# US 2021/0276190 A1 · Method for monitoring balanced state of biped robot

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

- Number: US 2021/0276190 A1
- Document ID: US20210276190A1
- Status: Published application
- Kind: A1
- Application number: 16/914,836
- Filed: 2020-06-29
- Published: 2021-09-09
- Earliest claimed: 2020-03-06
- Assignees: Beijing Institute of Technology
- Applicants: Beijing Institute of Technology
- Inventors: Qiang HUANG; Xuechao CHEN; Zhangguo YU; Tong WU; Mingyue QIN; Qingqing LI; Libo MENG
- Web page: https://patentgrep.com/patent/US20210276190A1

## CPC

- B25J 9/1674: {characterised by safety, monitoring, diagnostic}
- B62D 57/032: with alternately or sequentially lifted supporting base and legs; with alternately or sequentially lifted feet or skid (B62D57/024 takes precedence)
- B25J 9/1605: {Simulation of manipulator lay-out, design, modelling of manipulator}
- B25J 9/163: {learning, adaptive, model based, rule based expert control}
- B25J 9/1653: {parameters identification, estimation, stiffness, accuracy, error analysis}
- B25J 13/088: {with position, velocity or acceleration sensors}
- 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})
- B62D: MOTOR VEHICLES; TRAILERS (steering, or guiding on a desired track, of agricultural machines or implements A01B69/00; wheels, castors, axles, increasing wheel adhesion B60B; vehicle tyres, tyre inflation or tyre changing B60C; connections between vehicles of a train or the like B60D; vehicles for use on rail and road, amphibious or convertible vehicles B60F; suspension arrangements B60G; heating, cooling, ventilating or other air treating devices B60H; windows, windscreens, non-fixed roofs, doors or similar devices, protective coverings for vehicles not in use B60J; propulsion plant arrangements, auxiliary drives, transmissions, controls, instrumentation or dashboards B60K; electric equipment or propulsion of electrically-propelled vehicles B60L; power supply for electrically-propelled vehicles B60M; passenger accommodation not otherwise provided for B60N; adaptations for load transportation or to carry special loads or objects B60P; arrangement of signalling or lighting devices, the mounting or supporting thereof or circuits therefor, for vehicles in general B60Q; vehicles, vehicle fittings or vehicle parts not otherwise provided for B60R; servicing, cleaning, repairing, supporting, lifting, or manoeuvring, not otherwise provided for, B60S; brake arrangements, brake control systems or parts thereof B60T; air-cushion vehicles B60V; motor-cycles, accessories therefor B62J, B62K; testing of vehicles G01M)

## Abstract

The present invention provides a method for monitoring a balanced state of a humanoid robot, comprising: acquiring state data of the robot falling in different directions and being stable, forming a support vector machine (SVM) training data set and obtaining, by training, an initial SVM classifier; inputting the state data of the robot to the trained SVM classifier, so that the SVM classifier outputs a classification result; taking statistics on a proportion of cycles judged to have an impending fall in the total number of control cycles within a judgment buffer time after the SVM classifier outputs the classification result, and finally determining a monitoring result of the balanced state of the robot according to the proportion and finally extracting state data of misjudged cycles within the buffer time, adding the state data to the current training data set and updating the SVM classifier, eventually enabling the classifier to achieve the effects of matching motion capabilities of the robot and monitoring the balanced state.

## Claims

### Claim 1

A method for monitoring a balanced state of a humanoid robot, comprising: acquiring state data of the robot falling in different directions and being stable, forming a support vector machine (SVM) training data set and obtaining, by training, an initial SVM classifier; inputting the state data of the robot to the trained SVM classifier, so that the SVM classifier outputs a classification result; taking statistics on a proportion of cycles judged to have an impending fall in the total number of control cycles within a judgment buffer time after the SVM classifier outputs the classification result, and finally determining a monitoring result of the balanced state of the robot according to the proportion and finally extracting state data of misjudged cycles within the buffer time, adding the state data to the current training data set and updating the SVM classifier, eventually enabling the classifier to match motion capabilities of the robot and realize monitoring of the balanced state of the humanoid robot.

### Claim 2 (depends on 1)

The method for monitoring a balanced state of a humanoid robot according to claim 1, wherein a method for obtaining the state data is: when the robot stands stably and is free of motion control, pushing the robot by a force in a certain direction to make it fall, capturing all state data from starting pushing to the robot falling completely to a ground to serve as a fall data set and placing a fall label; selecting a proper pushing force so that the robot gets as close to instability as possible but does not fall, also obtaining state data to serve as a stable data set and placing a stable label, and using the labelled fall data set and stable data set as the training data set of the initial SVM classifier.

### Claim 3 (depends on 2)

The method for monitoring a balanced state of a humanoid robot according to claim 2, wherein the state data comprises an angular velocity and an accelerated linear velocity of an x axis, a y axis, and a z axis of a robot body and a distance between a projection point of a center of mass of the body on the ground and a boundary of a current support surface, and forms a seven-dimensional feature vector.

