Photovoltaic panel cleaning trolley laser radar SLAM autonomous navigation system

Abstract

The invention discloses a photovoltaic panel cleaning trolley laser radar SLAM autonomous navigation system, and belongs to the technical field of autonomous navigation, and the system comprises a map construction module which completes SLAM map construction through a Gmapping algorithm; the positioning module adopts self-adaptive Monte Carlo positioning based on a particle filtering algorithm to realize self positioning of the cleaning trolley; the path planning module comprises global path planning and local path planning; according to the global path planning, global path planning is carried out based on a global cost map according to a specified target point; the local path planning is to adjust a local planning path based on a local cost map and obstacle information in the moving process of the cleaning trolley so as to avoid obstacles. According to the cleaning trolley laser radar SLAM autonomous navigation system for the photovoltaic panel, the cleaning trolley can sense the environment in real time, position the position of the cleaning trolley and construct an accurate map so as to plan a path and avoid obstacles.

Description

Technical Field

[0001] The present invention relates to the field of autonomous navigation technology, and in particular to a laser radar SLAM autonomous navigation system for a photovoltaic panel cleaning vehicle. Background Art

[0002] During the use of photovoltaic panels, a layer of sand and dust will be covered on the photovoltaic panels. Under normal illumination conditions, uncleaned photovoltaic panels will reduce the power generation efficiency by 15% and aggravate the aging of the panels. In severe cases, it may cause a fire. Therefore, the photovoltaic panels need to be cleaned. The development of intelligent cleaning devices has become a hot topic in the current scientific and technological field, especially in the photovoltaic cleaning industry. Facing the diversified photovoltaic arrays, photovoltaic cleaning faces important challenges.

[0003] The navigation technology of mobile cars is the future development prospect of mobile cars, and it is also a basic function that intelligent mobile cars should have. The mobile car collects information about the surrounding environment and its own position and posture based on the sensor, and plans a path in the map constructed by the mobile car, so that the car can avoid obstacles and reach the corresponding target position according to the instructions. The main function of the navigation system is to plan a safe, efficient, fast and convenient path for the target user. By loading the automatic navigation system on the intelligent cleaning device, it can plan the corresponding autonomous navigation path according to the array layout of the current photovoltaic field, and quickly, accurately and safely reach the cleaning task execution position, reducing the labor employment cost, and having the ability of static and dynamic obstacle avoidance to avoid collision damage to the cleaning equipment. Therefore, the autonomous navigation system is of great significance for photovoltaic intelligent cleaning. Summary of the invention

[0004] The purpose of the present invention is to provide a photovoltaic panel cleaning trolley laser radar SLAM autonomous navigation system, which can plan the path for the cleaning trolley and has the ability to plan the path and automatically avoid obstacles.

[0005] To achieve the above object, the present invention provides a photovoltaic panel cleaning vehicle laser radar SLAM autonomous navigation system, comprising:

[0006] Map construction module, which completes SLAM map construction through Gmapping algorithm;

[0007] The positioning module uses adaptive Monte Carlo positioning based on particle filter algorithm to realize the positioning of the cleaning trolley itself;

[0008] The path planning module includes global path planning and local path planning; global path planning is to perform global path planning based on the global cost map according to the specified target point; local path planning is to adjust the local planned path based on the local cost map and the obstacle information during the movement of the cleaning car to avoid obstacles.

[0009] Preferably, the specific operation process of the Gmapping algorithm is as follows:

[0010] S1. Initialize particles. In the initial stage, initialize N particles so that the weight of each particle is initialized to the average weight, as follows

[0011]

[0012] S2, sampling, using the odometer model as the proposed distribution of the cleaning car for sampling;

[0013] S3, calculate the particle weight, when generating a new generation of sample particles, the sensor observation data Z s Integrating it into the proposed distribution can focus the sampling process on areas with high observation likelihood, thereby improving sampling efficiency and accuracy, as follows

[0014]

[0015] S4, resampling, is the process of re-collecting particles based on particle weights, and whether to resample is determined by judging the size of the sum of the squares of the particle weights.

