CN108820157A - A kind of Ship Intelligent Collision Avoidance method based on intensified learning - Google Patents
A kind of Ship Intelligent Collision Avoidance method based on intensified learning Download PDFInfo
- Publication number
- CN108820157A CN108820157A CN201810378954.4A CN201810378954A CN108820157A CN 108820157 A CN108820157 A CN 108820157A CN 201810378954 A CN201810378954 A CN 201810378954A CN 108820157 A CN108820157 A CN 108820157A
- Authority
- CN
- China
- Prior art keywords
- ship
- collision
- strategy
- intensified learning
- parameter
- 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
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B63—SHIPS OR OTHER WATERBORNE VESSELS; RELATED EQUIPMENT
- B63B—SHIPS OR OTHER WATERBORNE VESSELS; EQUIPMENT FOR SHIPPING
- B63B43/00—Improving safety of vessels, e.g. damage control, not otherwise provided for
- B63B43/18—Improving safety of vessels, e.g. damage control, not otherwise provided for preventing collision or grounding; reducing collision damage
Landscapes
- Health & Medical Sciences (AREA)
- Public Health (AREA)
- Chemical & Material Sciences (AREA)
- Engineering & Computer Science (AREA)
- Combustion & Propulsion (AREA)
- Mechanical Engineering (AREA)
- Ocean & Marine Engineering (AREA)
- Traffic Control Systems (AREA)
- Feedback Control In General (AREA)
Abstract
The Ship Intelligent Collision Avoidance method based on intensified learning that the invention discloses a kind of obtains the static data and dynamic data of two ships first;Then the legitimacy of inspection data judges whether to need to start collision prevention program;Related collision prevention parameter is calculated, judges whether that dangerous situation can be generated;If risk of collision will not be generated, speed and direction is kept to advance according to " collision regulation ";If risk of collision can be generated, learn collision prevention strategy with intensified learning method, input data is that the parameter after calculating is trained, and exports the strategy to generate after training, obtains the rudder angle turned needed for this ship;Then implementation strategy, dynamic updates the dynamic data of two ships in step 1, and returns to a reward value;After strategy execution, opportunity of restoring navigation is determined according to " collision regulation " and then is restored navigation.The present invention realizes autonomous learning and the improvement of ship collision prevention, avoids sailor etc. purely by unfavorable situation caused by experience.
Description
Technical field
The invention belongs to field of artificial intelligence, are related to a kind of Ship Intelligent Collision Avoidance method, specifically a kind of based on strong
The Ship Intelligent Collision Avoidance method that chemistry is practised.
Background technique
During navigation, ship collision prevention is the problem of can not ignore, and there are many different solutions, benefits for this problem
With the ship collision prevention intelligent decision based on AIS, ship collision prevention using intelligent algorithm based on evolution genetic algorithm is based on Bayes
Ship collision prevention algorithm of network etc., these algorithms all have it is certain solve the problems, such as the ability of ship collision prevention, but also have them
Limitation, they can not self-teaching and improvement to collision prevention strategy.
Currently, ship is related between more ships in open waters Avoidance, the ship collision prevention mode of existing open waters
Mainly still it is based on《International Regulations for Preventing Collisions at Sea》, currently, because " collision regulation " related evacuation clause is mostly qualitative description, in reality
During the anti-collision behavior of border, Ordinary Practice of Seaman, the practical ship experience etc. of grasping of driver can keep away specific decision scheme and ship
Effect generation is touched to significantly affect.
In actual conditions, ship collision prevention is mainly the manipulation for depending on people, is extremely fixed against usual way and the driver of sailor
Practical behaviour's ship experience, thus have many unstability.
Summary of the invention
In order to solve the above-mentioned technical problem, the present invention realizes the optimization to collision prevention strategy and algorithm using intensified learning, mentions
A kind of Ship Intelligent Collision Avoidance method based on intensified learning has been supplied, autonomous learning and the improvement of ship collision prevention has been realized, avoids
Sailor etc. is purely by unfavorable situation caused by experience.
