CN109241552A - A kind of underwater robot motion planning method based on multiple constraint target - Google Patents
A kind of underwater robot motion planning method based on multiple constraint target Download PDFInfo
- Publication number
- CN109241552A CN109241552A CN201810764979.8A CN201810764979A CN109241552A CN 109241552 A CN109241552 A CN 109241552A CN 201810764979 A CN201810764979 A CN 201810764979A CN 109241552 A CN109241552 A CN 109241552A
- Authority
- CN
- China
- Prior art keywords
- robot
- training
- constraint
- target
- underwater
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/04—Constraint-based CAD
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
- Manipulator (AREA)
Abstract
A kind of underwater robot motion planning method based on multiple constraint target belongs to machine learning and underwater robot motion planning field.The model construction stage: current environment is converted by the flow velocity signal of the signal of robot obstacle-avoiding sonar and flow sensor;According to Dynamic Constraints, discrete movement space is established;Using underwater obstacle as constraint, reward functions are established;Markovian decision process is established based on multi-objective restriction, is realized for algorithm and establishes basis;Training stage: being trained based on Q learning algorithm, in current environment, executes movement based on Greedy strategy, primary strategy is assessed based on original strategy and updated to one step strategy of every execution, and improvement strategy realizes planning purpose until adapting to environment.The present invention considers the multiple constraints target such as water flow, obstruction, target, by intensified learning method in conjunction with underwater multiple constraint target, realizes the motion planning of underwater robot, has stronger real-time, and can be adapted for a variety of environment.
Description
Technical field
The invention belongs to machine learning and underwater robot motion planning fields, and in particular to one kind is based on multiple constraint mesh
Target underwater robot motion planning method.
Background technique
Intelligent Underwater Robot scientific research of seas, ocean development, Underwater Engineering and in terms of have extensively
Application prospect.Intelligent Underwater Robot is generally operational under complicated marine environment, is made to preferably complete various operations
Life task and its own safety of guarantee, need it under circumstances not known to there is autokinetic movement to plan ability, can be in circumstances not known
Middle avoiding barrier, navigation to target point.
Traditional underwater robot motion planning technology needs to construct global map in advance.It is needed when environment changes
Connection model is re-established, adaptability is poor, not very practical.Intensified learning is a kind of unsupervised learning method, it is one
A process constantly attempted.It obtains knowledge with evaluation by constantly action, and improvement strategy is made final with adapting to environment
Evaluation function value is maximum, reaches the destination of study.
Intensified learning is but traditional based on the underwater of intensified learning it is verified that can use in robot under water
Robot motion planning method, it is contemplated that single constrained objective is not considered water flow constraint, goal constraint simultaneously and hindered
Influence in the case of the multi-objective restrictions such as boat object constraint to underwater robot movement.
Summary of the invention
The purpose of the present invention is to provide a kind of underwater robot motion planning methods based on multiple constraint target.This method
By the underwater human occupant dynamic model under the influence of construction water flow, and the method for combining intensified learning carries out multiple constraint target
Fusion, constructs reasonable prize signal and motion space, exports underwater robot optimal control policy by training.Furthermore this hair
It is bright also to combine underwater multiple constraint target with the Q learning algorithm in intensified learning, underwater robot can be allowed unknown
Environmental characteristic is obtained under underwater environment, is carried out Policy iteration, is completed the motion planning of underwater robot.
