CN109916400A - A kind of unmanned boat barrier-avoiding method combined based on gradient descent algorithm with VO method - Google Patents
A kind of unmanned boat barrier-avoiding method combined based on gradient descent algorithm with VO method Download PDFInfo
- Publication number
- CN109916400A CN109916400A CN201910283660.8A CN201910283660A CN109916400A CN 109916400 A CN109916400 A CN 109916400A CN 201910283660 A CN201910283660 A CN 201910283660A CN 109916400 A CN109916400 A CN 109916400A
- Authority
- CN
- China
- Prior art keywords
- unmanned boat
- speed
- barrier
- gradient
- value
- 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
Abstract
The invention belongs to unmanned boat technical fields, disclose a kind of unmanned boat barrier-avoiding method combined based on gradient descent algorithm with VO method.This method is according to current unmanned boat position, motion information and Obstacle Position, velocity information, judge between unmanned boat and barrier with the presence or absence of risk of collision, risk of collision is such as not present, the speed and rudder angle that the speed and rudder angle exported using unmanned boat Present navigation is exported as avoidance, if there is risk of collision, then using the speed of unmanned boat Present navigation output and rudder angle as initial value, input gradient descent algorithm program, the speed and angle that the output of unmanned boat avoidance is obtained by gradient descent algorithm program make unmanned boat carry out avoidance.This method finds locally optimal solution by gradient descent algorithm, make unmanned boat with more smooth track avoiding obstacles, solves the problems, such as that existing VO method causes the mutation of two near-optimal solution of front and back that unmanned boat unstability is caused to navigate by water during finding optimal solution since the region VO changes.
Description
Technical field
The invention belongs to unmanned boat technical fields, and in particular to a kind of nothing combined based on gradient descent algorithm with VO method
People's ship barrier-avoiding method.
Background technique
Unmanned surface vehicle, abbreviation unmanned boat are a kind of light-duty intelligent surface vehicles, have small in size, low cost, speed
The features such as degree is fast, mobility strong.With control technology, the progress of sensing technology, wireless communication technique, unmanned surface vehicle is obtained
Very big development.By carrying different equipment, unmanned boat can be applied in different fields, for example, when carrying simple beam, more
When the sonar sets such as wave beam, shallow bottom section plotter, can be used for seabed mapping, detect a mine it is antisubmarine etc.;When carrying water quality sampling or inspection
When measurement equipment, it can be used for environment measuring;When carrying weapon, it can be used for Regional patrolling, coast environmental protection, convoy, operation etc.
Task.
To guarantee that unmanned boat can normally and safely navigate by water in ocean, unmanned boat allows for meeting to during navigation
Other barriers such as ship of island, submerged reef, beacon, buoy and the navigation arrived carry out automatic obstacle avoiding.This is unmanned boat navigation
The most important thing, and obstacle avoidance algorithm is the emphasis of safe navigation.So unmanned boat is to have one to the premise of safe navigation
Outstanding obstacle avoidance algorithm.The obstacle avoidance algorithm that unmanned boat generallys use at present has very much, such as Artificial Potential Field Method, Velocity
Obstacle method (Speed Obstacles method, abbreviation VO method) etc., compared with other obstacle avoidance algorithms, VO method is relatively reliable.Before VO method
Mentioning is that the speed for setting barrier and direction are to maintain constant, and the shape size of unmanned boat is integrated to obstacle nitride layer
Then unmanned boat is equivalent to a particle by face, the shape of barrier is considered as circle, i.e., by the collision of barrier and unmanned boat
Problem is equivalent to the contact problems of particle and disk.The core of VO method is exactly in the opposite of the position and barrier for passing through unmanned boat
Displaced position forms a conical region, which is referred to as the region VO, such as according to the integrating shape of barrier and unmanned boat
Shown in Fig. 1, shade fill part is the region VO in Fig. 1, if falling in the area VO relative to the relative velocity of barrier in unmanned boat
In domain, then cause unmanned boat and barrier to collide in the certain time in future, belongs to dangerous speed area, if nobody
Ship is located at except the region VO relative to the relative velocity of barrier, then unmanned boat will not collide with barrier, belongs to safety
Speed area.VO method is a kind of to select the algorithm of optimal solution other than the non-region VO to be based on that is, other than the region VO based on the velocity space
Navigation avoidance task obtains optimal avoidance velocity magnitude and rudder angle, and the optimal velocity size and rudder angle that will acquire are input to nothing
People's ship bottom control module, control unmanned boat are travelled according to the optimal avoidance velocity magnitude and rudder angle of acquisition, avoiding obstacles.
