CN114684293B - Robot walking simulation algorithm - Google Patents
Robot walking simulation algorithm Download PDFInfo
- Publication number
- CN114684293B CN114684293B CN202011579027.2A CN202011579027A CN114684293B CN 114684293 B CN114684293 B CN 114684293B CN 202011579027 A CN202011579027 A CN 202011579027A CN 114684293 B CN114684293 B CN 114684293B
- Authority
- CN
- China
- Prior art keywords
- robot
- training
- gait
- data
- biped
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B62—LAND VEHICLES FOR TRAVELLING OTHERWISE THAN ON RAILS
- B62D—MOTOR VEHICLES; TRAILERS
- B62D57/00—Vehicles characterised by having other propulsion or other ground- engaging means than wheels or endless track, alone or in addition to wheels or endless track
- B62D57/02—Vehicles characterised by having other propulsion or other ground- engaging means than wheels or endless track, alone or in addition to wheels or endless track with ground-engaging propulsion means, e.g. walking members
- B62D57/032—Vehicles characterised by having other propulsion or other ground- engaging means than wheels or endless track, alone or in addition to wheels or endless track with ground-engaging propulsion means, e.g. walking members with alternately or sequentially lifted supporting base and legs; with alternately or sequentially lifted feet or skid
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1602—Programme controls characterised by the control system, structure, architecture
- B25J9/1605—Simulation of manipulator lay-out, design, modelling of manipulator
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1628—Programme controls characterised by the control loop
- B25J9/163—Programme controls characterised by the control loop learning, adaptive, model based, rule based expert control
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1656—Programme controls characterised by programming, planning systems for manipulators
- B25J9/1661—Programme controls characterised by programming, planning systems for manipulators characterised by task planning, object-oriented languages
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Robotics (AREA)
- Chemical & Material Sciences (AREA)
- Combustion & Propulsion (AREA)
- Transportation (AREA)
- Automation & Control Theory (AREA)
- Manipulator (AREA)
Abstract
In order to improve the stability control effect of the biped robot in a cooperative state, an algorithm is optimized, and the method comprises the following steps of: establishing a neural network model based on a nerve cell layer, training gait point value data of the first biped robot obtained by utilizing a gait measuring device, and training gait balance of the first biped robot; and adjusting according to the balance result. The biped robot disclosed by the invention not only can promote the training depth of the gait control model based on the experience data of the biped robot or other biped robots, but also can effectively cooperate with other biped robots to realize mutual experience sharing, so that the effects of jointly carrying the same goods, quickly adapting to a new environment, facilitating more important tasks to give enough running time and the like are achieved.
Description
Technical Field
The invention relates to the technical field of automatic control.
Background
The bipedal robot has better mobility than the conventional wheeled robot. Bipedal walking systems have very rich dynamics and require very low requirements on the walking environment. Most of the buildings and tools are designed according to the height and shape of the person, so that the bipedal robot has better use flexibility as a robot platform. Meanwhile, the stability control of the gait of the biped robot is the premise and the basis of smooth walking of the robot. Gait refers to a temporal and spatial interrelation of the joints during standing or walking, which can be described by the motion trajectories of the joints.
The existing stability study of the gait of the bipedal robot is based on a zero moment point (ZeroMomentPoint, ZMP) method, a mathematical model of the bipedal robot is built, and a control rule is deduced according to the fact that ZMP must fall in a stable area, such as the sole range of the robot foot. However, due to road conditions and other factors, the robot is likely to fail to walk, and even the bipedal robot falls down. This phenomenon is fatal to a disaster when a plurality of bipedal robots cooperate.
Disclosure of Invention
In order to improve the stability control effect of the biped robot in a cooperative state, the invention provides a robot walking simulation algorithm, which comprises the following steps of simulation on a computer: establishing a neural network model based on a nerve cell layer, training gait data of the first biped robot obtained by utilizing a gait measuring device, and training gait balance of the first biped robot; and adjusting according to the balance result.
Further, the robot has three degrees of freedom in the two feet, the ankle joint and the knee joint.
Further, the gait data includes angular velocity and acceleration, which are the angular velocity and acceleration in the forward direction.
Further, the gait measurement device comprises a gyroscope and an accelerometer.
Further, training of gait data includes:
obtaining road surface information data, wherein the road surface information data at least comprises route information and experience data which are in communication relation with the current position in an electronic map;
gait data are obtained, and training is carried out on the pre-constructed neural network model based on the gait data to obtain a primary training model:
wherein p and q represent a positive integer randomly selected from 1 to 10, and 0 is taken when the two types of corner marks have no corresponding physical meaning;
and performing deep learning on the experience data by using the first-stage training model, wherein a learning result is used for performing Hopfield network training to obtain a second-stage training model.
Further, training gait balance includes:
obtaining gait data of at least one second bipedal robot;
and performing deep learning on gait data of each second biped robot by using the second training model of the first biped robot, summarizing learning results, and finally training the learning results by using a sparse self-encoder to obtain a third training model of the first biped robot.
