CN113190017A - Harvesting robot operation path planning method based on improved ant colony algorithm - Google Patents

Abstract

The invention discloses a method for planning a working path of a harvesting robot based on an improved ant colony algorithm, comprising the steps of: 1. Establishing a mathematical model of an irregular quadrilateral farmland, and determining the most 2. Abstract the full-coverage path planning of farmland into a vehicle routing problem (VRP), and establish a corresponding VRP model according to the distribution of different unloading positions; 3. According to the harvester capacity, total driving distance, and full-load driving distance 4. According to the traversal order of the operation row and the farmland model, solve the expression of each path, and generate the full coverage path of the farmland, which is the harvester The path tracing provides reference. This method can design the full-coverage path of the farmland with the minimum full-load driving distance according to the distribution of different grain unloading positions.

Description

Path planning method for harvesting robot based on improved ant colony algorithm

technical field

The invention belongs to the field of intelligent harvester operating path planning, in particular to a harvesting robot operating path planning method based on an improved ant colony algorithm, and in particular relates to a harvester granary capacity, total travel distance, full-load travel distance and unloading position distribution constraints conditions, using the improved ant colony algorithm to design the method of full coverage path.

Background technique

The path planning of full coverage of farmland is a key technology to realize the autonomous operation of harvester, which can provide a reasonable path for harvester field operation, effectively improve the problem of repeated operation and missing operation, and improve the operation efficiency of harvester. At present, there is no mature and general method for full coverage path planning of farmland. In practical applications, there are two main ways to plan the full coverage path of farmland: one is to specify the initial working path by the driver, and to achieve full coverage of the farmland by constantly shifting it; In the farmland environment, such as rectangular farmland, the operation is carried out by the shuttle method or the spiral method. The above two path planning methods have low intelligence, adaptability to different farmland and operating efficiency. Therefore, it is necessary to study the work path planning method.

By analyzing the current research status of the full coverage path planning algorithm at home and abroad, the optimization goals such as the driving distance, the proportion of effective operation area, operation time, and energy consumption are used to design the optimal operation by using optimization algorithms such as greedy algorithm and ant colony algorithm. path. The full-load travel distance and the distribution of unloading positions have a great influence on the planning of agricultural machinery operation paths, and there are relatively few specific studies that comprehensively consider the above two factors. Therefore, when planning a full-coverage path, the following two factors need to be considered: the weight of the large harvester is large and the rolling of the land is large, so the driving distance of the full-load harvester in the field should be reduced; in actual operation, there are intermittent Grain unloading method, the grain truck cannot enter the field and can only park at the edge of the field. It is necessary to fully consider the location distribution of the unloading point and design different coverage paths independently.

SUMMARY OF THE INVENTION

Aiming at the problems of low effectiveness and weak generality of harvester autonomous path planning, the present invention provides a harvesting robot operation path planning method based on improved ant colony algorithm, establishes a farmland model, and constructs operation lines and unloading positions according to the distribution of five kinds of unloading positions. The distance model of grain location, with the harvester’s granary capacity, total travel distance, and full-load travel distance as constraints, uses the improved ant colony algorithm to solve the traversal sequence of the operation row, plan the operation path, reduce the degree of rolling on the land, and improve the path. The adaptability of the planning algorithm to the distribution of different unloading locations.

The present invention provides a method for planning a working path of a harvesting robot based on an improved ant colony algorithm, which is characterized by comprising the following steps:

Step 1. Establish a farmland model;

Let the four vertices of the farmland be A, B, C and D, and the coordinates of the four points are known, and the mathematical expressions of the boundaries are solved respectively, and the preliminary modeling of the farmland is completed. Determine the working direction according to the condition of the maximum degree, take the side and diagonal of the quadrilateral as the working direction respectively, and solve the required number of turns;

When AD is the working direction, and the cutting width of the harvester is l _{cut_width} , then a series of parallel working lines are expressed as

Among them, k _AD is the slope of AD side, b _AD is the intercept of AD side, and i is the sequence number of the job row;

Solve the intersection points of the group of parallel lines and the quadrilateral boundary respectively, if there are two intersection points, add one to the number of turns, until there is no intersection, then keep the current number of turns as the result of this work direction;