The object of the present invention is achieved like this:
A kind of underwater robot motion planning method based on multiple constraint target is divided into model construction stage and algorithm training
Stage, specifically includes the following steps:
(1) the model construction stage, the model construction of Markovian decision process E is referred specifically to, intensified learning task usually may be used
To be described with Markovian decision process.Because of the particularity of underwater environment, environmental constraints, obstruction constraint and target are considered
The multi-objective restrictions such as point constraint construct Markovian decision process, specifically includes the following steps:
(1-1) establishes current environment x according to sensor signalt;If the obstacle distance of the i-th freedom degree direction of robot is
liIf barrier is not present in i freedom degree, it is set as infinite;The flow velocity of robot present position is set as vc;Real-time localization machine
The Euclidean distance d of device people position, calculating robot and target point;
The maximum value that (1-2) can advance according to underwater robot, establishes the motion space A of robot, and A is moved by five
Order composition, before respectively advancing, is left front, right, left side pushes away and right side pushes away, speed va, angular speed ωa;
(1-3) considers obstruction constraint, if the i-th freedom degree guards against safe distance h under wateriIf the l detectedi< hi,
Then think collision, a negative reward r is setter;
(1-4) considers target point constraint, and target point threshold value is d', if detecting that d becomes larger, a negative reward is arranged
roppIf detecting that d becomes smaller, a positive reward r is setmoveIf detecting d < d', robot arrives at target point, if
Set a positive reward rarr。
(2) the algorithm training stage refers specifically to robot and carries out continuous trial and error in Computer Simulation, learning strategy, specifically
The following steps are included:
(2-1) initializes t=0, and t represents the step time of robot training movement each time;Initialize rt=0, rtRepresent machine
People executes reward obtained when t-th of movement;
(2-2) initializes a matrix Q, and (x, a), when being recorded in state x, the Q value that selection movement a can be obtained is initialized
It is 0;
(2-3) initialization counter count=0, the total degree of recorder people training;M value is set, it is total to represent robot
It needs to train M times altogether;
(2-4) is executed (2-5) when count is less than specified frequency of training M, otherwise executes (2-14);
(2-5) obtains sensor signal, obtains current state xt, including obstruction information, the freedom degree side i of robot
To barrier distance li, set infinite for distance if without barrier;The ocean current flow rate information vc of current locationt;
Own location information, and Euclidean distance d of the target point apart from robot is calculated;
(2-6) acts a according to matrix Q, selectiont;
(2-7) considers that kinematical constraint and water flow constraint will be selected according to the speed that target formula is actually externally shown
The movement a selectedtSpeed in conjunction with flow velocity, according to combining what is obtained emulate, and update li;
(2-8) is if li< hi, execute (2-9), otherwise execute (2-10);
(2-9) collides, rt=rter, terminate this training, enable xt+1For sky, matrix Q is updated, and by count+1, is enabled
T=0 re-executes training from (2-4);
(2-10) is executed (2-11) if d'< d, is otherwise arrived at target point, terminate this training, enable rt=rarr, enable
xt+1For sky, and matrix Q is updated by count+1, enable t=0, re-execute training from (2-4);
(2-11) is if dt< dt-1, execute (2-12), otherwise execute (2-13);
(2-12) d reduces, and enables rt=rmove, update xt+1, and matrix Q is updated, by t+1, trained stream is re-executed from (2-5)
Journey;
(2-13) d increases, and enables rt=ropp, update xt+1, and matrix Q is updated, by t+1, trained stream is re-executed from (2-5)
Journey;
(2-14) terminates training, the matrix Q after being trained;
(2-15) exports underwater robot motion planning strategy.
The kinematical constraint, i.e., the kinematic constraint of underwater robot itself in training process are as follows: assuming that aircraft
Center of gravity is (x, y) in the coordinate of fixed coordinate system, then robot fixed coordinate system speed are as follows:
Wherein, θ is Angle of Trim, and φ is Angle of Heel, and α is influence coefficient of the kinematic constraint to underwater robot speed.
The water flow constraint is accounted for when selection acts in the training process with following methods: in learning training process
In, xtUnder state by ADCP obtain flow velocity be vct, a movement a according to strategy, in Robot Selection behavior aggregatet,
The speed of body isWhen robot execution acts, water flow constraint, the practical route speed externally shown are considered
Are as follows: vit=vat+βvct, wherein β is influence coefficient of the water flow to underwater robot speed.
The selection acts atMethod particularly includes: use Greedy strategy, given threshold ε, using computer generate with
Machine number ε ', if random number is less than threshold value, i.e. ε ' < ε, then robot executes state Q (x in Q matrixt, a) in element maximum value
Corresponding movement, i.e. at=maxa Q(xt,a);If random number is greater than threshold value, i.e. ε ' > ε, then robot randomly chooses one and moves
It executes, i.e. at=randomQ (xt, a).
The method of the update matrix Q are as follows: it is assumed that the state belonging to before robot execution movement is xt, what should be executed is dynamic
As at, according to the obtained award coefficient r of feedbackt, the state arrived at after execution movement is xt+1, then
Q(xt,at)←(1-α)*rt+α*(rt+γmaxa'Q(xt+1,a'))
Wherein α is learning efficiency, and wherein γ is discount factor.
The beneficial effects of the present invention are:
(1) present invention considers the multiple constraints target such as water flow, obstruction, target, and traditional intensified learning planing method
A variety of constrained objectives are not considered simultaneously, and the method for the training has practicability and robustness;
(2) present invention realizes the movement rule of underwater robot by intensified learning method in conjunction with underwater multiple constraint target
It draws, there is stronger real-time, and can be adapted for a variety of environment.
