of a manipulator, which is intended to exert or exert a force and / or a torque by means of a tool on a work object or its surroundings. The invention further relates to a manipulator comprising at least one movable articulation and a tool.

Industrial manipulators can have different joints. For example, there are swivel joints. There, the limiting factor in manipulation tasks is often the torque that a single pivot can deliver. There are also, for example, prismatic joints, which are displacement joints. In this type of joint, the limiting factor in manipulation tasks is often the force that must support the joint. There are manipulators that have only swivel joints, and those that have only prismatic joints, and those that have both swivel and prismatic joints.

Torques that may be provided by pivot joints of a manipulator, particularly a robotic manipulator or robotic arm, often limit its ability to manipulate.

Especially with lightweight robots and manipulators for low payloads, such as a "YuMi" (brand) manipulator, low maximum torques of the hinges limit the achievable forces, particularly the contact forces, and / or torques that a robot or robotic arm exerts on the environment , in particular a work object, can exercise.

If a robot or robotic arm of this kind for manufacturing or in an arrangement overload errors of articulated joints may occur as faults. These errors typically end an operation and result in the loss of productivity. Another important application is the lifting of objects in which torques of a joint are often also limiting factors.

The invention is therefore based on the object to operate a manipulator so that its manipulation capability is not disturbed by its individual structural design and individual performance as possible disruptive. Further, it is the object of the present invention to minimize the load in a joint at a given external load on the tool, so the torque load in a rotary joint or the force load in a displacement joint.

According to the invention the above object is achieved by a method having the features of claim 1.

Thereafter, the method mentioned above is characterized in that the following steps are carried out:

Determining the amount and / or the direction of the maximum force and / or the maximum torque with which or with which the at least one articulated connection is to be subjected to maximum load,

Predetermining the amount and / or the direction of the expected force and / or the expected torque which the tool is to and / or has an effect on,

- Specifying at least a first spatial orientation and / or spatial position of the tool, which should not change during the application or application of force and / or torque, and

Determining a configuration of the articulated connection, which assumes these, when the tool in compliance with the predetermined first spatial orientation and / or spatial position, the predetermined force and / or the predetermined torque exerts and / or learns such that the specified amount and / or fixed direction of maximum force and / or the maximum torque with which or with which the at least one articulated connection is to be loaded to the maximum, is not exceeded.

A configuration of the articulated connection is to be understood here as the spatial orientation and spatial position of the articulated connection.

According to the invention, it has been recognized that a configuration of the articulated connections of a manipulator can be optimized in order to reduce the load on the articulated connections by forces and / or torques when forces and torques, in particular contact forces, and contact torques occur.

According to the invention, a new type of operation of a manipulator is proposed. This new type allows reducing selected torque loads of a connection, in particular a hinge connection, preferably with an expected static external load. In response, the forces and torques that a manipulator can exert on his environment in specific Cartesian directions are maximized based on given maximum torques of joint connections.

From an application point of view, minimizing loads on joints is usually the goal. However, the method described here is not limited to articulated joints. It allows the minimization of loads on any selected connections, in particular movable connections, of a manipulator.

Advantageously, the steps of the method are performed to determine the spatial orientations and spatial locations of multiple links relative to each other. Thus, the best possible configuration, namely spatial arrangement, of connections, in particular of articulated joints, can be produced in order to optimally exert forces and / or torques by means of a tool. The method is applicable to redundant and non-redundant manipulators.

With redundant manipulators, the configuration of the connections, in particular the articulated connections, can be optimized while maintaining a constant TCP position (orientation and position).

TCP stands for "Tool Center Point", this term combination describes in German a reference point of a mounted tool.

Usually, this reference point is the origin of a Cartesian coordinate system. The TCP position describes the possible spatial orientation and spatial position of a tool.