The technical solution adopted by the present invention to solve the technical problems is:A kind of Ship Intelligent Collision Avoidance based on intensified learning
Method, which is characterized in that include the following steps:
Step 1:Obtain the static data and dynamic data of two ships;
Step 2:The legitimacy of inspection data calculates related collision prevention parameter, judges whether that dangerous situation can be generated, starting is kept away
Touch program;
Step 3:If risk of collision will not be generated, speed and direction is kept to advance according to " collision regulation ";Such as
Fruit can generate risk of collision, then learn collision prevention strategy with intensified learning method, and input data is that the parameter after calculating carries out
Training, exports the strategy to generate after training, obtains the rudder angle turned needed for this ship;
Step 4:The strategy that step 3 generates is executed, then dynamic updates the dynamic data of two ships in step 1, and returns
One reward value;The reward value is used to evaluate the quality of collision prevention strategy;
Step 5:After strategy execution, opportunity of restoring navigation is determined according to " collision regulation " and then is restored navigation.
The invention has the advantages that it uses intensified learning to carry out the optimization of strategy, effective auxiliary operation people
Member reduces the operation error as caused by intuition and experience, effectively improves the efficiency of ship collision prevention, has used engineering
The method of habit enables ship collision prevention self improvement strategy with the process of autonomous learning compared with traditional avoidance algorithm.By plan
After slightly optimizing, it is convenient to by machine learning to optimal policy be supplied to operator reference, make high quality certainly
Plan avoids the generation for more exempting from close quarters situation,.
Detailed description of the invention
Fig. 1 is the schematic diagram of the embodiment of the present invention.
Specific embodiment
Understand for the ease of those of ordinary skill in the art and implement the present invention, with reference to the accompanying drawings and embodiments to this hair
It is bright to be described in further detail, it should be understood that implementation example described herein is merely to illustrate and explain the present invention, not
For limiting the present invention.
In machine learning field, intensified learning is as a kind of artificial intelligence approach, using DeepMind team grinding as representative
Study carefully team and be put forward for the first time the deeply learning method based on DQN (Deep Q-Network), and is swum using the part Atari2600
Play is used as test object, as a result can be more than human player, significant effect.2012, Lange further started to apply, and mentioned
Deep Fitted Q study is gone out for vehicle control.Experiments have shown that this method be suitable for intelligent control, robot and analysis,
The fields such as prediction, new thinking and opportunity are provided to ship collision prevention optimized handling.The present invention can be very good fitting mankind sea
The action of member makes Ship Intelligent Collision Avoidance decision be provided with the characteristics of autonomous learning is with improving.
Referring to Fig.1, a kind of Ship Intelligent Collision Avoidance method based on intensified learning provided by the invention, includes the following steps:
Step 1:Obtain the static data and dynamic data of two ships;
The static data and dynamic data of two ships include this ship information and object ship information;This ship information includes ship shape
State, ship turning indices, ship follow sex index, course made good, stem to, ground speed, speed through water, longitude, latitude, rudder
Angle, drinking water;Object ship information include name of vessel, MMSI, catchword, Ship Types, captain, the beam, course made good, stem to, to ground velocity
Degree, speed through water, longitude, latitude, distance, true bearing, relative bearing.
Step 2:The legitimacy of inspection data calculates related collision prevention parameter, judges whether that dangerous situation can be generated, starting is kept away
Touch program;
Related collision prevention parameter includes time to closest point of approach (TCPA:Time to Closest Point of
Approaching), distance to closest point of approach (DCPA:Distance of Closest Point of Approaching), safety
Meeting distance (SDA:Safety Distance of Approaching), distance of close quarter situation (CQS:Close-quarters
Situation Distance), immediate danger distance (IMD:Immediate Danger Distance), speed of related movement
(VR:Relative Velocity) and direction of relative movement (AR:Relative Angle);
Judge whether that dangerous situation can be generated, works as TCPA>0, and DCPA<When SDA, then risk of collision can be generated.