Detailed description of the invention
Fig. 1 is a kind of model construction schematic diagram of underwater robot motion planning method based on multiple constraint target;
Fig. 2 is a kind of training stage execution flow chart of underwater robot motion planning method based on multiple constraint target.
Specific embodiment
Specific embodiments of the present invention will be further explained with reference to the accompanying drawing:
The present invention relates to a kind of underwater robot motion planning methods, specifically by multi-objective restriction and intensified learning method
Combine, in underwater robot motion planning.The model construction stage: the signal of robot obstacle-avoiding sonar and flow velocity are passed
The flow velocity signal of sensor is converted into current environment;Dynamic Constraints based on underwater robot establish discrete movement space;With water
Lower barrier establishes reward functions as constraint;Markovian decision process is established based on multi-objective restriction, is built for algorithm realization
Vertical basis.Training stage: being trained based on Q learning algorithm, in current environment, executes movement, every execution based on Greedy strategy
One step strategy, is assessed based on original strategy and is updated primary strategy, and improvement strategy realizes planning mesh until adapting to environment
's.Intensified learning method in conjunction with underwater multiple constraint target, is realized the motion planning of underwater robot by the present invention, have compared with
Strong real-time, and can be adapted for a variety of environment.
The present invention is directed to the particularity of underwater environment, considers multiple constraint target combination intensified learning method, the underwater machine of training
Device people's motion planning strategy.It is divided into model construction stage and Strategies Training stage, comprising the following steps:
1. the model construction stage, as shown in Figure 1, the specific steps are as follows:
Intensified learning task can usually be described with Markovian decision process.Because of the particularity of underwater environment, consider
The multi-objective restrictions such as environmental constraints, obstruction constraint and target point constraint construct Markovian decision process.
The concrete composition of state space X: the one, obstruction of avoidance sonar handling machine people's local environment of underwater robot
Information, i.e. the obstruction clearance information l of robot i freedom degree directioni;Two, the ocean current letter of ADCP handling machine people local environment
Breath, i.e. the flow velocity vc of robot present position;Three, the relative position information of GPS handling machine people and target point, i.e. robot away from
Euclidean distance d from target point.
The concrete composition of motion space A: the motion space in the present invention includes four kinds of control commands, and title is respectively a left side
Before, it is preceding, right before, left side pushes away and right side pushes away.The linear velocity of robot is fixed value va。
The concrete composition of reward function R: robot once collides, reward value rter;There is no collisions for robot
But distance objective point is more and more remoter, reward value ropp;Robot there is no collision and distance objective point it is more and more closer, prize
Encouraging value is rmove;Robot arrives at target point, reward value rarr。
2. Strategies Training stage, process are as shown in Figure 2, the specific steps are as follows:
Virtual environment is initially set up for training, the specific method is as follows:
Using robot motion simulation software establish one emulation marine environment, in virtual environment set barrier,
Target point and ocean current.Barrier can define at random with target point, and define 6-12 different robot starting points.
Two-dimensional surface space is subjected to rasterizing processing, the ocean current in each grid can be considered identical, and flow field is flowed with one
Function Ψ (x, y) is generated at random, and the velocity field of ocean current can be obtained by flow field function:
Due to the Incoercibility of fluid
Vc in formulax, vcyRespectively ocean current is taken as each in the position (x, y) along the x axis with the velocity component of Y direction
The central point of grid.
Carry out Strategies Training, the specific steps are as follows:
1) t=0 is initialized, t represents the step time of robot training movement each time;Rt=0 is initialized, rt represents robot
Execute t-th of movement reward obtained.Defining a matrix Q, (x, a), when being recorded in state x, selection movement a's can be obtained
The Q value obtained, is initialized as 0.Initialization counter count=0, the total degree of recorder people training.M value is set, machine is represented
Device people needs to train M times in total.Initialize the radius of safety h of underwater robot i freedom degree directioni.D' value is set, machine is represented
The threshold value of people and target point distance.
2) state for initializing robot, randomly selects a starting point and starts to be explored.
3) robot obtains environmental information xt, including obstruction information, the barrier of robot i freedom degree direction with
The distance l of roboti, set infinite for distance if without barrier;The ocean current flow rate information vc of current location;Itself
Location information, and Euclidean distance d of the target point apart from robot is calculated.