Against this background, it is further advantageous to specify a plurality of spatial orientations and / or spatial positions of the tool which are to be permitted during the exertion and / or application of a force and / or a torque. In this way, it can be considered that the tool, if it is to exert a force along an axis, can certainly be rotated about this axis. It can be used as degrees of freedom in the alignment of the tool in order to obtain the optimal configuration, namely spatial arrangement of the compounds.

In a non-redundant manipulator, the method may also be used if one or more components of the TCP position may change.

The method realizes a new way of using redundancy. For manipulators having seven or more degrees of freedom, an unlimited number of configurations of the connections, especially the hinges, can exist which result in the same TCP position.

The method described here resolves the redundancy by selecting a configuration of the links selected to minimize the load Connections at a given static external load leads. For example, concrete joints can be selected.

Furthermore, the degree of redundancy is increased if the TCP position is not fully specified, for example, if only certain Cartesian translations or orientations are specified. In such a situation, even a traditional six-degree-of-freedom manipulator may become redundant with respect to the specified components of the TCP position.

The best possible configuration of the connections, in particular the hinges, is automatically calculated to maximize the manipulation capability of a given manipulator, particularly a robot or robotic arm.

Academic results in the field of online redundancy resolution to minimize stress on joints are already known. By way of example, the documents [1] to [4] are mentioned at the end of the description of the figures. However, these approaches are aimed at dynamic redundancy resolution during runtime and require direct control of torque.

The steps of the method are advantageously carried out on a device, in particular a computer, which can be operated independently by the manipulator. The particularly simple method described here is based on minimizing the static torques which can be carried out completely off-line and result in optimized configuration of the connections, in particular the articulated connections, against the background of a given task.

"Offline", unlike "online", means that a program can be developed on a computer that is independent of the manipulator. During development, therefore, the manipulator can continue to operate, so that downtime of the manipulator can be avoided. As a result, a robot system having improved manipulability can be provided.

In particular, it is an increase in the maximum force and / or a torque ment of a manipulator allows, which can be applied to the environment or can, with constant maximum torques of the joints are given. This is particularly important in manipulators with relatively small torques of the articulated joints, such as in a manipulator of the type YuMi (trademark).

In manufacturing tasks, increasing the contact forces and torques exerted on the environment can be of great benefit. Furthermore, the method described here is advantageous in order to minimize the torque of joints in lifting tasks.

It is given a simplified use of manipulators, which have a non-negligible environmental contact. The user does not have to manually locate a configuration to minimize the stress on connections, especially articulations. The user is rather freed from this task. It is conceivable to offer a software to a manipulator, which executes the described method.

Advantageously, geometrical or spatial constraints can be taken into account when determining the spatial orientation and spatial position of the connection. For example, angles or distances can be specified, which must not exceed components of the manipulator. As a result, overloads or too rigid configurations can be avoided.

Preferably, a joint is used as the compound. Articulated joints are particularly frequently subjected to stress.

Further preferred as a manipulator, a robot arm is used with multiple connections, wherein the compounds are designed as articulated joints on which segments of the robot arm are connected relative to each other movable. A robotic arm must be able to work in a particularly flexible and power-optimized manner.

Advantageously, at least one compound is selected from a plurality of compounds whose load is to be minimized. So can some connections, the are particularly sensitive, preferably be spared.

Further advantageously, at least one compound of several compounds is particularly weighted when determining the spatial orientation and spatial position of the compound is performed. As a result, individual compounds which are particularly impaired or severely limited in their maximum torque can be treated preferentially

A manipulator according to the invention, which also solves the problem, comprises at least one movable connection and a tool, wherein the spatial arrangement of the connection is determined or determined by the method described here. Preferably, all connections are optimally configured.

In the drawing, Fig. 1 shows an initial configuration of the articulated joints of a

Manipulator according to a first embodiment, namely a robot arm with a tool,

FIG. 2 shows the worst case configuration of the articulated joints of the robot arm according to FIG. 1, FIG.