Further, the adjusting according to the balance result comprises: and inputting gait data of the first biped robot into a three-stage training model for differential training, and outputting motion settings of each degree of freedom of the biped, ankle joint and knee joint of the first biped robot by the obtained four-stage training model.
Further, the first-level training model is selected by using a random selection mode in a Monte Carlo search tree optimization algorithm.
Further, the empirical data is gait data from the first bipedal robot or the second bipedal robot when the first bipedal robot has previously passed through a route having the highest similarity to the route information.
The invention has the beneficial effects that: the biped robot not only can promote the training depth of the gait control model based on the experience data of the biped robot or other biped robots, but also can effectively cooperate with other biped robots to realize mutual experience sharing, thereby achieving the effects of carrying the same goods together, quickly adapting to new environments, being convenient for giving more important tasks to enough operation time, and the like.
Drawings
Fig. 1 shows a block flow diagram of the present method.
Detailed Description
A robot walking simulation algorithm comprising the steps of, on a computer: establishing a neural network model based on a nerve cell layer, training gait data of the first biped robot obtained by utilizing a gait measuring device, and training gait balance of the first biped robot; and adjusting according to the balance result.
Preferably, the robot has three degrees of freedom in the two feet, ankle joint and knee joint.
Preferably, the gait data includes angular velocity and acceleration, which are angular velocity and acceleration in the forward direction.
Preferably, the gait measurement device comprises a gyroscope and an accelerometer.
Preferably, the training of gait data comprises:
obtaining road surface information data, wherein the road surface information data at least comprises route information and experience data which are in communication relation with the current position in an electronic map;
gait data are obtained, and training is carried out on the pre-constructed neural network model based on the gait data to obtain a primary training model:
wherein i represents the degree of freedom and has a value of 1, 2 or 3; p and q represent a positive integer randomly selected from 1 to 10, and 0 is taken when the two types of corner marks have no corresponding physical meaning;
and performing deep learning on the experience data by using the first-stage training model, wherein a learning result is used for performing Hopfield network training to obtain a second-stage training model.
Preferably, training gait balance comprises:
obtaining gait data of at least one second bipedal robot:
and performing deep learning on gait data of each second biped robot by using the second training model of the first biped robot, summarizing learning results, and finally training the learning results by using a sparse self-encoder to obtain a third training model of the first biped robot.
Preferably, the adjusting according to the balancing result comprises: and inputting gait data of the first biped robot into a three-stage training model for differential training, and outputting motion settings of each degree of freedom of the biped, ankle joint and knee joint of the first biped robot by the obtained four-stage training model.
Preferably, the first-level training model is selected by using a random selection mode in a Monte Carlo search tree optimization algorithm.
Preferably, the empirical data is gait data from a past route of the first bipedal robot or the second bipedal robot having the highest similarity to the route information.
The above examples illustrate only a few embodiments of the invention, which are described in detail and are not to be construed as limiting the scope of the invention. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the invention, which are all within the scope of the invention. Accordingly, the scope of protection of the present invention is to be determined by the appended claims.
Claims (5)
1. A robot walking simulation algorithm comprising the steps of, on a computer: establishing a neural network model based on a nerve cell layer, training gait data of the first biped robot obtained by utilizing a gait measuring device, and training gait balance of the first biped robot; adjusting according to the balance result;
the double feet, the ankle joints and the knee joints of the robot are all three degrees of freedom;
the gait data includes angular velocity and acceleration, which are the angular velocity and acceleration in the forward direction;
the gait measuring device comprises a gyroscope and an accelerometer;
the training of gait data comprises:
obtaining road surface information data, wherein the road surface information data at least comprises route information and experience data which are in communication relation with the current position in an electronic map;
gait data are obtained, and training is carried out on the pre-constructed neural network model based on the gait data to obtain a primary training model:
wherein p and q represent a positive integer randomly selected from 1 to 10, and 0 is taken when the two types of corner marks have no corresponding physical meaning;
and performing deep learning on the experience data by using the first-stage training model, wherein a learning result is used for performing Hopfield network training to obtain a second-stage training model.
2. The robotic walking simulation algorithm of claim 1, wherein training gait balance comprises:
obtaining gait data of at least one second bipedal robot;
and performing deep learning on gait data of each second biped robot by using the second training model of the first biped robot, summarizing learning results, and finally training the learning results by using a sparse self-encoder to obtain a third training model of the first biped robot.
3. The robot walking simulation algorithm of claim 2, wherein the adjusting based on the balancing result comprises: and inputting gait data of the first biped robot into a three-stage training model for differential training, and outputting motion settings of each degree of freedom of the biped, ankle joint and knee joint of the first biped robot by the obtained four-stage training model.
4. A robot walking simulation algorithm as claimed in claim 3, wherein the first-level training model is selected using a random selection scheme in a monte carlo search tree optimization algorithm.