4) threshold epsilon is set, generates a random number ε ' by computer, if ε ' < ε, robot randomly selects dynamic
The movement made in space executes, i.e. at=randomQ (xt, a);If ε ' > ε, robot according to matrix Q (x, a), choosing
It selects in state xtUnder, it is worth maximum movement a, i.e. at=maxa Q(xt,a)。
5) robot considers kinematical constraint and water flow constraint, according to the speed that target formula is actually externally shown,
In a simulated environment, according to speed vitMovement.
6) robot has executed movement atAfterwards, environmental information x is obtained againt+1。
If 6-1) li< hi, explanation collides, this training terminates, counter count+1, according to
Q(xt,at)←(1-α)*rt+α*(rt+γmaxa'Q(xt+1,a'))
Update matrix Q, train epochs t=0, if count < M, the re -training since step 2), if count=
M continues to execute step 7).
If 6-2) li > hi, illustrate to continue to judge whether it arrives at target point there is no collision.
If 6-2-1) d≤d', illustrating to arrive at target point, this training terminates, counter count+1, matrix Q is updated,
Train epochs t=0, if count < M, the re -training since step 2) continue to execute step 7) if count=M.
If 6-2-2) d > d', illustrate that, without arriving at target point, t+1 continues this training from step 3).
7) training is completed, and exports underwater robot motion planning strategy.
This method advantage is to consider the multiple constraints target such as water flow, obstruction, target, traditional intensified learning planning side
Method does not consider a variety of constrained objectives simultaneously, and trained method lacks practicability and robustness.The present invention passes through intensified learning pair
Multiple constraint target carries out Fusion Features, can train more practical underwater robot motion planning strategy.
Claims (5)
1. a kind of underwater robot motion planning method based on multiple constraint target, which is characterized in that this method is divided into model structure
Build stage and algorithm training stage, comprising:
(1) the model construction stage;The model construction of Markovian decision process E, comprising the following steps:
(1-1) establishes current environment x according to sensor signalt;If the obstacle distance of the i-th freedom degree direction of robot is liIf
Barrier is not present in i freedom degree, then by liIt is set as infinite;The flow velocity of robot present position is set as vc;Real-time localization machine
The Euclidean distance d of device people position, calculating robot and target point;
The maximum value that (1-2) can advance according to underwater robot establishes the motion space A of robot;The A includes five
Motion command, before respectively advancing, is left front, right, left side pushes away and right side pushes away;Speed is va, angular speed ωa;
(1-3) considers obstruction constraint;If the i-th freedom degree guards against safe distance h under wateriIf the l detectedi< hi, then it is assumed that
Collision occurs, and a negative reward r is arrangedter;
(1-4) considers target point constraint;If target point threshold value is d', if detecting, d becomes larger, and a negative reward r is arrangedopp,
If detecting, d becomes smaller, and a positive reward r is arrangedmoveIf detecting d < d', robot arrives at target point, setting one
A positive reward rarr。
(2) algorithm training rank;Robot carries out continuous trial and error, learning strategy in Computer Simulation, comprising the following steps:
(2-1) initializes t=0, and t represents the step time of robot training movement each time;Initialize rt=0, rtRobot is represented to hold
Reward obtained when t-th of movement of row;
(2-2) initializes a matrix Q, and (x, a), when being recorded in state x, selection acts the Q value that a is obtained;
(2-3) initialization counter count=0, the total degree of recorder people training;M value is set, represent robot needs in total
It trains M times;
(2-4) is executed (2-5) when count is less than specified frequency of training M, otherwise executes (2-14);
(2-5) obtains sensor signal, obtains current state xt;The current state xtIncluding obstruction information, robot i
The distance l of the barrier of freedom degree directioni, current location ocean current flow rate information vct, own location information, target is calculated
Euclidean distance d of the point apart from robot;
(2-6) acts a according to matrix Q, selectiont;
(2-7) considers kinematical constraint and water flow constraint, by the movement a of selectiontSpeed in conjunction with flow velocity, according to combine
To the route speed that externally shows of reality emulated, and update li;
(2-8) is if li< hi, execute (2-9), otherwise execute (2-10);
(2-9) collides, rt=rter, terminate this training, enable xt+1For sky, matrix Q is updated, and by count+1, enables t=0,
Training is re-executed from (2-4);
(2-10) is executed (2-11) if d'< d, is otherwise arrived at target point, terminate this training, enable rt=rarr, enable xt+1For
Sky, and matrix Q is updated by count+1, t=0 is enabled, re-executes training from (2-4);
(2-11) is if dt< dt-1, execute (2-12), otherwise execute (2-13);
(2-12) d reduces, and enables rt=rmove, update xt+1, and matrix Q is updated, by t+1, trained process is re-executed from (2-5);
(2-13) d increases, and enables rt=ropp, update xt+1, and matrix Q is updated, by t+1, trained process is re-executed from (2-5);
(2-14) terminates training, the matrix Q after being trained;
(2-15) exports underwater robot motion planning strategy.