Fig. 3 shows the best configuration of the joints of the robot arm according to

1, 4 the initial configuration of the articulated joints of the robot arm with a tool,

Fig. 5 shows the best configuration of the joints of the robot arm according to

Fig. 4, when a rotation of the tool about the x-axis is possible, and the best configuration of the joints of the robot arm of FIG. 4, when a translational movement of the tool along the y-axis and the z-axis is possible.

Fig. 7 schematically and by way of example a manipulator in a second

embodiment,

Fig. 8 schematically and by way of example a manipulator in a third

Embodiment.

The method is based on an input-output diagram:

The intended function to minimize load on connections in a robotic system accepts the following inputs from a user: Input 1:

A desired TCP position, which is described as follows:

Input 2:

A description of which components of the TCP position should be fixed. This is encoded in a diagonal matrix as follows:

TlJ j '- ^■ i. , .

Wen inn ^r ^ ^{t 4} ,, i isstt set ddiie corresponding Cartesian direction.

\ - -

If l. 0, the corresponding direction can be modified to

Reduce stress on compounds, in particular joints. Input 3:

The expected Cartesian force loading, described by a six-dimensional vector consisting of the anticipated contact forces and torques at the TCP position. This is designated as follows:

Optionally you can enter:

The description of items to be considered to minimize the load on connections. This is encoded in the following diagonal matrix:

Λ = diag (Ä _lf , X _t > 0 Vi = 1, ..., JV.

Where N is the number of degrees of freedom and ^t specifies how much an individual connection is weighted.

By default, the weighting matrix can be proposed in the following way:

Λ = d giQßß lß)

Here, all joints are weighted equally.

The expected Cartesian force or moment load f ' ^e . ^x ^ does not have to be specified exactly. It is sufficient to indicate the direction of the vector. For simple insertion or lifting tasks, the expected displacement typically has only one component, which is not zero, namely the force in the direction of movement.

The output function includes a configuration of the connections which takes into account the specified limitations of the TCP position and minimizes the stresses of the connections according to the specified weighting at the given Cartesian displacement. To calculate the optimized position, the following optimization problem is solved: mimmize

Here, the weighted root mean square of the stress of the compound f

minimized, which results from the external load * ^ex ^ and gravity.

The sentence denotes the null space with respect to the given TCP position

TCP ^ _nd the specified predetermined directions which are coded.

x = Φ (σ)

When used to describe forward kinematic of a manipulator, the null space can be described by:

With an arbitrarily small ^c .

One set contains all permissible configurations of the connections. This takes into account a sufficient distance to angular limits of connections and singularities. That is: c ϊ ^■ c ljointlimits ^ ^ singular ities>

^~

^ Singular ities ^~ {θ. I ^ min ^ ζΐ ·

The larger the value of ^ 1, the more distance must be kept to limits of connections. Simply increases a larger value of the distance to singularities.

Solving the optimization problem according to equation (1) is generally not trivial because of its non-convexity. For one-dimensional null spaces, for example a manipulator with seven

Degrees of freedom and Ψ = lr °, becomes a simple discretized

Nullraumtraversal (line search) can be used.

For higher-dimensional null spaces, a suboptimal solution can be found by using, for example, so-called mountaineering algorithms.

Both approaches require an initial configuration of the connections

Qinit _^ which takes into account the limitations of TCP positions. Such a configuration can be any inverse kinematics solution.

It should be noted here that the optimization is solved offline to find a new destination arrangement of the connections. There is no need for optimization, which is done online.

Alternatively, the following problem can be solved:

(2) minimize max λ, - qe rrvi} i = i, ..., N τ τηαχ, ι

Here τ is max. ⁱ d _as maximum torque which erzeu- a compound gene may una and G ¹ · are row vectors corresponding to individual connections. The parameters λ ¹ 'are the diagonal elements of λ as described with reference to introduced (1). Therefore, the equation (2) is designed to minimize the load of a compound in the worst case.

Figures 1 to 6 show a manipulator 1, namely a robot arm with joints.