In the same way, take AB, BC, CD, AC, and BD as the working directions, respectively, to solve the corresponding turning times. By comparison, the working direction with the fewest turning times is the desired one. If there are multiple sets of working directions that satisfy the minimum turning times. , then compare the included angle between the working direction and the boundary. Since the turning complexity and turning distance are small in the vertical case, the working direction with an included angle close to 90° is selected. Finally, the mathematical expression of each working row is obtained by solving and improving the farmland. Model;

Step 2. Abstract the farmland full coverage path planning as a vehicle routing problem, and establish a corresponding VRP model according to the distribution of different grain unloading locations;

Step 3. According to the harvester capacity, total driving distance, full-load driving distance and unloading position distribution constraints, the improved ant colony algorithm is used to design the optimal traversal sequence of job rows;

Step 4. According to the traversal sequence of the job row and the farmland model, the expressions of each path are solved to generate a farmland full coverage path, which provides a reference for the path tracking of the harvester.

Further, the step 2 abstracts the farmland full coverage path planning as a vehicle route problem, and establishes a corresponding VRP model according to the distribution of different grain unloading positions, including the following steps:

(2.1) Define five distributions of grain unloading positions:

The research is based on the intermittent grain unloading method, and the grain truck cannot drive in the farmland and needs to be parked on the side of the road. According to the location and number of unloading grains, it is divided into the following five situations:

"a": There is only one unloading position S ₁ , at the beginning or end of the operation line;

"b": There is only one unloading position S ₁ , on the side of the road parallel to the operation row;

"c": There are two grain unloading positions S ₁ and S ₂ , which are at the head and tail of the operation line respectively;

"d": There are two grain unloading positions S ₁ and S ₂ , which are respectively on the two sides of the road parallel to the operation row;

"e": There are two grain unloading positions S ₁ , S ₂ , one is at the head or tail of the operation row, and the other is on the road parallel to the operation row;

(2.2) Establish a VRP model:

到作业行尾端的距离为

卸粮点S₂到作业行首端的距离为

到作业行尾端的距离为

作业行之间的转弯距离为

根据五种卸粮位置分布情况，分别建立作业行与卸粮位置间的距离模型如下：Full-coverage path planning of farmland refers to planning the operation path of the harvester so that it traverses the entire farmland, and each operation row only passes through once, and the loading capacity of the harvester cannot exceed its own capacity. To unload the grain at the unloading position, let the set of path points be E, which includes the start point i _up and the tail point _{i down} of each operation line, the unloading positions S ₁ , S ₂ , and each operation line has lengths li and There are two attributes of the grain volume V _i that can be harvested. The harvester starts from the unloading position, traverses each operation line in a certain order, and finally returns to the unloading position. Let the capacity of the harvester be V _granary , and the serial numbers of the operation lines are i, j, where i,j∈[1,n _line ], the distance from the unloading point S ₁ to the beginning of the operation line is

The distance to the end of the job line is

The distance from the unloading point S2 to the _beginning of the operation row is

The distance to the end of the job line is

The turn distance between job rows is

According to the distribution of five grain unloading positions, the distance models between the operation line and the grain unloading position are respectively established as follows:

(2.2.1) The harvester is unloaded at the beginning of the job row:

Among them, l _i is the length of the i-th job line;

(2.2.2) The harvester is unloaded on the side parallel to the work row:

Among them, w is the width of the work row, θ _up is the angle between the upper boundary and the work row, and θ _down is the angle between the lower boundary and the work row;

(2.2.3) The harvester can be unloaded at the beginning and end of the row:

(2.2.4) The harvester can be unloaded on both sides parallel to the work row:

Among them, n _line is the number of job lines;

(2.2.5) The harvester is unloaded at the beginning and left of the row:

为The turn distance between job rows is

for

Among them, S _T is the turning distance of the T-turn, S _Ω is the turning distance of the Ω-turn, and S _U is the turning distance of the U-turn.

Further, the step 3 adopts the improved ant colony algorithm to design the optimal traversal sequence of work rows according to the harvester capacity, total travel distance, full load travel distance and unloading position distribution constraints, which specifically includes the following steps:

(3.1) Ant colony algorithm initialization:

Set the size of the ant colony as M, the importance factor of the legacy pheromone as α, the importance factor of the heuristic function as β, the pheromone volatility factor as ρ, the total amount of pheromone released by ants in one iteration as Q, and the maximum number of iterations as I _max ;

距离越小，函数值越大；In the first iteration, the initial pheromone content on each path is the same. The ant colony starts from the unloading point and selects a reasonable path according to the pheromone content and heuristic function of each path. The heuristic function of each path pq is