[0016]

[0017] In actual calculation, N eff Can be approximately equal to

[0018]

[0019] Among them, N eff represents the particle weight dispersion measure; N represents the number of particles; Represents the particle weight difference; it can be seen that N eff and It is an inverse relationship; when the N value is small, the difference in particle weights is large, and the gap between the sampled particle distribution and the actual distribution is also large. In this case, the collected particles may be relatively inaccurate, and most particles may deviate from the actual distribution value, indicating that there may be problems with the sampling of these particles. At this time, resampling is required to replace the previously sampled particles. This strategy can improve the positioning accuracy of particles and make the generated map more accurate;

[0020] S5, update the map according to the particle set The optimal particle is selected according to the particle weight in , the trajectory of the cleaning vehicle is determined, the map is updated in real time, and the constructed map is stored to complete the map construction.

[0021] Preferably, in the process of calculating the particle weight in S3, the long-term change mean of the particle weight ω is also tracked slow and the short-term mean ω fast , according to the preset long-term change rate α slow and the short-term rate of change α fast , for slow and ω fast Maintain and update according to the following formula:

[0022]

[0023] Preferably, in the particle resampling stage of S4, random sampling is performed on any particle according to the above formula to increase random particles. When the short-term change mean ω fast Equal to or greater than the long-term mean ω slow When , the cleaning car can be accurately positioned, and the particle group tends to converge. However, if the average measurement probability of each particle in the particle set decreases, the cleaning car may encounter positioning problems. At this time, random particles need to be added to solve this problem, as follows

[0024]

[0025] Preferably, in order to optimize the calculation speed of the algorithm, Monte Carlo positioning adopts KLD (Kullback-Leibler Divergence) sampling, and the size of the particle set is determined by calculating the distance between the approximate value of the particle filter and the actual probability distribution. The expression is as follows:

[0026]

[0027] Among them, M T represents the size of the particle set; k represents the degree to which the particle set covers the global map; α represents the error threshold between the approximate probability distribution and the true distribution; β represents a point on the normal distribution.

[0028] Preferably, the global path planning adopts A * algorithm, and add self-adjusting weighting coefficients weights, the mathematical formula is:

[0029] f(n)=g(n)+w(n)·h(n)

[0030] Among them, n represents the location identifier of the current scene search; f(n) represents the minimum cost from the initial location of the cleaning car to the target location; g(n) represents the actual minimum cost of the cleaning car from the initial location to the current location n; h(n) is a heuristic function, which represents the estimated cost of the cleaning car from the current location n to the target location; w(n) represents the self-adjusting weight coefficient weights, and w(n)≥1.

[0031] Preferably, in the global path planning, the path smoothing design is used to realize the smoothing of the cleaning vehicle's path, specifically:

[0032] Assume there is a series of control sites P0, P1, ..., P n , the planned path passes through the first point P0 and the last point P n , its mathematical expression formula is as follows:

[0033]

[0034] in, represents the binomial coefficient; t represents the parameter, and t≥0; P i represents the control site; the value of n is proportional to the smoothness of the path.

[0035] Preferably, the local path planning adopts a dynamic window method, specifically: the linear velocity v and angular velocity w of the cleaning trolley are used to simulate the travel path; first, multiple groups of (v, w) data are sampled in the velocity space of the cleaning trolley, and then the travel trajectory of the cleaning trolley in a certain set of linear velocity and angular velocity per unit time is derived according to the dynamic model of the trolley; secondly, a scoring function is used to evaluate the derived trajectory, and the evaluation criteria include the speed of the trolley, the distance to the obstacle, and the angle between the trolley's movement direction and the target site; finally, the optimal travel trajectory is selected as the final travel path of the cleaning trolley in unit time.