2. a kind of underwater robot motion planning method based on multiple constraint target according to claim 1,
It is characterized in that: the kinematical constraint, i.e., the kinematic constraint of underwater robot itself in training process are as follows: assuming that aircraft
Center of gravity is (x, y) in the coordinate of fixed coordinate system, then robot fixed coordinate system speed are as follows:
Wherein, θ is Angle of Trim, and φ is Angle of Heel, and α is influence coefficient of the kinematic constraint to underwater robot speed.
3. a kind of underwater robot motion planning method based on multiple constraint target according to claim 1, feature exist
In: the water flow is constrained and is accounted for when selection in the training process acts with following methods: during learning training, xt
Under state by ADCP obtain flow velocity be vct, a movement a according to strategy, in Robot Selection behavior aggregatet, the speed of itself
Degree isWhen robot execution acts, water flow constraint, the practical route speed externally shown are as follows: vi are consideredt
=vat+βvct, wherein β is influence coefficient of the water flow to underwater robot speed.
4. a kind of underwater robot motion planning method based on multiple constraint target according to claim 1, feature exist
In: the selection acts atMethod particularly includes: Greedy strategy is used, given threshold ε generates random number using computer
ε ', if random number is less than threshold value, i.e. ε ' < ε, then robot executes state Q (x in Q matrixt, a) in element maximum value it is corresponding
Movement, i.e. at=maxaQ(xt,a);If random number is greater than threshold value, i.e. ε ' > ε, then robot randomly chooses a movement and holds
Row, i.e. at=randomQ (xt, a).
5. a kind of underwater robot motion planning method based on multiple constraint target according to claim 1, feature exist
In: the method for the update matrix Q are as follows: it is assumed that the state belonging to before robot execution movement is xt, the movement that should be executed is
at, according to the obtained award coefficient r of feedbackt, the state arrived at after execution movement is xt+1, then
Q(xt,at)←(1-α)*rt+α*(rt+γmaxa'Q(xt+1,a'))
Wherein α is learning efficiency, and wherein γ is discount factor.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810764979.8A CN109241552B (en) | 2018-07-12 | 2018-07-12 | Underwater robot motion planning method based on multiple constraint targets |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810764979.8A CN109241552B (en) | 2018-07-12 | 2018-07-12 | Underwater robot motion planning method based on multiple constraint targets |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109241552A true CN109241552A (en) | 2019-01-18 |
CN109241552B CN109241552B (en) | 2022-04-05 |
Family
ID=65072571
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810764979.8A Active CN109241552B (en) | 2018-07-12 | 2018-07-12 | Underwater robot motion planning method based on multiple constraint targets |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109241552B (en) |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110196605A (en) * | 2019-04-26 | 2019-09-03 | 大连海事大学 | A kind of more dynamic object methods of the unmanned aerial vehicle group of intensified learning collaboratively searching in unknown sea area |
CN110333739A (en) * | 2019-08-21 | 2019-10-15 | 哈尔滨工程大学 | A kind of AUV conduct programming and method of controlling operation based on intensified learning |
CN110765686A (en) * | 2019-10-22 | 2020-02-07 | 中国人民解放军战略支援部队信息工程大学 | Method for designing shipborne sonar sounding line by using limited wave band submarine topography |
CN110779132A (en) * | 2019-11-13 | 2020-02-11 | 垚控科技(上海)有限公司 | Water pump equipment operation control system of air conditioning system based on reinforcement learning |
CN110955239A (en) * | 2019-11-12 | 2020-04-03 | 中国地质大学(武汉) | Unmanned ship multi-target trajectory planning method and system based on inverse reinforcement learning |
CN112052511A (en) * | 2020-06-15 | 2020-12-08 | 成都蓉奥科技有限公司 | Air combat maneuver strategy generation technology based on deep random game |