With reference to FIGS. 1 to 6, a method for minimizing the loading of at least one connection 3 of the manipulator 1, which exerts a force and / or a torque on a work object or its surroundings by means of a tool 2, comprises the following steps:

Determining the amount and / or the direction of the maximum force and / or the maximum torque with which or with which the connection 3 is to be loaded to the maximum,

Predetermining the amount and / or the direction of the expected force and / or the expected torque which is to be exerted and / or acted upon by the tool 2,

- Specify at least one spatial orientation and / or spatial

Position of the tool 2 to be permitted during the application or application of force and / or torque, and

Determining a spatial orientation and spatial position of the

Compound 3, which this assumes when the tool 2 exerts and / or experiences the force and / or the torque.

The steps are performed to determine the spatial orientations and spatial locations of multiple links 3 relative to one another. Several spatial orientations and spatial positions of the tool 2 are specified which are to be permitted during the exertion and / or the action of the force and / or the torque.

The steps are carried out "offline", namely on a device which is operated independently by the manipulator 1.

When determining the spatial orientation and spatial position of the connection 3, geometric or spatial constraints are taken into account.

As compound 3, a hinge connection is used. As a manipulator 1, a robot arm is used with multiple connections 3, wherein the compounds 3 are designed as articulated joints on which segments of the robot arm are connected relative to each other movable. At least one connection 3 is selected from several connections whose load is to be minimized.

At least one connection 3 of several connections is particularly weighted when the determination of the spatial orientation and spatial position of the connection 3 is performed.

The manipulator 1 comprises at least one movable connection 3 and a

Tool 2, wherein the spatial arrangement of the compound 3 is determined or determined by the method described here.

By way of example, configurations, namely spatial arrangements of connections 3, which are illustrated in FIGS. 1 to 6, are described below.

In Fig. 1 an initial configuration is shown, in Fig. 2 the worst case configuration and in Fig. 3 the best configuration of a manipulator 1, namely a robot arm.

Fig. 2 shows the minimization of the load of a connection 3, namely a hinge connection, with fully defined TCP position. The center open arrow shows the direction of the expected Cartesian shift.

In this example, the manipulator 1, namely the robot arm, has to exert a force along the x-axis of its base coordinate system with the tool 2, for example to carry out a tapping or piercing action.

The TCP position ' ^Lr is given and all Cartesian directions are fixed. That means Ψ = °. The expected Cartesian displacement is aligned only along the x-axis of the base coordinate stems.

The goal now is to minimize the root mean square load of compound 3. Therefore:

Λ = diag (0,0,0 ^, l ^, l, l, 0)

The initial configuration is shown in FIG. The quadratic load of compound 3 in this configuration at a given external

Load is = 0.2926 Newton meters.

The worst configuration with respect to the root mean square load of the connection 3 for the given external load is shown in FIG.

There applies: RMS ₌ o, 2988 Newton meters.

Fig. 3 shows the best possible configuration.

T

There applies: = 0.0787 Newton meters.

Notably, the worst case configuration of FIG. 2 is intuitive and humane but, in terms of joint loading, given a given external displacement, it comes very close to the worst case.

The optimized configuration allows reducing the root mean square load of the connection 3 substantially by 73%.

It should be noted that illustrated deviations in the TCP position may be due to imprecise calibration and they are intended to schematically describe a test arrangement.

As part of an extension, the dimensions of null space are reduced to two enlarged. In the first path, rotation about the x-axis of the base coordinate system is no longer excluded. This is reasonable for pressing, tapping or piercing tasks, since orientation with respect to the x-axis is arbitrary.

If ^ diag (l, 1,1,0, 1,1 ^ _{the new} optimized configuration is shown in FIG.

The resulting compound 3 stress mean is only

RMS = 0.0092 Newton meters, i. Compound 3 stress has been reduced by 96.86% as compared to its load in the initial configuration.

Shown in Figure 6 is another optimized configuration at a fixed TCP position where arbitrary translation along the y and z axes is possible.