[0036] Therefore, the present invention adopts the above-mentioned photovoltaic panel cleaning car laser radar SLAM autonomous navigation system, based on the improved PBPF Gmapping map construction algorithm and AMCL positioning algorithm, and A * The algorithm made heuristic function improvements and path smoothing processing, and verified its effectiveness through corresponding simulation comparison experiments, and adopted DWA local path planning. Finally, the cleaning car combined with the map environment and path planning algorithm was able to complete static obstacle avoidance and dynamic obstacle avoidance experiments and realize autonomous navigation.

[0037] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a schematic diagram of the structure of a laser radar SLAM autonomous navigation system for a photovoltaic panel cleaning vehicle of the present invention;

[0039] Figure 2 This is a schematic diagram of the SLAM working state of the present invention;

[0040] Figure 3 A in the present invention * Algorithm flow chart;

[0041] Figure 4 30×30 grid map heuristic function optimization before and after comparison of the embodiment of the present invention; wherein (a) is the original A * Schematic diagram of the path planned by the algorithm; (b) Schematic diagram of the path optimized by the heuristic function;

[0042] Figure 5 60×60 grid map heuristic function optimization before and after comparison of the embodiment of the present invention; wherein (a) is the original A * Schematic diagram of the path planned by the algorithm; (b) Schematic diagram of the path optimized by the heuristic function;

[0043] Figure 6 Comparison of the path smoothing optimization before and after the 30×30 grid map in the embodiment of the present invention; wherein (a) is a schematic diagram of the path optimized by the heuristic function; (b) is a schematic diagram of the path smoothing optimization;

[0044] Figure 7 The figure is a comparison of the path smoothing optimization before and after the 60×60 grid map in the embodiment of the present invention; wherein (a) is a schematic diagram of the path optimized by the heuristic function; (b) is a schematic diagram of the path smoothing optimization;

[0045] Figure 8 This is a schematic diagram of the motion model of the cleaning trolley in an embodiment of the present invention;

[0046] Fig. 9 Schematic diagram of the local path planning simulation process of an embodiment of the present invention; wherein, (a) is a schematic diagram of process one; (b) is a schematic diagram of process two; (c) is a schematic diagram of process three; and (d) is a schematic diagram of process four. DETAILED DESCRIPTION

[0047] Example

[0048] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0049] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] like Figure 1 As shown, a photovoltaic panel cleaning vehicle laser radar SLAM autonomous navigation system includes:

[0051] The map construction module completes the SLAM map construction through the Gmapping algorithm; Figure 2 As shown in the figure, the location information of the photovoltaic intelligent cleaning car and landmarks (markers, i.e., objects or areas with identifiable features in the mapping environment) needs to be obtained during operation. Since the position in the real world cannot be directly measured, the observation result is to infer the actual location information relative to the position relationship between the photovoltaic intelligent cleaning car and the landmarks. In the figure, X represents the position trajectory of the photovoltaic intelligent cleaning car; U represents the historical amount of the control input; m represents the set of all landmarks; Z represents the set of all observed landmark information; s represents a certain point in time; X represents the location ... a certain point in time; s The vector representing the posture and position of the photovoltaic intelligent cleaning vehicle; U s represents the control variable applied at time s-1, and this control variable can make the robot reach the position X at time s-1; Z k,i It indicates that the location information of the i-th landmark is observed from the position of the photovoltaic intelligent cleaning car at time s, and the observation result is expressed as Z k Indicates; m i The vector representing the position of the i-th landmark. It is assumed here that the actual position of the landmark remains unchanged.

[0052] The positioning module uses adaptive Monte Carlo positioning based on particle filter algorithm to realize the positioning of the cleaning trolley itself;

[0053] The path planning module includes global path planning (globalplanner) and local path planning (localplanner); global path planning is to perform global path planning based on the global costmap according to the specified target point; local path planning is to adjust the local planned path based on the local costmap and the obstacle information during the movement of the cleaning car to avoid obstacles.