The smaller the distance, the larger the function value;

(3.2) Path selection:

In the calculation of the nth iteration, the probability of the mth ant transferring from the current point p to the next point q is:

Among them, τ _pq (n) is the pheromone of the path pq in the nth iteration;

并记录总路径长度

通过对比，获得最小的总路径长度

以及对应的路径

Select the path with the largest transition probability as the target path, and judge whether it exceeds the capacity constraint. If it exceeds, the point will not be added, and the next target point will continue to be found; if not, it will be added to the current path. Repeat this step until all job lines are traversed, complete one iteration, and record the path node of each ant

and record the total path length

By comparison, the minimum total path length is obtained

and the corresponding path

(3.3) Adjust the route length based on the weight weight factor:

The accumulation of pheromone is directly related to the length of the route. By changing the length of the route and changing the content of pheromone, the optimization process of the ant colony algorithm is affected. Through the optimization algorithm, the feasible solution that makes the function reach the minimum value is obtained. In addition to considering the total driving distance, the full load distance constraint is added. When the ants pass through a work row, the volume of the harvested grain is updated, and the value is related to the path's value. The length is associated to guide the ants to optimize according to the full load distance constraint, and the distance of the path is improved as follows:

Among them, weight _mass is the weight weight factor, V _now is the currently harvested grain volume, V _granary is the _granary capacity, and km is the scale factor. The distance will be greater than the real distance, thereby reducing the pheromone of the path, inducing the ants to find a shorter path under full load;

(3.4) Update pheromone:

Every time an ant passes a path, it will leave a pheromone on the path. At the next iteration, the pheromone content of the path pq is

τ _pq (n+1)=(1-ρ)τ _pq (n)+Δτ _pq

Among them, (1-ρ) τ _pq (n) is the remaining pheromone after volatilization, and Δτ _pq is the pheromone left by all ants passing through this path in this iteration:

Among them, the pheromone content left by each ant is:

The value is determined by the distance of the path passed, the shorter the distance, the higher the pheromone content;

(3.5) Determine whether to terminate the iteration:

If the end condition of the ant colony algorithm is satisfied, the path traversed by the ants in this iteration is the traversal sequence of the job row, and go to step 4;

If the end condition of the ant colony algorithm is not satisfied, set the number of iterations i=i+1, and jump to step (3.2) to continue searching for the next path.

Further, the end condition of the ant colony algorithm is:

The current number of iterations i>I _max , or the optimal solution meets the accuracy requirements:

是第i次迭代后，得到的最短路径长度，ξ是预设的精度阈值。in,

is the shortest path length obtained after the ith iteration, and ξ is the preset accuracy threshold.

Further, the step 4 solves the expression of each path according to the traversal sequence of the job row and the farmland model, and generates a farmland full coverage path, which provides a reference for the path tracking of the harvester, as follows:

Using step 3, according to the distribution constraints of harvester capacity, total driving distance, full-load driving distance and unloading position, the improved ant colony algorithm is used to design the optimal traversal order of job rows, and each job obtained in step 1 is solved by using the traversal order. The row expressions are linked together to form a path expression covering the entire field.

Beneficial effects: Compared with the prior art, the method for planning the working path of the harvesting robot based on the improved ant colony algorithm disclosed in the present invention has the following advantages: five kinds of grain unloading position distributions are proposed, and the operation row and grain unloading are correspondingly established. The distance model of the location improves the adaptability of the path planning algorithm to different operating environments; the ant colony algorithm is used for path search, which has strong robustness and stronger global search ability; the "weight weight factor" is used to adjust the path length , the improvement of the ant colony algorithm can reduce the full-load driving distance, reduce the degree of rolling of the land by the harvester, and improve the operation efficiency.

Description of drawings

FIG. 1 is a flowchart of a method for planning a working path of a harvesting robot based on an improved ant colony algorithm disclosed in the present invention.

Detailed ways

The present invention will be described in further detail below in conjunction with the accompanying drawings and specific embodiments:

The invention provides an operation path planning method for a harvesting robot based on an improved ant colony algorithm. A farmland model is established. According to the distribution of five kinds of grain unloading positions, a distance model between the operation row and the unloading position is constructed. The full-load driving distance is the constraint condition. The improved ant colony algorithm is used to solve the traversal sequence of the operation row, plan the operation path, reduce the degree of rolling of the land, and improve the adaptability of the path planning algorithm to the distribution of different unloading positions.

As shown in FIG. 1 , the present invention discloses a method for planning a working path of a harvesting robot based on an improved ant colony algorithm, which includes the following steps:

Step 1. Establish a farmland model:

Let the four vertices of the farmland be A, B, C, and D. Knowing the coordinates of the four points, the mathematical expressions of the boundaries can be solved respectively, and the preliminary modeling of the farmland can be completed. The working direction is determined on the condition that the number of turns is the least and the perpendicularity between the working row and the boundary is the maximum, and the sides and diagonals of the quadrilateral are used as the working direction to solve the required number of turns.

When AD is the working direction, set the cutting width of the harvester as l _{cut_width} , then a series of parallel working lines can be expressed as

Among them, k _AD is the slope of AD side, b _AD is the intercept of AD side, and i is the sequence number of the job row;

Solve the intersection of the set of parallel lines and the quadrilateral boundary respectively. If there are two intersections, add one to the number of turns until there is no intersection, then keep the current number of turns as the result of this job direction.

In the same way, take AB, BC, CD, AC, and BD as the working directions to solve the corresponding turning times. By comparison, the working direction with the least number of turns is the desired one. If there are multiple sets of working directions that satisfy the minimum number of turns, compare the angle between the working direction and the boundary. Since the turning complexity and turning distance are small in the vertical case, the working direction with an included angle close to 90° is selected. Finally, the mathematical expression of each operation row is obtained by solving, and the farmland model is perfected.

Step 2. Abstract the full-coverage path planning of farmland into a vehicle routing problem, and establish a corresponding VRP model according to the distribution of different unloading positions. It includes the following steps:

(2.1) Define the distribution of five grain unloading locations

The present invention is researched based on the intermittent grain unloading method, and the grain transport vehicle cannot run in the farmland and needs to be parked on the side of the road. According to the location and number of unloading grains, it can be divided into the following five situations:

"a": There is only one unloading position S ₁ , at the head (or tail) of the operation line;

"b": There is only one unloading position S ₁ , on the side of the road parallel to the operation row;

"c": There are two grain unloading positions S ₁ and S ₂ , which are at the head and tail of the operation line respectively;

"d": There are two grain unloading positions S ₁ and S ₂ , which are respectively on the two sides of the road parallel to the operation row;

"e": There are two grain unloading positions S ₁ , S ₂ , one is at the head (or tail) of the operation row, and the other is on the road parallel to the operation row.

(2.2) Establish a VRP model

到作业行尾端的距离为

卸粮点S₂到作业行首端的距离为

到作业行尾端的距离为

作业行之间的转弯距离为

根据五种卸粮位置分布情况，分别建立作业行与卸粮位置间的距离模型如下：Farmland full coverage path planning refers to planning the operation path of the harvester so that it traverses the entire farmland, and each operation row only passes through once, and the loading capacity of the harvester cannot exceed its own capacity. When the load is full or close to full load, it needs to go to the unloading position to unload the grain. Let the set of path points be E, which includes the start point i _up , the end point i _down , and the unloading positions S ₁ and S ₂ of each job row. Each job row has two attributes _: length li and harvestable grain volume _Vi . The harvester starts from the unloading position, traverses each operation line in a certain order, and finally returns to the unloading position. Let the capacity of the harvester be V _granary , the serial numbers of the operation lines are i, j (i,j∈[1,n _line ]), and the distance from the unloading point S1 to the beginning _of the operation line is

The distance to the end of the job line is

The distance from the unloading point S2 to the _beginning of the operation row is

The distance to the end of the job line is

The turn distance between job rows is

According to the distribution of five grain unloading positions, the distance models between the operation line and the grain unloading position are respectively established as follows:

(2.2.1) The harvester is unloaded at the beginning of the operation row

where li is the length of the _i -th job line.

(2.2.2) The harvester is unloaded on the side parallel to the working row

Among them, w is the width of the work row, θ _up is the angle between the upper boundary and the work row, and θ _down is the angle between the lower boundary and the work row.

(2.2.3) The harvester can be unloaded at the beginning and end of the operation row

(2.2.4) The harvester can be unloaded on both sides parallel to the operation row

Among them, n _line is the number of job lines.

(2.2.5) The harvester is unloaded at the head and left side of the operation row

为The turn distance between job rows is

for

Among them, S _T is the turning distance of the T-turn, S _Ω is the turning distance of the Ω-turn, and S _U is the turning distance of the U-turn.