For the unmanned boat in ocean navigation, not only to guarantee the safe avoidance of unmanned boat, but also to guarantee unmanned boat
It is on an even keel, we to guarantee during Real Time Obstacle Avoiding upper layer navigation obstacle avoidance module export to the speed of bottom control module
It spends size and rudder angle is consecutive variations, because if there are wide-angle variations to be easy to cause for adjacent navigation avoidance output twice
The rollover of unmanned boat, and during avoidance, if interruption variation occurs for rudder angle output, it is easy to cause being swung left and right for hull,
So guaranteeing that the stability change of unmanned boat avoidance output is most important for the stability navigation of unmanned boat.VO method is based on VO
Region obtains globally optimal solution (optimal avoidance velocity magnitude and rudder angle), since the region VO that VO method obtains can be with unmanned boat
With variation that the relative position of barrier occurs and change, when the region VO changes, then the overall situation obtained based on the region VO
Optimal solution will also change, and optimal algorithm can find global optimum, and the corresponding minimum more than one of VO method, to draw
It plays the optimal solution that front and back obtains twice and corresponds to different minimums, the optimal solution for causing front and back to obtain twice mutates (before i.e.
Wide variation occurs for the velocity magnitude of avoidance output twice and rudder angle afterwards), and then the unstability of unmanned boat is caused to navigate by water (ginseng
See that Fig. 2, Fig. 2 are scatter chart of the VO method avoidance emulation experiment adjacent two frames moment corresponding cost function in the velocity space,
By the figure we can be found that cost function space there are two extreme points, it is optimal during cost function changes
Value can beat in two extreme points and then lead to the angle and speed jitter of corresponding avoidance output).
Summary of the invention
Aiming at the problems and shortcomings existing in the prior art, the object of the present invention is to provide one kind to be based on gradient descent algorithm
The unmanned boat barrier-avoiding method combined with VO method.
To realize goal of the invention, The technical solution adopted by the invention is as follows:
A kind of unmanned boat barrier-avoiding method combined based on gradient descent algorithm with VO method, comprising the following steps:
(1) obtain current position, movement and the posture information of unmanned boat, obtain the current position of barrier, movement and
Dimension information obtains the speed and rudder angle (v of the output of unmanned boat Present navigationlos,θlos);
(2) judge with the presence or absence of risk of collision between unmanned boat and barrier, if do not deposited between unmanned boat and barrier
In risk of collision, then unmanned boat is continued to navigate by water by the speed and rudder angle of navigation output, if existed between unmanned boat and barrier
Risk of collision then performs the next step operation;
(3) with the speed and rudder angle (v of the output of unmanned boat Present navigationlos,θlos) calculated as the decline of initial value input gradient
Method program, the four direction up and down using gradient descent algorithm program in initial value carry out the iterative search based on gradient,
Initial value is obtained in the reference variable (v of four direction up and downi+1,θj)、(vi-1,θj)、(vi,θj+And (v 1)i,θj-1);It will
In the four reference variable input gradient value function formula obtained, the corresponding gradient value of each reference variable is calculated, is then sentenced
Whether the corresponding gradient value of each reference variable of breaking meets the iterative cycles termination condition of gradient descent algorithm program;Through judging
It was found that the corresponding gradient value of any one reference variable meets iterative cycles termination condition, then iterative cycles are terminated, is changed with meeting
The speed and rudder angle exported for the corresponding reference variable of gradient value of loop termination condition as unmanned boat avoidance is transmitted to nobody
The control module of ship;Find that iterative cycles termination condition is not satisfied in the corresponding gradient value of each reference variable through judgement, then
Using four reference variables as the initial value of gradient descent algorithm program, continue four four, reference variable upper and lower, left and right
Direction carries out the iterative search based on gradient, until finding the gradient value for meeting iterative cycles termination condition.
According to above-mentioned unmanned boat barrier-avoiding method, it is preferable that the step-length of gradient descent algorithm program described in step (3)
Are as follows: velocity magnitude 0.1, angular dimension are 0.5 °;The iterative cycles termination condition of gradient descent algorithm program are as follows: velocity magnitude
Gradient value be less than setting minimum speed Grads threshold 0.01, while directional velocity (directional velocity, that is, rudder angle) gradient value be less than set
Fixed minimum direction gradient threshold value 0.05.
According to above-mentioned unmanned boat barrier-avoiding method, it is preferable that the specific building of gradient value function formula described in step (3)
Step are as follows:
1) velocity space is carried out it is discrete, it is discrete after Speed Two Dimensions variable be (vi,θj), wherein viFor velocity magnitude,
θjFor directional velocity, viRange be [0,20m/s], θjRange be [0,360 °];
2) using the deviation of the Speed Two Dimensions variable of the Speed Two Dimensions variable and unmanned boat of the velocity space navigation output as leading
The cost value of boat task, value interval are [0,0.5];
3) according to Speed Obstacles method, the corresponding cost value of Speed Two Dimensions variable positioned at the region VO is 1, be located at the region VO with
The outer corresponding cost value of Speed Two Dimensions variable is 0;
4) there are flexibilities for the cost value of maritime affairs rule constraint, and according to Speed Obstacles method, there are optimal other than the region VO
Avoidance Speed Two Dimensions variable when, the corresponding Speed Two Dimensions variable of maritime affairs rule constraint is excluded in optimal avoidance Speed Two Dimensions
Except variable, when optimal avoidance Speed Two Dimensions variable being not present except the region VO, with the corresponding speed of maritime affairs rule constraint
Two-dimentional variable is exported as optimal avoidance Speed Two Dimensions variable, therefore, the value area of the corresponding cost value of maritime affairs rule constraint
Between be [0.5,1];
5) according to navigation task, Speed Obstacles method, the cost value value interval and the velocity space of maritime affairs rule constraint three
Speed Two Dimensions range of variables between relationship, building obtain objective cost function calculation formula, as shown in formula (I):
6) by objective cost function J to viAnd θjLocal derviation is sought respectively, obtains corresponding partial derivative, as gradient value function,
The gradient value function includes velocity magnitude gradient value function and directional velocity gradient value function, wherein J pairs of objective cost function
viThe partial derivative for asking local derviation to obtain is velocity magnitude gradient value function, and as shown in formula (II), objective cost function J is to θjAsk inclined
The partial derivative led is directional velocity gradient value function, as shown in formula (III):
According to above-mentioned unmanned boat barrier-avoiding method, it is preferable that judge whether deposit between unmanned boat and barrier in step (2)
In the concrete operations of risk of collision are as follows: calculate the time to closest point of approach TCPA and distance to closest point of approach of unmanned boat and barrier
DCPA, as TCPA≤tmaxAnd DCPA≤dminWhen, wherein tmax、dminIt is known parameter, is deposited between unmanned boat and barrier
In risk of collision, if TCPA, DCPA cannot meet above-mentioned condition simultaneously, there is no collision wind between unmanned boat and barrier
Danger;The calculation formula of described TCPA, DCPA are as follows:
DCPA=| | (PA+vA·TCPA)-(PB+vBTCPA)|| (V)
Wherein, PBAnd PARespectively indicate the position vector of barrier and unmanned boat, vBAnd vAIt is illustrated respectively in earth coordinates
Under barrier and unmanned boat velocity vector.
According to above-mentioned unmanned boat barrier-avoiding method, it is preferable that the concrete operations of step (1) are as follows: sensed by GPS and inertial navigation
Position, movement and the posture information of device acquisition unmanned boat;Ring is carried out by the self-contained Fusion of unmanned boat
Border models to obtain the position of barrier, movement and dimension information, and is modeled as barrier according to the different sizes of barrier
Round barrier of different sizes;The speed and rudder angle (v of the output of unmanned boat Present navigation are obtained according to LOS algorithmlos,
θlos)。
Compared with prior art, what the present invention obtained has the beneficial effect that
(1) present invention finds the optimal avoidance speed and rudder angle of unmanned boat avoidance output, gradient using gradient descent algorithm
Descent algorithm is a kind of algorithm for finding locally optimal solution, gives initial value, and gradient descent algorithm can carry out near initial value
Search obtains the reference variable near initial value, by the way that its gradient is calculated in reference variable input gradient value function formula
Value, then judges whether gradient value meets for nearest local minimum, if it is local minimum, then terminates iterative cycles, will
Optimal velocity and rudder angle that the corresponding reference variable of gradient value is exported as unmanned boat avoidance are transmitted to the control module of unmanned boat,
Unmanned boat is controlled by the optimal velocity and rudder angle traveling of avoidance output, avoiding obstacles.Therefore, gradient descent algorithm passes through searching
Locally optimal solution can make unmanned boat with more smooth track avoiding obstacles, solve existing VO method and find optimal solution
Cause two near-optimal solution of front and back to mutate to cause the unstability of unmanned boat to navigate by water since the region VO changes in the process
Problem;Moreover, gradient descent algorithm principle is simple, control parameter is few, it is easy to accomplish, it can greatly accelerate the search speed of algorithm
Degree.
(2) present invention comprehensively considers the factor of three aspect of unmanned boat navigation task, VO method and maritime affairs rule constraint, constructs
The reasonable objective cost function of unmanned boat avoidance, the advantage of the cost function be to fully consider realistic task feature, and there are poles
Big flexibility, being embodied in navigation task is that avoidance feasible solution obtains foundation, and cost value section is [0,0.5], VO method
Avoidance task is then boolean's characteristic, and cost value non-zero i.e. 1 is in override grade, flexibility is embodied in maritime affairs rule constraint and exists
In the case that feasible solution is not present in navigation area, maritime affairs rule, generation can be suitably violated to a certain extent under not impact conditions
It is presented as that its cost value section is [0.5,1] in valence function;And navigation task is that override obtains cost value, followed by maritime affairs
Rule constraint region.By objective cost function respectively to viAnd θjLocal derviation is sought, velocity magnitude gradient value function and speed side have been obtained
To gradient value function, the objective cost function of two-dimentional variable is become into one-dimensional gradient value function, make to calculate it is simpler, and
Gradient direction is that functional value changes most fast direction, can be improved search speed, most searches optimal solution fastly.
(3) as independent algorithm, the obstacle avoidance algorithm that do not need and navigate is coupled gradient descent algorithm in the present invention, is only needed
Optimal solution can be obtained by gradient descent algorithm by providing gradient value function formula, ensure that the independence and portable of algorithm
Property;And gradient descent algorithm not only can solve the locally optimal solution problem of one-dimensional variable, can also solve the part of multidimensional
Optimal solution problem, so the variation for complicated speed and rudder angle, gradient descent algorithm can also search out optimal solution quickly.
Detailed description of the invention
Fig. 1 is the VO regional distribution chart that VO method is formed;
Fig. 2 is distribution curve of the VO method avoidance emulation experiment adjacent two frames moment corresponding cost function in the velocity space
Figure;
Fig. 3 is the unmanned boat barrier-avoiding method flow chart combined the present invention is based on gradient descent algorithm with VO method;
Fig. 4 is the avoidance simulation result of single barrier;
Fig. 5 is the avoidance simulation result of multi-obstacle avoidance;
The unmanned boat barrier-avoiding method avoidance velocity magnitude and angle that Fig. 6 is combined with VO method the present invention is based on gradient descent algorithm
Degree output change curve.
Specific embodiment
Below by way of specific embodiment, invention is further described in detail, but does not limit the scope of the invention.
Embodiment 1:
A kind of unmanned boat barrier-avoiding method combined based on gradient descent algorithm with VO method, as shown in figure 3, including following step
It is rapid:
(1) position, movement and the posture information of unmanned boat are obtained by GPS and inertial navigation sensor;Pass through unmanned boat itself
The sensors such as radar, vision, laser radar, the sonar of carrying obtain environmental information, carry out data fusion and obtain with environmental modeling
The position of barrier, movement and dimension information, and according to the different sizes of barrier be modeled as barrier of different sizes
Round barrier;The speed and rudder angle (v of the output of unmanned boat Present navigation are obtained according to LOS algorithmlos,θlos)。
(2) judge with the presence or absence of risk of collision between unmanned boat and barrier, if do not deposited between unmanned boat and barrier
In risk of collision, then the speed and rudder angle biography that are exported using the speed of unmanned boat Present navigation output and rudder angle as unmanned boat avoidance
Transport to unmanned boat control module (i.e. unmanned boat is continued to navigate by water by the speed and rudder angle of navigation output);If unmanned boat and barrier
Between there are risk of collision, then perform the next step operation.
Wherein, judge between unmanned boat and barrier whether there is risk of collision concrete operations are as follows: calculate unmanned boat with
The time to closest point of approach TCPA and distance to closest point of approach DCPA of barrier, as TCPA≤tmaxAnd DCPA≤dminWhen, wherein tmax、
dminIt is known parameter, there are risk of collision between unmanned boat and barrier, if TCPA, DCPA cannot meet simultaneously
Condition is stated, then risk of collision is not present between unmanned boat and barrier;The calculation formula of described TCPA, DCPA are as follows:
DCPA=| | (PA+vA·TCPA)-(PB+vBTCPA)|| (V)
PBAnd PARespectively indicate the position vector of barrier and unmanned boat, vBAnd vAIt is illustrated respectively under earth coordinates
The velocity vector of barrier and unmanned boat.
(3) with the speed and rudder angle (v of the output of unmanned boat Present navigationlos,θlos) calculated as the decline of initial value input gradient
Method program, the four direction up and down using gradient descent algorithm program in initial value carry out the iterative search based on gradient,
Initial value is obtained in reference variable (vi+1, the θ of four direction up and downj)、(vi-1,θj)、(vi,θj+ 1) and (vi, θj-1);
In the four reference variable input gradient value function formula that will acquire, the corresponding gradient value of each reference variable is calculated, then
Judge whether the corresponding gradient value of each reference variable meets the iterative cycles termination condition of gradient descent algorithm program;Through sentencing
It is disconnected to find that the corresponding gradient value of any one reference variable meets iterative cycles termination condition and (meets iterative cycles at this time and terminate item
The gradient value of part is the extreme point nearest from initial value, i.e. locally optimal solution), then iterative cycles are terminated, to meet iterative cycles
Speed and rudder angle that the corresponding reference variable of the gradient value of termination condition is exported as unmanned boat avoidance are transmitted to the control of unmanned boat
Molding block, the speed and rudder angle for exporting unmanned boat according to avoidance carry out traveling avoiding obstacles;Through each ginseng of judgement discovery
It examines the corresponding gradient value of variable and iterative cycles termination condition is not satisfied, then using four reference variables as gradient descent algorithm journey
The initial value of sequence continues to carry out the iterative search based on gradient in four reference variable upper and lower, left and right four directions, until looking for
To the gradient value for meeting iterative cycles termination condition.Wherein, the step-length of the gradient descent algorithm program are as follows: velocity magnitude is
0.1, angular dimension is 0.5 °;The iterative cycles termination condition of gradient descent algorithm program is that velocity magnitude gradient value is less than setting
Minimum speed Grads threshold 0.01, while directional velocity (directional velocity, that is, rudder angle) gradient value be less than setting minimum direction ladder
Spend threshold value 0.05.
The specific construction step of the gradient value function formula are as follows:
1) velocity space is carried out it is discrete, it is discrete after Speed Two Dimensions variable be (vi,θj), wherein viFor velocity magnitude,
θjFor directional velocity, since the dynamics of unmanned boat limits, viRange be 0~20m/s, θjRange be 0~360 °;
2) using the deviation of the Speed Two Dimensions variable of the Speed Two Dimensions variable and unmanned boat of the velocity space navigation output as leading
The cost value of boat task, value interval are [0,0.5];
3) according to Speed Obstacles method, the corresponding cost value of Speed Two Dimensions variable positioned at the region VO is 1, be located at the region VO with
The outer corresponding cost value of Speed Two Dimensions variable is 0;
4) there are flexibilities for the cost value of maritime affairs rule constraint, and according to Speed Obstacles method, there are optimal other than the region VO
Avoidance Speed Two Dimensions variable when, the corresponding Speed Two Dimensions variable of maritime affairs rule constraint is excluded in optimal avoidance Speed Two Dimensions
(the corresponding Speed Two Dimensions variable of maritime affairs rule constraint is not considered) except variable, and optimal avoidance is not present except the region VO
It is defeated using the corresponding Speed Two Dimensions variable of maritime affairs rule constraint as optimal avoidance Speed Two Dimensions variable when Speed Two Dimensions variable
Out, therefore, the value interval of the corresponding cost value of maritime affairs rule constraint is [0.5,1];
5) according to navigation task, Speed Obstacles method avoidance task, the cost value value interval of maritime affairs rule constraint three and
Relationship between the Speed Two Dimensions range of variables of the velocity space, building obtains objective cost function calculation formula, such as formula (I) institute
Show:
6) by objective cost function J to viAnd θjLocal derviation is sought respectively, obtains corresponding partial derivative, as gradient value function,
The gradient value function includes velocity magnitude gradient value function and directional velocity gradient value function, wherein J pairs of objective cost function
viThe partial derivative for asking local derviation to obtain is velocity magnitude gradient value function, and as shown in formula (II), objective cost function J is to θjAsk inclined
The partial derivative led is directional velocity gradient value function, as shown in formula (III):
(4) avoidance simulated effect figure is obtained, while printing avoidance output curve diagram, passes through curve in avoidance output curve diagram
The stability and validity of the continuity verification gradient descent algorithm avoidance of variation.Fig. 4 is that unmanned boat is kept away under single barrier scene
(ordinate and abscissa indicate distance, unit m), as seen from Figure 4, in single barrier to the effect picture of barrier emulation in figure
Under scene, using barrier-avoiding method of the invention it is possible to prevente effectively from the collision of unmanned boat and barrier.Fig. 5 is multi-obstacle avoidance scene
Lower unmanned boat avoidance simulated effect figure (ordinate and abscissa indicate distance in figure, unit m), as seen from Figure 5,
Under the complex scene of multi-obstacle avoidance, barrier-avoiding method of the invention still can be realized the avoidance of unmanned boat.Fig. 6 is unmanned boat avoidance
The change curve of output speed and the change curve of avoidance output angle, as seen from Figure 6, using avoidance of the invention
Method, unmanned boat avoidance output speed change curve and angle change curve are continuous linearity curve, are thus illustrated, of the invention
Barrier-avoiding method can make unmanned boat with more smooth track avoiding obstacles, solve existing VO method and finding optimal solution preocess
In due to the region VO change cause two near-optimal solution of front and back mutate cause unmanned boat unstability navigate by water the problem of.