5. The robot walking simulation algorithm according to claim 4, wherein the empirical data is gait data from a past route of the first bipedal robot or the second bipedal robot having the highest similarity to the route information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011579027.2A CN114684293B (en) | 2020-12-28 | 2020-12-28 | Robot walking simulation algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011579027.2A CN114684293B (en) | 2020-12-28 | 2020-12-28 | Robot walking simulation algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114684293A CN114684293A (en) | 2022-07-01 |
| CN114684293B true CN114684293B (en) | 2023-07-25 |
Family
ID=82130043
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011579027.2A Active CN114684293B (en) | 2020-12-28 | 2020-12-28 | Robot walking simulation algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114684293B (en) |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101074494B1 (en) * | 2009-02-17 | 2011-10-17 | 동아대학교 산학협력단 | Method for Generating Optimal Trajectory of a Biped Robot for Walking Up a Staircase |
| US9221170B2 (en) * | 2013-06-13 | 2015-12-29 | GM Global Technology Operations LLC | Method and apparatus for controlling a robotic device via wearable sensors |
| KR101813697B1 (en) * | 2015-12-22 | 2017-12-29 | 한국항공대학교산학협력단 | Unmanned aerial vehicle flight control system and method using deep learning |
| EP3504034A1 (en) * | 2016-09-15 | 2019-07-03 | Google LLC. | Deep reinforcement learning for robotic manipulation |
| CN109987169B (en) * | 2017-12-29 | 2021-01-08 | 深圳市优必选科技有限公司 | Gait control method and device for biped robot, terminal device and medium |
| US20210162589A1 (en) * | 2018-04-22 | 2021-06-03 | Google Llc | Systems and methods for learning agile locomotion for multiped robots |
| CN108983804B (en) * | 2018-08-27 | 2020-05-22 | 燕山大学 | Biped robot gait planning method based on deep reinforcement learning |
| CN109968355A (en) * | 2019-03-08 | 2019-07-05 | 北京工业大学 | A kind of method that humanoid robot gait's balance model is established |
| CN110262511B (en) * | 2019-07-12 | 2022-08-09 | 同济人工智能研究院(苏州)有限公司 | Biped robot adaptive walking control method based on deep reinforcement learning |
| CN111546349A (en) * | 2020-06-28 | 2020-08-18 | 常州工学院 | New deep reinforcement learning method for humanoid robot gait planning |
-
2020
- 2020-12-28 CN CN202011579027.2A patent/CN114684293B/en active Active
Non-Patent Citations (1)
| Title |
|---|
| 基于ADAMS的双足机器人拟人行走动态仿真;王立权;俞志伟;计算机仿真;第26卷(第3期);160-182 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114684293A (en) | 2022-07-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gehring et al. | Control of dynamic gaits for a quadrupedal robot | |
| US7881824B2 (en) | System and method of controlling a legged locomotion robot | |
| Vundavilli et al. | Dynamically balanced optimal gaits of a ditch-crossing biped robot | |
| Wang et al. | Drc-hubo walking on rough terrains | |
| CN104298243A (en) | Humanoid robot uneven ground walking stability control method | |
| Zhong et al. | Trajectory planning for biped robot walking on uneven terrain–Taking stepping as an example | |
| CN106094817B (en) | Intensified learning humanoid robot gait's planing method based on big data mode | |
| US20220324109A1 (en) | Method and apparatus for controlling multi-legged robot, and storage medium | |
| Vundavilli et al. | Soft computing-based gait planners for a dynamically balanced biped robot negotiating sloping surfaces | |
| Silva et al. | Using reinforcement learning to improve the stability of a humanoid robot: Walking on sloped terrain | |
| Xu et al. | Research on kinematics and stability of a bionic wall-climbing hexapod robot | |
| Lee et al. | Time-dependent genetic algorithm and its application to quadruped’s locomotion | |
| CN114684293B (en) | Robot walking simulation algorithm | |
| Mănoiu-Olaru et al. | Stability analysis software platform dedicated for a hexapod robot | |
| CN112698656B (en) | Gait stability control method for biped robot | |
| Ferreira et al. | Human-like gait adaptation to slippery surfaces for the nao robot wearing instrumented shoes | |
| Clever et al. | A new template model for optimization studies of human walking on different terrains | |
| CN114690788A (en) | Robot gait optimization method | |
| Tarokh et al. | Systematic method for kinematics modeling of legged robots on uneven terrain | |
| Nahrendra et al. | Robust recovery motion control for quadrupedal robots via learned terrain imagination | |
| Fedorov | Humanoid bipedal balancing and locomotion | |
| Lathan et al. | Analytical approach for generating dynamically balanced gaits for obstacle crossing biped robot | |
| Spitz et al. | Minimal feedback to a rhythm generator improves the robustness to slope variations of a compass biped | |
| Leng et al. | Parameter design of biped robot motion system based on multi-objective optimization | |
| Belter | Adaptive foothold selection for a hexapod robot walking on rough terrain |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |