# US 2018/0107174 A1 · Human collaborative robot system having improved external force detection accuracy by machine learning

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2018/0107174 A1
- Document ID: US20180107174A1
- Status: Published application
- Kind: A1
- Application number: 15/782,371
- Filed: 2017-10-12
- Published: 2018-04-19
- Earliest claimed: 2016-10-19
- Assignees: Fanuc Corporation
- Applicants: Fanuc Corporation
- Inventors: Shougo TAKAHASHI
- Web page: https://patentgrep.com/patent/US20180107174A1

## CPC

- G05B 13/027: {using neural networks only}
- B25J 9/163: {learning, adaptive, model based, rule based expert control}
- G05B 9/02: electric
- G05B 13/04: involving the use of models or simulators
- G05B 2219/36489: Position and force
- G05B 2219/39164: Embodied evolution, evolutionary robots with basic ann learn by interactions with each other
- G05B 2219/39165: Evolution, best performing control strategy is transmitted to other robots
- G05B 2219/40202: Human robot coexistence
- G05B: CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS (systems for controlling or regulating non-electric variables G05D; systems for regulating electric or magnetic variables G05F; control devices or systems insofar as characterised by mechanical features only G05G)
- B25J: MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES ({manipulators specially adapted for use in surgery A61B34/70; manipulators used in cleaning hollow articles B08B9/04}; manipulators associated with rolling mills B21B39/20; manipulators associated with forging machines B21J13/10; {manipulators associated with picking-up and placing mechanisms B23P19/007}; means for holding wheels or parts thereof B60B30/00; {vehicles with ground-engaging propulsion means, e.g. walking members B62D57/02, B62D57/032; devices for picking-up and depositing articles or materials between conveyors B65G47/90, B65G47/91; manipulators with gripping or holding means for transferring packages B65H67/065}; cranes B66C; {manipulators used in the protection or supervision of pipe-line installations F17D5/00; walking equipment adapted for nuclear steam-generators F22B37/006}; manipulators specially adapted for, or associated with, nuclear reactors G21C; {apparatus used for handling wafers during manufacture or treatment of semiconductor H10P72/50})

## Abstract

A human collaborative robot system having a function of detecting a force includes a human collaborative robot and a learning unit into which sensing data, internal data, and calculation data are input. The learning unit outputs a first force component applied to the human collaborative robot from outside, a second force component occurring in an operation of the human collaborative robot, and a third force component categorized as noise; and performs learning using supervised data in which inputs and correct labels obtained in advance are collected in pairs, wherein the correct labels of the supervised data are obtained by exerting a force on the human collaborative robot from outside, operating the human collaborative robot over a plurality of paths, and applying noise to the human collaborative robot, and the operation of the human collaborative robot is controlled based on the first force component output from the learning unit.

## Claims

### Claim 1

A human collaborative robot system having a function of detecting a force comprising:
a human collaborative robot for performing operation in collaboration with a human; and
a learning unit into which sensing data for calculating the force, internal data of control software for controlling the human collaborative robot, and calculation data obtained based on one of the sensing data and the internal data are input,
the learning unit outputting a first force component applied to the human collaborative robot from outside, a second force component occurring in the operation of the human collaborative robot, and a third force component categorized as noise,
the learning unit for performing learning using supervised data in which the inputs and correct labels obtained in advance are collected in pairs, wherein
the correct labels of the supervised data are obtained by exerting a force on the human collaborative robot from outside, operating the human collaborative robot over a plurality of paths, and applying noise to the human collaborative robot, and
the operation of the human collaborative robot is controlled based on the first force component output from the learning unit.

### Claim 2 (depends on 1)

The human collaborative robot system according to claim 1, wherein forces having different magnitudes and different directions are exerted on a plurality of portions of the human collaborative robot stopped in a plurality of postures, and the exerted forces and the inputs to the learning unit are collected in pairs as the supervised data of the first force component.

### Claim 3 (depends on 1)

The human collaborative robot system according to claim 1, wherein force components occurring when the human collaborative robot operates over the plurality of paths, while holding a plurality of types of works, and the inputs to the learning unit are collected in pairs as the supervised data of the second force component.

### Claim 4 (depends on 1)

The human collaborative robot system according to claim 1, wherein a noise source is applied to the human collaborative robot stopped in a plurality of postures, and fourth force components occurring by the noise source and the inputs to the learning unit are collected in pairs as the supervised data of the third force component.

### Claim 5 (depends on 2)

The human collaborative robot system according to claim 2, wherein a mode for teaching at least one of the first force component and the second force component is prepared.

### Claim 6 (depends on 2)

The human collaborative robot system according to claim 2, wherein the human collaborative robot includes:
a function of choosing a model group to be learned by the learning unit from model groups of the collected supervised data; or
a function of recommending a model group to be used in accordance with an application, from model groups of a recorded force.

### Claim 7 (depends on 1)

The human collaborative robot system according to claim 1, wherein the learning unit performs online learning to extract a feature of the detected force, and separates the detected force into the first force component, the second force component and the third force component based on the extracted feature.

### Claim 8 (depends on 1)

The human collaborative robot system according to claim 1, wherein the human collaborative robot is put in a non-collaborative operation mode in which no operation area is shared with the human, and in the non-collaborative operation mode, the learning unit performs online learning on the condition that a correct label of the first force component is set at “0”.

### Claim 9 (depends on 1)

The human collaborative robot system according to claim 1, wherein
the human collaborative robot includes a control device having a function of stopping the human collaborative robot, when the separated first force component exceeds a certain threshold value, and
after the stop of the human collaborative robot, when a first input signal indicating that the stop is caused by a false detection of the first force component by the learning unit is input, the learning unit performs online learning using input data immediately before the stop of the human collaborative robot and a correct label of an external force set at a value lower than the certain threshold value in a pair.

### Claim 10 (depends on 1)

The human collaborative robot system according to claim 1, wherein
the human collaborative robot includes a control device having a function of stopping the human collaborative robot, when the separated first force component exceeds a certain threshold value, and
after the stop of the human collaborative robot, when a second input signal indicating that the stop is caused by a normal detection of the first force component by the learning unit is input, an input signal to and an output signal from the learning unit immediately before the stop of the human collaborative robot are stored in a memory as the supervised data, and the supervised data stored in the memory is used in online learning thereafter.

### Claim 11 (depends on 10)

The human collaborative robot system according to claim 10, wherein the first input signal and the second input signal also function as operation restart signals for the human collaborative robot.

### Claim 12 (depends on 10)

The human collaborative robot system according to claim 10, wherein the second input signal is generated by the human, or using an image recognition result by a camera for imaging the human collaborative robot or a signal from an approach sensor.

### Claim 13 (depends on 1)

The human collaborative robot system according to claim 1, wherein
the human collaborative robot system includes a function of stopping the human collaborative robot when the separated first force component exceeds a certain threshold value, and a function of detecting an urgent force pattern when the human stops the human collaborative robot by application of an external force,
when the urgent force pattern is detected, the learning unit performs online learning using input data immediately before the detection of the urgent force pattern and a correct label of an external force set at a value higher than the certain threshold value in a pair.

### Claim 14 (depends on 1)

The human collaborative robot system according to claim 1, further comprising:
a plurality of human collaborative robots, wherein
each of the human collaborative robots performs learning in an independent manner, and
the human collaborative robots mutually exchange or share a plurality of results of learning independently performed by each of the human collaborative robots through a network.

### Claim 15 (depends on 1)

The human collaborative robot system according to claim 1, wherein
the first force component is an external force exerted by the human on the human collaborative robot;
the second force component is an internal force generated in the operation of the human collaborative robot; and
the third force component is noise applied to the human collaborative robot, or
the human collaborative robot performs:
storing a log of a force detected by the human collaborative robot in one cycle of a taught operation program;
storing logs of the force detected in a plurality of cycles;
applying signal synthesis or signal processing to the logs of the force detected in the cycles, to extract a force component occurring in the operation of the robot; and
performing online learning by the learning unit, using the extracted force component occurring in the operation of the human collaborative robot as a correct label.