CN112149354A (en) * | 2020-09-24 | 2020-12-29 | 哈尔滨工程大学 | Reinforced learning algorithm research platform for UUV cluster |
CN112631269A (en) * | 2019-10-08 | 2021-04-09 | 国立大学法人静冈大学 | Autonomous mobile robot and control program for autonomous mobile robot |
CN112650246A (en) * | 2020-12-23 | 2021-04-13 | 武汉理工大学 | Ship autonomous navigation method and device |
CN112925319A (en) * | 2021-01-25 | 2021-06-08 | 哈尔滨工程大学 | Underwater autonomous vehicle dynamic obstacle avoidance method based on deep reinforcement learning |
CN112945232A (en) * | 2019-12-11 | 2021-06-11 | 中国科学院沈阳自动化研究所 | Target value planning method for near-bottom terrain tracking of underwater robot |
CN113052257A (en) * | 2021-04-13 | 2021-06-29 | 中国电子科技集团公司信息科学研究院 | Deep reinforcement learning method and device based on visual converter |
CN113110493A (en) * | 2021-05-07 | 2021-07-13 | 北京邮电大学 | Path planning equipment and path planning method based on photonic neural network |
WO2021147192A1 (en) * | 2020-01-21 | 2021-07-29 | 厦门邑通软件科技有限公司 | Machine heuristic learning method, system and device for operation behavior record management |
CN114559439A (en) * | 2022-04-27 | 2022-05-31 | 南通科美自动化科技有限公司 | Intelligent obstacle avoidance control method and device for mobile robot and electronic equipment |
CN116295449A (en) * | 2023-05-25 | 2023-06-23 | 吉林大学 | Method and device for indicating path of autonomous underwater vehicle |
CN117079118A (en) * | 2023-10-16 | 2023-11-17 | 广州华夏汇海科技有限公司 | Underwater walking detection method and system based on visual detection |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102117071A (en) * | 2009-12-30 | 2011-07-06 | 中国科学院沈阳自动化研究所 | Multi-underwater robot semi-physical simulation system and control method thereof |
CN105512769A (en) * | 2015-12-16 | 2016-04-20 | 上海交通大学 | Unmanned aerial vehicle route planning system and unmanned aerial vehicle route planning method based on genetic programming |
CN106950969A (en) * | 2017-04-28 | 2017-07-14 | 深圳市唯特视科技有限公司 | It is a kind of based on the mobile robot continuous control method without map movement planner |
CN107065881A (en) * | 2017-05-17 | 2017-08-18 | 清华大学 | A kind of robot global path planning method learnt based on deeply |
CN107168309A (en) * | 2017-05-02 | 2017-09-15 | 哈尔滨工程大学 | A kind of underwater multi-robot paths planning method of Behavior-based control |
WO2017161632A1 (en) * | 2016-03-24 | 2017-09-28 | 苏州大学张家港工业技术研究院 | Cleaning robot optimal target path planning method based on model learning |
CN107883961A (en) * | 2017-11-06 | 2018-04-06 | 哈尔滨工程大学 | A kind of underwater robot method for optimizing route based on Smooth RRT algorithms |
CN107918396A (en) * | 2017-11-30 | 2018-04-17 | 深圳市智能机器人研究院 | A kind of underwater cleaning robot paths planning method and system based on hull model |
CN108268031A (en) * | 2016-12-30 | 2018-07-10 | 深圳光启合众科技有限公司 | Paths planning method, device and robot |
-
2018
- 2018-07-12 CN CN201810764979.8A patent/CN109241552B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102117071A (en) * | 2009-12-30 | 2011-07-06 | 中国科学院沈阳自动化研究所 | Multi-underwater robot semi-physical simulation system and control method thereof |
CN105512769A (en) * | 2015-12-16 | 2016-04-20 | 上海交通大学 | Unmanned aerial vehicle route planning system and unmanned aerial vehicle route planning method based on genetic programming |
WO2017161632A1 (en) * | 2016-03-24 | 2017-09-28 | 苏州大学张家港工业技术研究院 | Cleaning robot optimal target path planning method based on model learning |
CN108268031A (en) * | 2016-12-30 | 2018-07-10 | 深圳光启合众科技有限公司 | Paths planning method, device and robot |
CN106950969A (en) * | 2017-04-28 | 2017-07-14 | 深圳市唯特视科技有限公司 | It is a kind of based on the mobile robot continuous control method without map movement planner |
CN107168309A (en) * | 2017-05-02 | 2017-09-15 | 哈尔滨工程大学 | A kind of underwater multi-robot paths planning method of Behavior-based control |
CN107065881A (en) * | 2017-05-17 | 2017-08-18 | 清华大学 | A kind of robot global path planning method learnt based on deeply |
CN107883961A (en) * | 2017-11-06 | 2018-04-06 | 哈尔滨工程大学 | A kind of underwater robot method for optimizing route based on Smooth RRT algorithms |