Step 3:If risk of collision will not be generated, speed and direction is kept to advance according to " collision regulation ";Such as
Fruit can generate risk of collision, then learn collision prevention strategy with intensified learning method, and input data is that the parameter after calculating carries out
Training, exports the strategy to generate after training, obtains the rudder angle turned needed for this ship;
Learn collision prevention strategy with intensified learning method, specific implementation includes following sub-step:
Step 4.1:Input is used to the static parameter and dynamic parameter of the ship of training;
Step 4.2:All kinds of parameters input intensified learning DQN (Deep Q-learning Network) is carried out into training data;
It constantly updates Q value function and obtains best model until Q function convergence;
Step 4.3:The static parameter for the ship for being used to test and dynamic parameter are inputted into trained model;
Step 4.4:Export the rudder angle turned needed for this ship;
The present embodiment passes through the static data and dynamic data, ocean that the equipment such as various kinds of sensors obtain two ships first
Environmental data.A markov decision process four-tuple E=is generated at this time<S,A,P,R>, S is that state set describes ship's head
The speed of a ship or plane, A are that behavior aggregate describes the rudder angle that ship should turn over, For transfer function, state transfer is specified
Probability;For reward functions, award is specified.Existing algorithm generallys use DQN (Deep Q-
Learning Network) carry out training data.Q-Table is initialized first, and row and column is S and A respectively, and the value of Q-Table is used
To measure the quality for the movement a that current state s takes.The present embodiment is updated using Bellman equation in the training process
Q-Table:
Q (s, a)=r+ γ (max (Q (s ', a '))
Q (s a) is expressed as current s and takes the instant r after a, and at a discount the maximum reward max after γ (Q (s ',
a′))。
The present embodiment realizes that Q-Table, input state x export the Q value of different movement a in DQN by neural network.
Its corresponding algorithm is as follows
1. with a deep neural network as the network of Q value, parameter ω;
Q(s,a,ω)≈Qπ(s,a)
2. carrying out objective function objective function using mean square deviation mean-square error in Q value
Namely loss function;
L (ω)=E [(r+ γ maxa, Q (s, a, ω) and-Q (s, a, ω)2)]
Formula is s ' above, and a ' is next state and movement, has used the representation of David Silver here, has seen
Come than more visible.It can be seen that be exactly here used Q-Learning to be updated Q value as target value.There is target value,
There is current value again, then deviation can be calculated by mean square deviation.
3. gradient of the calculating parameter ω about loss function;
4. realizing the optimization aim of End-to-end using SGD;
Gradient above is calculated, andIt is calculated from deep neural network, accordingly, it is possible to use SGD
Stochastic gradient descent carrys out undated parameter, to obtain optimal Q value.
5. acting a with probability ε random selectiontOr the maximum movement a of Q value is selected by the Q value that network exportst, then
To execution atReward r afterwardstWith the input of next network, network calculates the output of subsequent time network further according to current value,
So circulation.
After iteration several times, training, is represented when Q value converges to maximum value and trained good model.It will train
Good model use in the collision prevention of two ships, it can at that time in emergency circumstances predict optimal collision prevention strategy, that is, turn
Rudder angle, assist operators carry out the control of ship, and change ship status terminates until collision prevention behavior.
Step 4:The strategy that step 3 generates is executed, then dynamic updates the dynamic data of two ships in step 1, and returns
One reward value;The reward value is used to evaluate the quality of collision prevention strategy;
Reward value includes minimum track deviation amount, most short evacuation time, most short avoidance path, most short, minimum evacuation amplitude;
The superiority and inferiority of strategy depends on executing the accumulation award obtained after this strategy for a long time, and strategy can be during training by several
After secondary iteration, training, optimization is continuously available when the Q value for representing award converges to maximum value.
Step 5:After strategy execution, opportunity of restoring navigation is determined according to " collision regulation " and then is restored navigation.
It should be understood that the part that this specification does not elaborate belongs to the prior art.
It should be understood that the above-mentioned description for preferred embodiment is more detailed, can not therefore be considered to this
The limitation of invention patent protection range, those skilled in the art under the inspiration of the present invention, are not departing from power of the present invention
Benefit requires to make replacement or deformation under protected ambit, fall within the scope of protection of the present invention, this hair
It is bright range is claimed to be determined by the appended claims.