The following applies: Ψ = diag (1,0,0,1,1,1)

The resulting root mean square of the stress of compound 3 is

^{Ai, J} = 0.1050 Newton meters. This again represents a significant reduction compared to the value at the initial configuration, namely a reduction of 96.40%.

Fig. 4 shows the initial configuration in this respect, in Fig. 5 a best configuration is shown allowing for rotation about the x-axis, and in Fig. 6 the best configuration is shown, with translation along the y and z - Axis is possible. FIGS. 4 to 6 show a minimization of the joint load with partially fixed TCP position. The open arrow indicates the direction of the expected Cartesian shift. The method described here could be implemented in a rapid chain of instructions of the following kind:

CalcBestConfig (TCP pose, fixed directions, Cartesian rench, [joint weighting]) This means in German language: Calculate best configuration (TCP position, fixed directions, Cartesian load, weighting of joints).

Instead of specifying the limitations of the TCP positions, either by fixing the directions or by allowing them to be modified during optimization, another constraint could apply. This limitation could set limits to the degree of change in an individual direction.

For example, a change in TCP position with respect to the z-axis of tool 2 could be limited to plus / minus 25%, rather than either setting it completely or allowing it to flow completely.

A further additional restriction could be made by limiting the arm angle in order to avoid a rigid configuration such as the optimized one according to FIG.

The approach could also be extended towards iterative online optimization. The described method according to the invention can be applied to different types of manipulators with different types of articulated joints.

The method has been explained above using the example of a manipulator 1 shown schematically and by way of example in FIG. 1, which only has pivots 3.

Likewise, the method can be applied to a manipulator 1 a shown schematically and by way of example in FIG. 7, which has only prismatic, that is to say displacement, joints 3 a, 3 a '. Furthermore, the method can also be applied to a manipulator 1b shown schematically and by way of example in FIG. 8, which has both swivel joints 3b and prismatic joints 3b '.

Tag List Manipulator

a manipulator

b manipulator

Tool

swivel

a Prismatic joint

a 'Prismatic joint

b swivel

b 'Prismatic joint

Claims

claims

1 . Method for minimizing the load on at least one connection (3) of a manipulator (1) which is intended to exert a force and / or a torque on a work object or its surroundings by means of a tool (2),

characterized in that the following steps are carried out:

Specifying the amount and / or the direction of the expected force and / or the expected torque, which is to exercise and / or act on the tool,

Determining a configuration of the articulated connection, which assumes these, when the tool in compliance with the predetermined first spatial orientation and / or spatial position, the predetermined force and / or the predetermined torque exerts and / or learns such that the specified amount and / or fixed direction of the maximum force and / or the maximum torque, with which or with which the at least one articulated joint is to be maximally loaded, is not exceeded.

2. The method according to claim 1, characterized in that the steps are carried out to determine the spatial orientations and spatial positions of several compounds (3) relative to each other.

3. The method according to claim 1 or 2, characterized in that a plurality of spatial orientations and / or spatial positions of the tool (2) are given, which are to be allowed during the exertion and / or application of force and / or torque.

4. Method according to one of the preceding claims, characterized net, that the steps are performed on a device which is independently operable by the manipulator (1).

Method according to one of the preceding claims, that geometric or spatial restrictions are taken into account when determining the spatial orientation and spatial position of the connection (3).

Method according to one of the preceding claims, characterized in that a joint connection is used as connection (3).

Method according to one of the preceding claims, characterized in that a robot arm with a plurality of connections (3) is used as the manipulator (1), wherein the connections (3) are designed as articulated connections to which segments of the robot arm are connected so as to be movable relative to each other.

Method according to one of the preceding claims, characterized in that at least one compound (3) is selected from a plurality of compounds whose load is to be minimized.

Method according to one of the preceding claims, characterized in that at least one compound (3) of a plurality of compounds is particularly weighted when determining the spatial orientation and spatial position of the compound (3) is performed.

Manipulator (1) comprising at least one movable connection (3) and a tool (2), the spatial arrangement of the connection (3) being determined or determined by the method according to one of the preceding claims.