CN107918396A (en) * | 2017-11-30 | 2018-04-17 | 深圳市智能机器人研究院 | A kind of underwater cleaning robot paths planning method and system based on hull model |
Non-Patent Citations (3)
Title |
---|
YUSHAN SUN: "Research on Recovery Strategy and Motion Control for Autonomous Underwater Vehicle", 《PROCEEDINGS OF 2018 INTERNATIONAL CONFERENCE ON ADVANCED CONTROL, AUTOMATION AND; ARTIFICIAL INTELLIGENCE (ACAAI2018)》 * |
ZHANG GUO-CHENG: "A Novel Adaptive Second Order Sliding Mode Path Following Control for a Portable AUV", 《OCEAN ENGINEERING》 * |
洪晔 等: "基于分层马尔可夫决策过程的AUV 全局路径规划研究", 《系统仿真学报》 * |
Cited By (27)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110196605B (en) * | 2019-04-26 | 2022-03-22 | 大连海事大学 | Method for cooperatively searching multiple dynamic targets in unknown sea area by reinforcement learning unmanned aerial vehicle cluster |
CN110196605A (en) * | 2019-04-26 | 2019-09-03 | 大连海事大学 | A kind of more dynamic object methods of the unmanned aerial vehicle group of intensified learning collaboratively searching in unknown sea area |
CN110333739A (en) * | 2019-08-21 | 2019-10-15 | 哈尔滨工程大学 | A kind of AUV conduct programming and method of controlling operation based on intensified learning |
CN110333739B (en) * | 2019-08-21 | 2020-07-31 | 哈尔滨工程大学 | AUV (autonomous Underwater vehicle) behavior planning and action control method based on reinforcement learning |
CN112631269A (en) * | 2019-10-08 | 2021-04-09 | 国立大学法人静冈大学 | Autonomous mobile robot and control program for autonomous mobile robot |
CN112631269B (en) * | 2019-10-08 | 2024-06-11 | 国立大学法人静冈大学 | Autonomous mobile robot and control program for autonomous mobile robot |
CN110765686A (en) * | 2019-10-22 | 2020-02-07 | 中国人民解放军战略支援部队信息工程大学 | Method for designing shipborne sonar sounding line by using limited wave band submarine topography |
CN110765686B (en) * | 2019-10-22 | 2020-09-11 | 中国人民解放军战略支援部队信息工程大学 | Method for designing shipborne sonar sounding line by using limited wave band submarine topography |
CN110955239A (en) * | 2019-11-12 | 2020-04-03 | 中国地质大学(武汉) | Unmanned ship multi-target trajectory planning method and system based on inverse reinforcement learning |
CN110955239B (en) * | 2019-11-12 | 2021-03-02 | 中国地质大学(武汉) | Unmanned ship multi-target trajectory planning method and system based on inverse reinforcement learning |
CN110779132A (en) * | 2019-11-13 | 2020-02-11 | 垚控科技(上海)有限公司 | Water pump equipment operation control system of air conditioning system based on reinforcement learning |
CN112945232A (en) * | 2019-12-11 | 2021-06-11 | 中国科学院沈阳自动化研究所 | Target value planning method for near-bottom terrain tracking of underwater robot |
CN112945232B (en) * | 2019-12-11 | 2022-11-01 | 中国科学院沈阳自动化研究所 | Target value planning method for near-bottom terrain tracking of underwater robot |
WO2021147192A1 (en) * | 2020-01-21 | 2021-07-29 | 厦门邑通软件科技有限公司 | Machine heuristic learning method, system and device for operation behavior record management |
CN112052511A (en) * | 2020-06-15 | 2020-12-08 | 成都蓉奥科技有限公司 | Air combat maneuver strategy generation technology based on deep random game |
CN112149354A (en) * | 2020-09-24 | 2020-12-29 | 哈尔滨工程大学 | Reinforced learning algorithm research platform for UUV cluster |
CN112650246A (en) * | 2020-12-23 | 2021-04-13 | 武汉理工大学 | Ship autonomous navigation method and device |
CN112925319A (en) * | 2021-01-25 | 2021-06-08 | 哈尔滨工程大学 | Underwater autonomous vehicle dynamic obstacle avoidance method based on deep reinforcement learning |
CN113052257B (en) * | 2021-04-13 | 2024-04-16 | 中国电子科技集团公司信息科学研究院 | Deep reinforcement learning method and device based on visual transducer |
CN113052257A (en) * | 2021-04-13 | 2021-06-29 | 中国电子科技集团公司信息科学研究院 | Deep reinforcement learning method and device based on visual converter |
CN113110493A (en) * | 2021-05-07 | 2021-07-13 | 北京邮电大学 | Path planning equipment and path planning method based on photonic neural network |
CN114559439A (en) * | 2022-04-27 | 2022-05-31 | 南通科美自动化科技有限公司 | Intelligent obstacle avoidance control method and device for mobile robot and electronic equipment |
CN114559439B (en) * | 2022-04-27 | 2022-07-26 | 南通科美自动化科技有限公司 | Mobile robot intelligent obstacle avoidance control method and device and electronic equipment |