Claims (5)
1. a kind of Ship Intelligent Collision Avoidance method based on intensified learning, which is characterized in that include the following steps:
Step 1:Obtain the static data and dynamic data of two ships;
Step 2:The legitimacy of inspection data calculates related collision prevention parameter, judges whether that dangerous situation can be generated, starts collision prevention journey
Sequence;
Step 3:If risk of collision will not be generated, speed and direction is kept to advance according to " collision regulation ";If meeting
Risk of collision is generated, then learns collision prevention strategy with intensified learning method, input data is that the parameter after calculating is trained,
Output is the strategy generated after training, obtains the rudder angle turned needed for this ship;
Step 4:The strategy that step 3 generates is executed, then dynamic updates the dynamic data of two ships in step 1, and returns to one
Reward value;The reward value is used to evaluate the quality of collision prevention strategy;
Step 5:After strategy execution, opportunity of restoring navigation is determined according to " collision regulation " and then is restored navigation.
2. the Ship Intelligent Collision Avoidance method according to claim 1 based on intensified learning, it is characterised in that:In step 1, two
The static data and dynamic data of ship include this ship information and object ship information;Described ship information includes ship status, ship
Oceangoing ship turning indices, ship follow sex index, course made good, stem to, ground speed, speed through water, longitude, latitude, rudder angle, eat
Water;The object ship information include name of vessel, MMSI, catchword, Ship Types, captain, the beam, course made good, stem to, to ground velocity
Degree, speed through water, longitude, latitude, distance, true bearing, relative bearing.
3. the Ship Intelligent Collision Avoidance method according to claim 1 based on intensified learning, it is characterised in that:In step 2, institute
Stating related collision prevention parameter includes time to closest point of approach TCPA, distance to closest point of approach DCPA, safe meeting distance SDA, close quarters situation
Distance CQS, immediate danger distance IMD, speed of related movement VR and direction of relative movement AR;
It is described to judge whether that dangerous situation is generated, work as TCPA>0, and DCPA<When SDA, then risk of collision can be generated.
4. the Ship Intelligent Collision Avoidance method according to claim 1 based on intensified learning, it is characterised in that:In step 3, fortune
Learn collision prevention strategy with intensified learning method, specific implementation includes following sub-step:
Step 4.1:Input is used to the static parameter and dynamic parameter of the ship of training;
Step 4.2:All kinds of parameters input intensified learning DQN is carried out into training data;Q value function is constantly updated, until Q function is received
It holds back, obtains best model;
Step 4.3:The static parameter for the ship for being used to test and dynamic parameter are inputted into trained model;
Step 4.4:Export the rudder angle turned needed for this ship.
5. the Ship Intelligent Collision Avoidance method according to any one of claims 1-4 based on intensified learning, it is characterised in that:
In step 4, the reward value includes minimum track deviation amount, most short evacuation time, most short avoidance path, minimum evacuation amplitude;
The superiority and inferiority of strategy depends on executing the accumulation award obtained after this strategy for a long time, and strategy can be during training by several
After secondary iteration, training, optimization is continuously available when the Q value for representing award converges to maximum value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810378954.4A CN108820157B (en) | 2018-04-25 | 2018-04-25 | Intelligent ship collision avoidance method based on reinforcement learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810378954.4A CN108820157B (en) | 2018-04-25 | 2018-04-25 | Intelligent ship collision avoidance method based on reinforcement learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108820157A true CN108820157A (en) | 2018-11-16 |
CN108820157B CN108820157B (en) | 2020-03-10 |
Family
ID=64155045
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810378954.4A Active CN108820157B (en) | 2018-04-25 | 2018-04-25 | Intelligent ship collision avoidance method based on reinforcement learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108820157B (en) |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110203325A (en) * | 2019-06-14 | 2019-09-06 | 上海外高桥造船有限公司 | The test method and system of the collision prevention function of ship autonomous navigation system |
CN110400491A (en) * | 2019-06-10 | 2019-11-01 | 北京海兰信数据科技股份有限公司 | A kind of Open sea area multiple target auxiliary Decision of Collision Avoidance method and decision system |
CN110648556A (en) * | 2019-10-29 | 2020-01-03 | 青岛科技大学 | Ship collision avoidance method based on ship collision avoidance characteristic |
CN111045445A (en) * | 2019-10-23 | 2020-04-21 | 浩亚信息科技有限公司 | Aircraft intelligent collision avoidance method, equipment and medium based on reinforcement learning |
CN111186549A (en) * | 2020-01-15 | 2020-05-22 | 大连海事大学 | Course autopilot control system with ship collision avoidance function |
CN111199272A (en) * | 2019-12-30 | 2020-05-26 | 同济大学 | Adaptive scheduling method for intelligent workshop |
CN111445086A (en) * | 2020-04-17 | 2020-07-24 | 集美大学 | Method for predicting fly-back time based on PIDVCA |
CN112149237A (en) * | 2020-10-15 | 2020-12-29 | 北京海兰信数据科技股份有限公司 | Real-time ship collision avoidance method and system |
CN112180950A (en) * | 2020-11-05 | 2021-01-05 | 武汉理工大学 | Intelligent ship autonomous collision avoidance and path planning method based on reinforcement learning |
CN112650246A (en) * | 2020-12-23 | 2021-04-13 | 武汉理工大学 | Ship autonomous navigation method and device |
WO2021082864A1 (en) * | 2019-10-30 | 2021-05-06 | 武汉理工大学 | Deep reinforcement learning-based intelligent collision-avoidance method for swarm of unmanned surface vehicles |
CN113012473A (en) * | 2021-02-07 | 2021-06-22 | 中电科(宁波)海洋电子研究院有限公司 | Low-cost wave glider marine navigation collision avoidance method |
CN113076338A (en) * | 2021-02-04 | 2021-07-06 | 中国船级社 | Rule-based intelligent ship collision avoidance automatic test scene generation method and system |
CN113093721A (en) * | 2021-02-04 | 2021-07-09 | 中国船级社 | Automatic concurrent ship collision avoidance testing method and system |
CN113221449A (en) * | 2021-04-27 | 2021-08-06 | 中国科学院国家空间科学中心 | Ship track real-time prediction method and system based on optimal strategy learning |
CN113799949A (en) * | 2020-06-11 | 2021-12-17 | 中国科学院沈阳自动化研究所 | AUV buoyancy adjusting method based on Q learning |
CN115107948A (en) * | 2022-06-24 | 2022-09-27 | 大连海事大学 | Efficient reinforcement learning autonomous ship collision avoidance method adopting multiplexing of internal excitation signals and learning experience |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH11175493A (en) * | 1997-12-16 | 1999-07-02 | Nippon Telegr & Teleph Corp <Ntt> | Experience enhancement type enhanced learning method using behavior selection network and record medium recorded with experience enhancement type enhanced learning program |
JP2000080673A (en) * | 1998-09-08 | 2000-03-21 | Ishikawajima Harima Heavy Ind Co Ltd | Route planning method for dredger |
CN105139072A (en) * | 2015-09-09 | 2015-12-09 | 东华大学 | Reinforcement learning algorithm applied to non-tracking intelligent trolley barrier-avoiding system |
CN106842925A (en) * | 2017-01-20 | 2017-06-13 | 清华大学 | A kind of locomotive smart steering method and system based on deeply study |
-
2018
- 2018-04-25 CN CN201810378954.4A patent/CN108820157B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH11175493A (en) * | 1997-12-16 | 1999-07-02 | Nippon Telegr & Teleph Corp <Ntt> | Experience enhancement type enhanced learning method using behavior selection network and record medium recorded with experience enhancement type enhanced learning program |
JP2000080673A (en) * | 1998-09-08 | 2000-03-21 | Ishikawajima Harima Heavy Ind Co Ltd | Route planning method for dredger |
CN105139072A (en) * | 2015-09-09 | 2015-12-09 | 东华大学 | Reinforcement learning algorithm applied to non-tracking intelligent trolley barrier-avoiding system |
CN106842925A (en) * | 2017-01-20 | 2017-06-13 | 清华大学 | A kind of locomotive smart steering method and system based on deeply study |
Non-Patent Citations (2)
Title |
---|
曹江丽: "《水下机器人路径规划问题的关键技术研究》", 《博士学位论文全文库》 * |
段群杰等: "《基于模糊神经网络的水下机器人局部路径规划方法》", 《船舶工程》 * |
Cited By (30)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110400491A (en) * | 2019-06-10 | 2019-11-01 | 北京海兰信数据科技股份有限公司 | A kind of Open sea area multiple target auxiliary Decision of Collision Avoidance method and decision system |
CN110203325A (en) * | 2019-06-14 | 2019-09-06 | 上海外高桥造船有限公司 | The test method and system of the collision prevention function of ship autonomous navigation system |
CN111045445A (en) * | 2019-10-23 | 2020-04-21 | 浩亚信息科技有限公司 | Aircraft intelligent collision avoidance method, equipment and medium based on reinforcement learning |
CN111045445B (en) * | 2019-10-23 | 2023-11-28 | 浩亚信息科技有限公司 | Intelligent collision avoidance method, equipment and medium for aircraft based on reinforcement learning |
CN110648556A (en) * | 2019-10-29 | 2020-01-03 | 青岛科技大学 | Ship collision avoidance method based on ship collision avoidance characteristic |
WO2021082864A1 (en) * | 2019-10-30 | 2021-05-06 | 武汉理工大学 | Deep reinforcement learning-based intelligent collision-avoidance method for swarm of unmanned surface vehicles |
US20220189312A1 (en) * | 2019-10-30 | 2022-06-16 | Wuhan University Of Technology | Intelligent collision avoidance method for a swarm of unmanned surface vehicles based on deep reinforcement learning |
US11990044B2 (en) * | 2019-10-30 | 2024-05-21 | Wuhan University Of Technology | Intelligent collision avoidance method for a swarm of unmanned surface vehicles based on deep reinforcement learning |
CN111199272A (en) * | 2019-12-30 | 2020-05-26 | 同济大学 | Adaptive scheduling method for intelligent workshop |
CN111199272B (en) * | 2019-12-30 | 2023-11-03 | 同济大学 | Self-adaptive scheduling method for intelligent workshops |
CN111186549A (en) * | 2020-01-15 | 2020-05-22 | 大连海事大学 | Course autopilot control system with ship collision avoidance function |
CN111186549B (en) * | 2020-01-15 | 2021-08-17 | 大连海事大学 | Course autopilot control system with ship collision avoidance function |
CN111445086A (en) * | 2020-04-17 | 2020-07-24 | 集美大学 | Method for predicting fly-back time based on PIDVCA |
CN111445086B (en) * | 2020-04-17 | 2022-07-26 | 集美大学 | Method for predicting fly-back time based on PIDVCA |
CN113799949B (en) * | 2020-06-11 | 2022-07-26 | 中国科学院沈阳自动化研究所 | AUV buoyancy adjusting method based on Q learning |
CN113799949A (en) * | 2020-06-11 | 2021-12-17 | 中国科学院沈阳自动化研究所 | AUV buoyancy adjusting method based on Q learning |
CN112149237A (en) * | 2020-10-15 | 2020-12-29 | 北京海兰信数据科技股份有限公司 | Real-time ship collision avoidance method and system |
CN112149237B (en) * | 2020-10-15 | 2024-07-16 | 北京海兰信数据科技股份有限公司 | Ship real-time collision prevention method and system |
CN112180950B (en) * | 2020-11-05 | 2022-07-08 | 武汉理工大学 | Intelligent ship autonomous collision avoidance and path planning method based on reinforcement learning |
CN112180950A (en) * | 2020-11-05 | 2021-01-05 | 武汉理工大学 | Intelligent ship autonomous collision avoidance and path planning method based on reinforcement learning |
CN112650246A (en) * | 2020-12-23 | 2021-04-13 | 武汉理工大学 | Ship autonomous navigation method and device |
CN113093721A (en) * | 2021-02-04 | 2021-07-09 | 中国船级社 | Automatic concurrent ship collision avoidance testing method and system |
CN113076338A (en) * | 2021-02-04 | 2021-07-06 | 中国船级社 | Rule-based intelligent ship collision avoidance automatic test scene generation method and system |
CN113093721B (en) * | 2021-02-04 | 2022-03-01 | 中国船级社 | Automatic concurrent ship collision avoidance testing method and system |
CN113012473A (en) * | 2021-02-07 | 2021-06-22 | 中电科(宁波)海洋电子研究院有限公司 | Low-cost wave glider marine navigation collision avoidance method |
CN113012473B (en) * | 2021-02-07 | 2022-04-19 | 中电科(宁波)海洋电子研究院有限公司 | Low-cost wave glider marine navigation collision avoidance method |
CN113221449B (en) * | 2021-04-27 | 2024-03-15 | 中国科学院国家空间科学中心 | Ship track real-time prediction method and system based on optimal strategy learning |
CN113221449A (en) * | 2021-04-27 | 2021-08-06 | 中国科学院国家空间科学中心 | Ship track real-time prediction method and system based on optimal strategy learning |
CN115107948A (en) * | 2022-06-24 | 2022-09-27 | 大连海事大学 | Efficient reinforcement learning autonomous ship collision avoidance method adopting multiplexing of internal excitation signals and learning experience |
CN115107948B (en) * | 2022-06-24 | 2023-08-25 | 大连海事大学 | Efficient reinforcement learning autonomous ship collision prevention method |
Also Published As
Publication number | Publication date |
---|---|
CN108820157B (en) | 2020-03-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108820157A (en) | A kind of Ship Intelligent Collision Avoidance method based on intensified learning | |
Shen et al. | Automatic collision avoidance of multiple ships based on deep Q-learning | |
Statheros et al. | Autonomous ship collision avoidance navigation concepts, technologies and techniques | |
Xu et al. | Intelligent collision avoidance algorithms for USVs via deep reinforcement learning under COLREGs | |
Xue et al. | Automatic simulation of ship navigation | |
CN110333739A (en) | A kind of AUV conduct programming and method of controlling operation based on intensified learning | |
CN108762281A (en) | It is a kind of that intelligent robot decision-making technique under the embedded Real-time Water of intensified learning is associated with based on memory | |
Liang et al. | Autonomous collision avoidance of unmanned surface vehicles based on improved A star and minimum course alteration algorithms | |
CN111063218A (en) | Ship collision avoidance decision method | |
He et al. | Dynamic adaptive intelligent navigation decision making method for multi-object situation in open water | |
CN110083155A (en) | Machine learning method for realizing ship anthropomorphic intelligent collision avoidance decision | |
Wang et al. | Local path optimization method for unmanned ship based on particle swarm acceleration calculation and dynamic optimal control | |
CN109597417A (en) | A kind of more USV multi-agent synergy collision-avoidance planning methods based on collision prevention criterion | |
Lisowski | Computational intelligence methods of a safe ship control | |
CN113534668B (en) | Maximum entropy based AUV (autonomous Underwater vehicle) motion planning method for actor-critic framework | |
Han et al. | A potential field-based trajectory planning and tracking approach for automatic berthing and COLREGs-compliant collision avoidance | |
Li et al. | Simulation on vessel intelligent collision avoidance based on artificial fish swarm algorithm | |
Han et al. | A COLREGs-compliant guidance strategy for an underactuated unmanned surface vehicle combining potential field with grid map | |
Liu et al. | Reinforcement learning-based collision avoidance: Impact of reward function and knowledge transfer | |
Wang et al. | A distributed model predictive control using virtual field force for multi-ship collision avoidance under COLREGs | |
Amendola et al. | Navigation in restricted channels under environmental conditions: Fast-time simulation by asynchronous deep reinforcement learning | |
Zhang et al. | Dynamic path planning algorithm for unmanned surface vehicle under island-reef environment | |
Yang et al. | Improved reinforcement learning for collision-free local path planning of dynamic obstacle | |
Lee et al. | Ship steering autopilot based on Anfis framework and conditional tuning scheme | |
Lisowski | Game and computational intelligence decision making algorithms for avoiding collision at sea |
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 |