Step 3. According to the harvester capacity, total driving distance, full-load driving distance and unloading position distribution constraints, an improved ant colony algorithm is used to design the optimal traversal sequence of job rows. Specifically include the following steps:

(3.1) Initialization of Ant Colony Algorithm

Set the size of the ant colony as M, the importance factor of the legacy pheromone as α, the importance factor of the heuristic function as β, the pheromone volatility factor as ρ, the total amount of pheromone released by ants in one iteration as Q, and the maximum number of iterations as I _max .

距离越小，函数值越大。In the first iteration, the initial pheromone content on each path is the same. The ant colony starts from the unloading point and selects a reasonable path according to the pheromone content and heuristic function of each route. The heuristic function of each path pq is

The smaller the distance, the larger the function value.

(3.2) Path selection

In the calculation of the nth iteration, the probability of the mth ant transferring from the current point p to the next point q is:

where τ _pq (n) is the pheromone of the path pq in the nth iteration.

并记录总路径长度

通过对比，获得最小的总路径长度

以及对应的路径

Select the path with the largest transition probability as the target path, and judge whether it exceeds the capacity constraint. If it exceeds, the point will not be added, and the next target point will continue to be found; if not, it will be added to the current path. Repeat this step until all job lines are traversed, complete one iteration, and record the path node of each ant

and record the total path length

By comparison, the minimum total path length is obtained

and the corresponding path

(3.3) Adjust the route length based on the weight weight factor

The accumulation of pheromone is directly related to the length of the route. By changing the length of the route, the content of pheromone can be changed, which affects the optimization process of the ant colony algorithm. In the current path planning algorithm, the total distance traveled is often used as the cost function, and through the optimization algorithm, the feasible solution that makes the function reach the minimum value is obtained. In addition to considering the total travel distance, this paper adds a full load distance constraint. In order to reduce the degree of compaction of the land by the harvester under full load, a weight weighting factor is added to the true distance of the path. When the ants pass through a job line, the volume of the harvested grain is updated, and the value is associated with the length of the path to guide the ants to optimize according to the full load distance constraint. The distance of the path is improved as follows:

Among them, weight _mass is the weight weight factor, V _now is the volume of the currently harvested grain, V _granary is the _granary capacity, and km is the scale factor. When the ants pass through a certain path with a large weight, the recorded distance of the path will be greater than the real distance, thereby reducing the pheromone of the path and inducing the ants to find a shorter path under full load.

(3.4) Update pheromone

Every time an ant passes a path, it will leave a pheromone on the path. At the next iteration, the pheromone content of the path pq is

τ _pq (n+1)=(1-ρ)τ _pq (n)+Δτ _pq

Among them, (1-ρ)·τ _pq (n) is the remaining pheromone after volatilization, and Δτ _pq is the pheromone left by all ants passing through the path in this iteration as

Among them, the pheromone content left by each ant is

This value is determined by the distance of the path traversed, the shorter the distance, the higher the pheromone content.

(3.5) Determine whether to terminate the iteration

If the end condition of the ant colony algorithm is satisfied, the path traversed by the ants in this iteration is the traversal sequence of the job row, and go to step 4;

If the end condition of the ant colony algorithm is not satisfied, set the number of iterations i=i+1, and jump to step (3.2) to continue searching for the next path.

The end condition of the ant colony algorithm is:

The current number of iterations i>I _max , or the optimal solution meets the accuracy requirements:

是第i次迭代后，得到的最短路径长度，ξ是预设的精度阈值。in,

is the shortest path length obtained after the ith iteration, and ξ is the preset accuracy threshold.

Step 4. According to the traversal sequence of the job row and the farmland model, the expressions of each path are solved to generate a farmland full coverage path, which provides a reference for the path tracking of the harvester. details as follows:

Using the optimal traversal order of the job row obtained by the solution in step 3, the expressions of each job row obtained by the solution in step 1 are connected to form a path expression covering the entire farmland.

The above is only one of the preferred embodiments of the present invention, and is not intended to limit the present invention in any other form, and any modification or equivalent change made according to the technical essence of the present invention still belongs to the protection claimed by the present invention. scope.

Claims (5)

1. The harvesting robot operation path planning method based on the improved ant colony algorithm is characterized by comprising the following steps of:

step 1, establishing a farmland model:

setting the four vertexes of the farmland as A, B, C and D, knowing the coordinates of the four points, solving the mathematical expression of the boundary, completing the preliminary modeling of the farmland, determining the operation direction under the conditions of the minimum turning times and the maximum vertical degree of the operation line and the boundary, and respectively taking the sides and the diagonal lines of the quadrangle as the operation direction to solve the required turning times;

when the AD is taken as the operation direction, the cutting width of the harvester is set as l_{cut_width}Then a series of parallel operation behaviors

Wherein k is_ADIs the slope of the AD side, b_ADThe intercept is AD edge intercept, i is the serial number of the operation line;

respectively solving the intersection points of the group of parallel lines and the quadrilateral boundary, if two intersection points exist, adding one to the turn times until no intersection point exists, and keeping the current turn times as the result of the operation direction;

similarly, respectively using AB, BC, CD, AC and BD as operation directions, solving corresponding turning times, wherein the operation direction with the minimum turning times is obtained by comparison, if a plurality of groups of operation directions all meet the requirement of the minimum turning times, comparing the included angle between the operation direction and the boundary, selecting the operation direction with the included angle close to 90 degrees because the turning complexity and the turning distance are smaller under the vertical condition, and finally solving to obtain the mathematical expression of each operation line so as to perfect the farmland model;

step 2, abstracting the farmland full-coverage path planning into a vehicle route problem, and establishing a corresponding VRP model according to different grain unloading position distribution;

step 3, designing an optimal operation line traversal sequence by adopting an improved ant colony algorithm according to the capacity of the harvester, the total travel distance, the full-load travel distance and the grain unloading position distribution constraint condition;

and 4, solving the expression of each path according to the traversal sequence of the operation lines and the farmland model to generate a farmland full-coverage path and provide reference for path tracking of the harvester.

2. The harvesting robot working path planning method based on the improved ant colony algorithm according to claim 1, wherein the step 2 abstracts the farmland full coverage path planning into a vehicle route problem, and establishes a corresponding VRP model according to different grain unloading position distributions, and comprises the following steps:

(2.1) defining five grain unloading position distribution conditions:

based on intermittent type formula mode of unloading is studied, and the fortune grain car can not travel in the farmland, need berth at the roadside, divide into following five kinds of condition according to the position and the figure of unloading:

"a": only one grain unloading position S₁At the head or tail of the operation line;

"b": only one grain unloading position S₁On a side road parallel to the work line;

"c": with two grain unloading positions S₁、S₂Respectively at the head and tail of the operation line;

"d": with two grain unloading positions S₁、S₂On the two side roads parallel to the operation line;

"e": with two grain unloading positions S₁、S₂One at the head or tail of the line and one on a road parallel to the line;

(2.2) establishing a VRP model:

the farmland full-coverage path planning means that the operation path of the harvester is planned to traverse the whole farmland, each operation line only passes through once, the loading capacity of the harvester cannot exceed the capacity of the harvester, when the harvester is fully loaded or nearly fully loaded, the harvester needs to unload grains at the grain unloading position, the set of path points is E, wherein the set of path points comprises a head end point i of each operation line_upEnd point i_downPosition S for unloading grain₁、S₂Each job row having a length of l_iAnd grain volume V that can be harvested_iTwo attributes, starting from the grain unloading position, traversing each operation line in a certain sequence, and finally returning to the grain unloading position, and setting the capacity of the harvester as V_granaryThe sequence number of the operation line is i, j, wherein i, j belongs to [1, n ]_line]Unloading point S₁A distance to the head end of the working line of

To the tail end of the working line at a distance of

Grain unloading point S₂A distance to the head end of the working line of

To the tail end of the working line at a distance of

The turning distance between the working lines is

According to five grain unloading positionsAnd (3) setting distribution conditions, and respectively establishing a distance model between an operation line and a grain unloading position as follows:

(2.2.1) unloading the harvester at the head of the working line:

wherein l_iIs the length of the ith job row;

(2.2.2) the harvester unloads on the side parallel to the working line:

wherein w is the working line width, θ_upIs the angle between the upper boundary and the line of operation, θ_downIs the included angle between the lower boundary and the operation line;

(2.2.3) the harvester can be unloaded at both the head and tail ends of the work line:

(2.2.4) the harvester can unload on both sides parallel to the working line:

wherein n is_lineIs the number of operation lines;

(2.2.5) harvester unloading at work line head end and left side:

the turning distance between the working lines is

Is composed of

Wherein S is_TTurning distance in T-turn mode, S_ΩTurning distance in omega-type turning mode, S_UThe turning distance of the U-shaped turning mode.

3. The method for planning the operation path of the harvesting robot based on the improved ant colony algorithm as claimed in claim 1, wherein the step 3 is to design an optimal operation traversal order by using the improved ant colony algorithm according to the constraints of the harvester capacity, the total travel distance, the full-load travel distance and the grain unloading position distribution, and specifically comprises the following steps:

(3.1) ant colony algorithm initialization:

setting the ant colony scale as M, the important degree factor of the left pheromone as alpha, the important degree factor of the heuristic function as beta, the pheromone volatilization factor as rho, the total pheromone released by the ant in one iteration as Q, and the maximum iteration number as I_max；

In the first iteration, the initial pheromone content on each path is the same, the ant colony starts from a grain unloading point, a reasonable path is selected according to the pheromone content and the heuristic function of each path, and the heuristic function of each section of path pq is

The smaller the distance, the larger the function value;

(3.2) path selection:

calculating the probability that the mth ant transfers from the current point p to the next point q in the nth iteration as

Wherein, tau_pq(n) is the pheromone of path pq in the nth iteration;

selecting the path with the maximum transition probability as a target path, judging whether the capacity constraint is exceeded or not, if so, not adding the point, and continuously searching the next target point; if not, it is added to the current path. Repeating the steps until all the operation lines are traversed, completing one iteration, and recording the path node of each ant

And recording the total path length

By contrast, the minimum total path length is obtained

And corresponding path

(3.3) adjusting the route length based on the weight factor:

the accumulation of pheromones has a direct relation with the length of a route, the searching process of an ant colony algorithm is influenced by changing the content of the pheromones by changing the length of the route, in the current path planning algorithm, the total distance of driving is taken as a cost function, a feasible solution for enabling the function to reach the minimum value is obtained by solving through an optimization algorithm, except for considering the total driving distance, a full load distance constraint is added, when an ant passes through an operation line, the grain volume harvested by the ant is updated, the value is associated with the length of the route, the ant is guided to search for the optimization according to the full load distance constraint, and the distance of the route is improved as follows:

wherein, light_massIs a weight factorSeed, V_nowFor the grain volume currently harvested, V_granaryIs the volume of the granary, k_mWhen ants pass through a certain path with larger weight, the recorded distance of the path is larger than the real distance, so that pheromone of the path is reduced, and the ants are induced to search for a shorter path under the condition of full load;

(3.4) update pheromone:

every time an ant passes a path, it leaves a pheromone on that path. On the next iteration, the pheromone content of path pq is

τ_pq(n+1)＝(1-ρ)τ_pq(n)+Δτ_pq

Wherein (1-rho). tau_pq(n) pheromones remaining after volatilization, Δ τ_pqThe pheromones left for all ants passing through this path in this iteration are:

wherein, the content of the pheromone left by each ant is as follows:

the value is determined by the distance of the path traversed, the shorter the distance, the higher the pheromone content;

(3.5) judging whether to terminate the iteration:

if the ant colony algorithm end condition is met, the path traveled by the iterative ants is the operation line traversal sequence, and the step 4 is carried out;

and if the ant colony algorithm ending condition is not met, enabling the iteration number i to be i +1, and jumping to the step (3.2) to continue to search the next path.

4. A harvesting robot working path planning method based on improved ant colony algorithm according to claim 3, characterized in that the ant colony algorithm end condition is:

the current iteration number I is more than I_maxOr the optimal solution meets the accuracy requirement:

wherein,

the shortest path length obtained after the ith iteration is shown, and xi is a preset precision threshold.

5. The method for planning the operation path of the harvesting robot based on the improved ant colony algorithm as claimed in claim 1, wherein the step 4 is to solve the expression of each path according to the traversal sequence of the operation lines and the farmland model to generate a farmland full-coverage path, and provide a reference for path tracking of the harvester, and specifically comprises the following steps:

and (3) designing an optimal operation line traversal sequence by adopting an improved ant colony algorithm according to the capacity of the harvester, the total travel distance, the full-load travel distance and the grain unloading position distribution constraint conditions in the step 3, and linking the operation line expressions obtained by the step 1 by utilizing the traversal sequence to form a path expression covering the whole farmland.

Images