Claims (5)
1. a kind of unmanned boat barrier-avoiding method combined based on gradient descent algorithm with VO method, which is characterized in that including following step
It is rapid:
(1) unmanned boat current position, movement and posture information are obtained, barrier current position, movement and size are obtained
Information obtains the speed and rudder angle (v of the output of unmanned boat Present navigationlos,θlos);
(2) judge with the presence or absence of risk of collision between unmanned boat and barrier, if there is no touch between unmanned boat and barrier
Risk is hit, unmanned boat is continued to navigate by water by the speed and rudder angle of navigation output, if there is collision wind between unmanned boat and barrier
Danger, then perform the next step operation;
(3) with the speed and rudder angle (v of the output of unmanned boat Present navigationlos,θlos) it is used as initial value input gradient descent algorithm journey
Sequence carries out the iterative search based on gradient in the four direction up and down of initial value using gradient descent algorithm program, obtains
Reference variable (v of the initial value in four direction up and downi+1,θj)、(vi-1,θj)、(vi,θj+1) and (vi,θj-1);It will acquire
Four reference variable input gradient value function formula in, calculate the corresponding gradient value of each reference variable, then judgement is every
Whether the corresponding gradient value of one reference variable meets the iterative cycles termination condition of gradient descent algorithm program;It is found through judgement
The corresponding gradient value of any one reference variable meets iterative cycles termination condition, then terminates iterative cycles, is followed with meeting iteration
The speed and rudder angle that the corresponding reference variable of the gradient value of ring termination condition is exported as unmanned boat avoidance are transmitted to unmanned boat
Control module;Find that iterative cycles termination condition is not satisfied in the corresponding gradient value of each reference variable through judgement, then with four
Initial value of a reference variable as gradient descent algorithm program continues in four reference variable upper and lower, left and right four directions
The iterative search based on gradient is carried out, until finding the gradient value for meeting iterative cycles termination condition.
2. unmanned boat barrier-avoiding method according to claim 1, which is characterized in that gradient descent algorithm described in step (3)
The step-length of program are as follows: velocity magnitude 0.1, angular dimension are 0.5 °;The iterative cycles termination condition of gradient descent algorithm program
Are as follows: velocity magnitude gradient value is less than the minimum speed Grads threshold 0.01 of setting, while directional velocity gradient value is less than setting
Minimum direction gradient threshold value 0.05.
3. unmanned boat barrier-avoiding method according to claim 2, which is characterized in that gradient value function described in step (3) is public
The specific construction step of formula are as follows:
1) velocity space is carried out it is discrete, it is discrete after Speed Two Dimensions variable be (vi,θj), wherein viFor velocity magnitude, θjFor speed
Spend direction, viRange be [0,20m/s], θjRange be [0,360 °];
2) deviation of the Speed Two Dimensions variable of the Speed Two Dimensions variable and unmanned boat of the velocity space navigation output is appointed as navigation
The cost value of business, value interval are [0,0.5];
3) according to Speed Obstacles method, the corresponding cost value of Speed Two Dimensions variable positioned at the region VO is 1, other than the region VO
The corresponding cost value of Speed Two Dimensions variable is 0;
4) according to Speed Obstacles method, other than the region VO there are when optimal avoidance Speed Two Dimensions variable, by maritime affairs rule constraint
Corresponding Speed Two Dimensions variable excludes except optimal avoidance Speed Two Dimensions variable;It is kept away except the region VO there is no optimal
It is defeated using the corresponding Speed Two Dimensions variable of maritime affairs rule constraint as optimal avoidance Speed Two Dimensions variable when hindering Speed Two Dimensions variable
Out, therefore, the value interval of the corresponding cost value of maritime affairs rule constraint is [0.5,1];
5) according to navigation task, Speed Obstacles method, the speed of the cost value value interval of maritime affairs rule constraint three and the velocity space
The relationship between two-dimentional range of variables is spent, building obtains objective cost function calculation formula, as shown in formula (I):
6) by objective cost function J to viAnd θjLocal derviation is sought respectively, obtains corresponding partial derivative, as gradient value function, it is described
Gradient value function includes velocity magnitude gradient value function and directional velocity gradient value function, wherein objective cost function J is to viIt asks
The partial derivative that local derviation obtains is velocity magnitude gradient value function, and as shown in formula (II), objective cost function J is to θjLocal derviation is asked to obtain
The partial derivative arrived is directional velocity gradient value function, as shown in formula (III):
4. unmanned boat barrier-avoiding method according to claim 3, which is characterized in that judge unmanned boat and obstacle in step (2)
It whether there is the concrete operations of risk of collision between object are as follows:
The time to closest point of approach TCPA and distance to closest point of approach DCPA for calculating unmanned boat and barrier, as TCPA≤tmaxAnd DCPA
≤dminWhen, wherein tmax、dminIt is known parameter, there are risk of collision between unmanned boat and barrier, if TCPA,
DCPA cannot meet above-mentioned condition simultaneously, then risk of collision is not present between unmanned boat and barrier;The meter of described TCPA, DCPA
Calculate formula are as follows:
DCPA=| | (PA+vA·TCPA)-(PB+vBTCPA)|| (V)
Wherein, PBAnd PARespectively indicate the position vector of barrier and unmanned boat, vBAnd vAIt is illustrated respectively under earth coordinates
The velocity vector of barrier and unmanned boat.