CN116295449A (en) * | 2023-05-25 | 2023-06-23 | 吉林大学 | Method and device for indicating path of autonomous underwater vehicle |
CN116295449B (en) * | 2023-05-25 | 2023-09-12 | 吉林大学 | Method and device for indicating path of autonomous underwater vehicle |
CN117079118A (en) * | 2023-10-16 | 2023-11-17 | 广州华夏汇海科技有限公司 | Underwater walking detection method and system based on visual detection |
CN117079118B (en) * | 2023-10-16 | 2024-01-16 | 广州华夏汇海科技有限公司 | Underwater walking detection method and system based on visual detection |
Also Published As
Publication number | Publication date |
---|---|
CN109241552B (en) | 2022-04-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109241552A (en) | A kind of underwater robot motion planning method based on multiple constraint target | |
CN109540151B (en) | AUV three-dimensional path planning method based on reinforcement learning | |
Chen et al. | A hybrid path planning algorithm for unmanned surface vehicles in complex environment with dynamic obstacles | |
Cao et al. | Target search control of AUV in underwater environment with deep reinforcement learning | |
CN108564202B (en) | Unmanned ship route optimization method based on environment forecast information | |
CN108803313B (en) | Path planning method based on ocean current prediction model | |
CN106338919A (en) | USV (Unmanned Surface Vehicle) track tracking control method based on enhanced learning type intelligent algorithm | |
Hao et al. | Dynamic path planning of a three-dimensional underwater AUV based on an adaptive genetic algorithm | |
Wang et al. | Improved quantum particle swarm optimization algorithm for offline path planning in AUVs | |
Xue et al. | Proximal policy optimization with reciprocal velocity obstacle based collision avoidance path planning for multi-unmanned surface vehicles | |
CN108549394A (en) | A kind of more AUV straight lines formation control methods based on pilotage people and virtual pilotage people | |
CN107168309A (en) | A kind of underwater multi-robot paths planning method of Behavior-based control | |
Cao et al. | Toward optimal rendezvous of multiple underwater gliders: 3D path planning with combined sawtooth and spiral motion | |
CN104794526A (en) | Automatic ship anti-collision method optimized by wolf colony search algorithm | |
CN109765929A (en) | A kind of UUV Real Time Obstacle Avoiding planing method based on improvement RNN | |
Lan et al. | Path planning for underwater gliders in time-varying ocean current using deep reinforcement learning | |
Sun et al. | A novel fuzzy control algorithm for three-dimensional AUV path planning based on sonar model | |
CN117215317B (en) | Unmanned ship local path planning method, equipment and storage medium | |
Fang et al. | Autonomous underwater vehicle formation control and obstacle avoidance using multi-agent generative adversarial imitation learning | |
CN113093804A (en) | Unmanned ship formation control method and control system based on inversion sliding mode control | |
Xia et al. | Research on collision avoidance algorithm of unmanned surface vehicle based on deep reinforcement learning | |
CN114973650A (en) | Vehicle ramp entrance confluence control method, vehicle, electronic device, and storage medium | |
Yan et al. | Real-world learning control for autonomous exploration of a biomimetic robotic shark | |
Wang et al. | Obstacle avoidance for environmentally-driven USVs based on deep reinforcement learning in large-scale uncertain environments | |
Liang et al. | Economic MPC-based planning for marine vehicles: Tuning safety and energy efficiency |
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 | ||
EE01 | Entry into force of recordation of patent licensing contract | ||
EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20190118 Assignee: Osenda (Shandong) Offshore Engineering Co.,Ltd. Assignor: HARBIN ENGINEERING University Contract record no.: X2023230000005 Denomination of invention: A Motion Planning Method of Underwater Vehicle Based on Multi-constrained Targets Granted publication date: 20220405 License type: Exclusive License Record date: 20230130 |