### Claim 4 (depends on 3)

The method for monitoring a balanced state of a humanoid robot according to claim 3, wherein the angular velocity and the accelerated linear velocity of the x axis, the y axis, and the z axis of the robot body are obtained by estimating a motion state of the robot body using extended Kalman filtering; a current state is predicted using a system equation according to an assumption that a linear velocity and an angular velocity of a foot in contact with the ground are zero and a triaxial angular velocity and a triaxial acceleration output by an inertial measurement unit (IMU); and a prior state is updated according to a position and an attitude of the contact foot in combination with joint angle measurement and forward kinematics analysis, and angular velocity and acceleration data of the x axis, the y axis, and the z axis of the robot is finally obtained.

### Claim 5 (depends on 3)

The method for monitoring a balanced state of a humanoid robot according to claim 3, wherein a calculation method for the distance between the projection point of the center of mass of the body on the ground and the boundary of the current support surface is: calculating a position of a projection of the center of mass of the robot body on the ground according to an attitude angle of the robot, and meanwhile calculating the distance between the projection point of the center of mass of the body on the ground and the boundary of the current support surface according to a velocity direction of the robot and joint angles of a leg.

### Claim 6 (depends on 1)

The method for monitoring a balanced state of a humanoid robot according to claim 1, wherein 0.2 s after each fall judgment made by the classifier is the judgment buffer time.

### Claim 7 (depends on 6)

The method for monitoring a balanced state of a humanoid robot according to claim 6, wherein when the proportion of the cycles judged by the SVM classifier to have a fall is greater than a set threshold, it is eventually judged that a fall is about to occur; otherwise, it is still judged that stability exists.

### Claim 8 (depends on 6)

The method for monitoring a balanced state of a humanoid robot according to claim 6, wherein each time a buffer time entry condition is reached and a fall judgment condition is not reached and it is eventually judged that the robot is stable, state data in misjudged cycles within the buffer time is extracted and added to the training data set, and retraining is performed to update the SVM classifier.

### Claim 9 (depends on 2)

The method for monitoring a balanced state of a humanoid robot according to claim 2, wherein 0.2 s after each fall judgment made by the classifier is the judgment buffer time.

### Claim 10 (depends on 9)

The method for monitoring a balanced state of a humanoid robot according to claim 9, wherein when the proportion of the cycles judged by the SVM classifier to have a fall is greater than a set threshold, it is eventually judged that a fall is about to occur; otherwise, it is still judged that stability exists.

### Claim 11 (depends on 9)

The method for monitoring a balanced state of a humanoid robot according to claim 9, wherein each time a buffer time entry condition is reached and a fall judgment condition is not reached and it is eventually judged that the robot is stable, state data in misjudged cycles within the buffer time is extracted and added to the training data set, and retraining is performed to update the SVM classifier.

### Claim 12 (depends on 3)

The method for monitoring a balanced state of a humanoid robot according to claim 3 wherein 0.2 s after each fall judgment made by the classifier is the judgment buffer time.

### Claim 13 (depends on 12)

The method for monitoring a balanced state of a humanoid robot according to claim 12, wherein when the proportion of the cycles judged by the SVM classifier to have a fall is greater than a set threshold, it is eventually judged that a fall is about to occur; otherwise, it is still judged that stability exists.

### Claim 14 (depends on 12)

The method for monitoring a balanced state of a humanoid robot according to claim 12, wherein each time a buffer time entry condition is reached and a fall judgment condition is not reached and it is eventually judged that the robot is stable, state data in misjudged cycles within the buffer time is extracted and added to the training data set, and retraining is performed to update the SVM classifier.

### Claim 15 (depends on 4)

The method for monitoring a balanced state of a humanoid robot according to claim 4, wherein 0.2 s after each fall judgment made by the classifier is the judgment buffer time.

### Claim 16 (depends on 15)

The method for monitoring a balanced state of a humanoid robot according to claim 15, wherein when the proportion of the cycles judged by the SVM classifier to have a fall is greater than a set threshold, it is eventually judged that a fall is about to occur; otherwise, it is still judged that stability exists.

### Claim 17 (depends on 15)

The method for monitoring a balanced state of a humanoid robot according to claim 15, wherein each time a buffer time entry condition is reached and a fall judgment condition is not reached and it is eventually judged that the robot is stable, state data in misjudged cycles within the buffer time is extracted and added to the training data set, and retraining is performed to update the SVM classifier.

### Claim 18 (depends on 5)

The method for monitoring a balanced state of a humanoid robot according to claim 5, wherein 0.2 s after each fall judgment made by the classifier is the judgment buffer time.

### Claim 19 (depends on 18)

The method for monitoring a balanced state of a humanoid robot according to claim 18, wherein when the proportion of the cycles judged by the SVM classifier to have a fall is greater than a set threshold, it is eventually judged that a fall is about to occur; otherwise, it is still judged that stability exists.

### Claim 20 (depends on 18)

The method for monitoring a balanced state of a humanoid robot according to claim 18, wherein each time a buffer time entry condition is reached and a fall judgment condition is not reached and it is eventually judged that the robot is stable, state data in misjudged cycles within the buffer time is extracted and added to the training data set, and retraining is performed to update the SVM classifier.