5. unmanned boat barrier-avoiding method according to claim 4, which is characterized in that the concrete operations of step (1) are as follows:
Position, movement and the posture information of unmanned boat are obtained by GPS and inertial navigation sensor;Pass through self-contained more of unmanned boat
Data Fusion of Sensor carries out environmental modeling and obtains the position of barrier, movement and dimension information, and not according to barrier
Barrier is modeled as to round barrier of different sizes with size;The output of unmanned boat Present navigation is obtained according to LOS algorithm
Speed and rudder angle (vlos,θlos)。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910283660.8A CN109916400B (en) | 2019-04-10 | 2019-04-10 | Unmanned ship obstacle avoidance method based on combination of gradient descent algorithm and VO method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910283660.8A CN109916400B (en) | 2019-04-10 | 2019-04-10 | Unmanned ship obstacle avoidance method based on combination of gradient descent algorithm and VO method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109916400A true CN109916400A (en) | 2019-06-21 |
CN109916400B CN109916400B (en) | 2020-08-25 |
Family
ID=66969288
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910283660.8A Active CN109916400B (en) | 2019-04-10 | 2019-04-10 | Unmanned ship obstacle avoidance method based on combination of gradient descent algorithm and VO method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109916400B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111984006A (en) * | 2020-07-24 | 2020-11-24 | 哈尔滨工程大学 | Unmanned ship multi-target meeting collision avoidance method integrating ocean current and scale difference influences |
CN112306090A (en) * | 2020-10-26 | 2021-02-02 | 中国人民解放军军事科学院国防科技创新研究院 | Unmanned aerial vehicle instant rapid obstacle avoidance method based on relative displacement and velocity vector synthesis |
CN114167880A (en) * | 2021-12-02 | 2022-03-11 | 大连海事大学 | Time-optimal-based multi-underwater glider path planning system |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6462739B1 (en) * | 1999-03-18 | 2002-10-08 | Abode Systems Incorporated | Curve smoothing without shrinking |
US8849483B2 (en) * | 2011-04-13 | 2014-09-30 | California Institute Of Technology | Target trailing with safe navigation with colregs for maritime autonomous surface vehicles |
JP2015157612A (en) * | 2014-02-25 | 2015-09-03 | 株式会社デンソー | vehicle behavior control device |
CN105589464A (en) * | 2016-03-28 | 2016-05-18 | 哈尔滨工程大学 | UUV dynamic obstacle avoidance method based on speed obstruction method |
US20160299507A1 (en) * | 2015-04-08 | 2016-10-13 | University Of Maryland, College Park | Surface vehicle trajectory planning systems, devices, and methods |
WO2017129863A1 (en) * | 2016-01-29 | 2017-08-03 | Rolls-Royce Oy Ab | Autonomous operation of a vessel |
CN108073176A (en) * | 2018-02-10 | 2018-05-25 | 西安交通大学 | A kind of modified D*Lite vehicle dynamic path planing methods |
CN108664020A (en) * | 2018-04-11 | 2018-10-16 | 上海大学 | A kind of unmanned boat dynamic obstacle avoidance algorithm based on Speed Obstacles method and dynamic window method |
CN109298712A (en) * | 2018-10-19 | 2019-02-01 | 大连海事大学 | A kind of autonomous Decision of Collision Avoidance method of unmanned ship based on the study of adaptive sailing situation |
CN109543225A (en) * | 2018-10-19 | 2019-03-29 | 东软集团股份有限公司 | Control program generation method, device, storage medium and the electronic equipment of vehicle |
-
2019
- 2019-04-10 CN CN201910283660.8A patent/CN109916400B/en active Active
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6462739B1 (en) * | 1999-03-18 | 2002-10-08 | Abode Systems Incorporated | Curve smoothing without shrinking |
US8849483B2 (en) * | 2011-04-13 | 2014-09-30 | California Institute Of Technology | Target trailing with safe navigation with colregs for maritime autonomous surface vehicles |
JP2015157612A (en) * | 2014-02-25 | 2015-09-03 | 株式会社デンソー | vehicle behavior control device |
US20160299507A1 (en) * | 2015-04-08 | 2016-10-13 | University Of Maryland, College Park | Surface vehicle trajectory planning systems, devices, and methods |
WO2017129863A1 (en) * | 2016-01-29 | 2017-08-03 | Rolls-Royce Oy Ab | Autonomous operation of a vessel |
CN105589464A (en) * | 2016-03-28 | 2016-05-18 | 哈尔滨工程大学 | UUV dynamic obstacle avoidance method based on speed obstruction method |
CN108073176A (en) * | 2018-02-10 | 2018-05-25 | 西安交通大学 | A kind of modified D*Lite vehicle dynamic path planing methods |
CN108664020A (en) * | 2018-04-11 | 2018-10-16 | 上海大学 | A kind of unmanned boat dynamic obstacle avoidance algorithm based on Speed Obstacles method and dynamic window method |
CN109298712A (en) * | 2018-10-19 | 2019-02-01 | 大连海事大学 | A kind of autonomous Decision of Collision Avoidance method of unmanned ship based on the study of adaptive sailing situation |
CN109543225A (en) * | 2018-10-19 | 2019-03-29 | 东软集团股份有限公司 | Control program generation method, device, storage medium and the electronic equipment of vehicle |
Non-Patent Citations (4)
Title |
---|
FU MING-YU等: ""Multi-Behavior Fusion Based Potential Field Method for Path Planning of Unmanned Surface Vessel"", 《CHINA OCEAN ENGINEERING》 * |
WANG, CHUNXIN等: ""A decoupling controller by hierarchical backstepping method for straight-line tracking of unmanned surface vehicle"", 《SYSTEMS SCIENCE & CONTROL ENGINEERING》 * |
吴博等: ""基于速度障碍原理的无人艇自动避碰算法"", 《大连海事大学学报》 * |
蒲华燕等: ""基于椭圆碰撞锥的无人艇动态避障方法"", 《仪器仪表学报》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111984006A (en) * | 2020-07-24 | 2020-11-24 | 哈尔滨工程大学 | Unmanned ship multi-target meeting collision avoidance method integrating ocean current and scale difference influences |
CN111984006B (en) * | 2020-07-24 | 2021-07-06 | 哈尔滨工程大学 | Unmanned ship multi-target meeting collision avoidance method integrating ocean current and scale difference influences |
CN112306090A (en) * | 2020-10-26 | 2021-02-02 | 中国人民解放军军事科学院国防科技创新研究院 | Unmanned aerial vehicle instant rapid obstacle avoidance method based on relative displacement and velocity vector synthesis |
CN114167880A (en) * | 2021-12-02 | 2022-03-11 | 大连海事大学 | Time-optimal-based multi-underwater glider path planning system |
Also Published As
Publication number | Publication date |
---|---|
CN109916400B (en) | 2020-08-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Wang et al. | Roboat: An autonomous surface vehicle for urban waterways | |
Liu et al. | Self-adaptive dynamic obstacle avoidance and path planning for USV under complex maritime environment | |
Mousazadeh et al. | Developing a navigation, guidance and obstacle avoidance algorithm for an Unmanned Surface Vehicle (USV) by algorithms fusion | |
CN110850403B (en) | Multi-sensor decision-level fused intelligent ship water surface target feeling knowledge identification method | |
CN107748561B (en) | Unmanned ship local obstacle avoidance system and method based on multiple sensing parameters | |
Stutters et al. | Navigation technologies for autonomous underwater vehicles | |
Braginsky et al. | Obstacle avoidance approaches for autonomous underwater vehicle: Simulation and experimental results | |
CN108445879A (en) | A kind of unmanned boat barrier-avoiding method based on prediction collision risk region | |
CN108681321A (en) | A kind of undersea detection method that unmanned boat collaboration is formed into columns | |
CN105759829A (en) | Laser radar-based mini-sized unmanned plane control method and system | |
CN107037809A (en) | A kind of unmanned boat collision prevention method based on improvement ant group algorithm | |
CN109916400A (en) | A kind of unmanned boat barrier-avoiding method combined based on gradient descent algorithm with VO method | |
TWI725677B (en) | Autonomous vessel simulation system and operating method thereof | |
CN104881045A (en) | Bionic robot fish three-dimensional tracking method based on embedded visual guidance | |
CN109765914A (en) | A kind of unmanned surface vehicle collision prevention method based on sliding window population | |
Zhang et al. | NavNet: AUV navigation through deep sequential learning | |
CN111090283B (en) | Unmanned ship combined positioning and orientation method and system | |
Lan et al. | Improved RRT algorithms to solve path planning of multi-glider in time-varying ocean currents | |
Shan et al. | LiDAR-based stable navigable region detection for unmanned surface vehicles | |
Tang et al. | OdoNet: Untethered speed aiding for vehicle navigation without hardware wheeled odometer | |
Feng et al. | Automatic tracking method for submarine cables and pipelines of AUV based on side scan sonar | |
CN112947438B (en) | AUV (autonomous Underwater vehicle) online path planning method based on full-oscillation type invasive weed optimization algorithm | |
Li et al. | Virtual-reality-based online simulator design with a virtual simulation system for the docking of unmanned underwater vehicle | |
Sang et al. | An autonomous underwater vehicle simulation with fuzzy sensor fusion for pipeline inspection | |
Hyland et al. | Mine avoidance techniques for underwater vehicles |
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 |