[0054] like Figure 3 As shown, the global path planning adopts A * Algorithm, A * The algorithm combines the concepts of the Djkstra algorithm and the BFS algorithm, using heuristic search to improve the efficiency of the algorithm while retaining the ability of the Djkstra algorithm to find the global shortest path. *The algorithm can effectively limit its blind iterative calculation, thereby improving the search efficiency, and design self-adjusting weighting coefficients weights. The mathematical formula is:

[0055] f(n)=g(n)+w(n)·h(n)

[0056] Where n is the location identifier of the current scene search; f(n) is the minimum cost from the initial location of the cleaning car to the target location; g(n) is the actual minimum cost of the cleaning car from the initial location to the current location n; h(n) is the heuristic function, which represents the estimated cost of the cleaning car from the current location n to the target location; w(n) is the self-adjusting weight coefficient weights, and w(n) ≥ 1, so that A * The algorithm can automatically adjust the weight coefficient of the heuristic function h(n) during the path planning process.

[0057] In order to more intuitively compare the visualization results and operation data, the simulation experiment of this embodiment is based on the MATLAB experimental platform. On two grid maps of 30×30 and 60×60 sizes, the control variable method is adopted, and the upper left corner circle mark is unified as the starting point, and the lower right corner square mark is unified as the end point. The starting point and the end point are in the same position under the same grid map specification. The number of obstacles generated accounts for 20% of the total number of squares, and the layout of obstacles is not randomly changed. The heuristic function before and after optimization is simulated for path planning, and the experimental results of the two cases are compared. Figure 4-5 As shown in Table 1;

[0058] Table 1 Comparison of heuristic function optimization before and after under different grid map specifications

[0059]

[0060] As shown in Table 1, under the 30×30 grid map, the algorithm optimized by the heuristic function increases the path length by 6.9%, but shortens the search time by 73.61%; under the 60×60 grid map, the optimized algorithm increases the path length by 14.29%, but shortens the search time by 79.3%. These experimental results show that the A optimized by the heuristic function is * The algorithm is consistent with the changing characteristics of path planning when h(n) is taken. Although it is not the shortest path, the path length increases, but the increase is small, and the search time is greatly reduced, which verifies the A * The effectiveness of the algorithm for heuristic function optimization.

[0061] In the global path planning, the path smoothing design is used to realize the smoothness of the cleaning vehicle's travel path, specifically:

[0062] The Bezier curve is a parametric curve that changes the shape of the curve by adjusting the position of the control points to achieve a smooth path. Assume there is a series of control points P0, P1, ..., P n , the planned path passes through the first point P0 and the last point P n , its mathematical expression formula is as follows:

[0063]

[0064] in, represents the binomial coefficient; t represents the parameter, and t≥0; P i represents the control site; the value of n is proportional to the smoothness of the path. The design of the comparative experiment is consistent with the above experimental design, and the experimental results are shown in Figure 6-7 As shown in Table 2:

[0065] Table 2 Comparison of corner smoothing optimization before and after under different specifications of raster maps

[0066]

[0067] As shown in Table 2, under the 30×30 grid map, the algorithm after corner smoothing optimization reduces the path length by 3.23%, the search time by 44.54%, and the number of corners by 56.52%; under the 60×60 grid map, the optimized algorithm reduces the path length by 10.71%, the search time by 19.1%, and the number of corners by 55%. These experimental results show that the A optimized corner smoothing algorithm can effectively reduce the search time of the A-axis. * The path length of the algorithm is reduced, and the search time and the number of corners are greatly reduced, which verifies the * The effectiveness of the algorithm for corner smoothing optimization.

[0068] The most commonly used local path planning algorithm is the Dynamic Window Approach (DWA), in addition to the Artificial Potential Field (APF), Model Predictive Control (MPC), Time Elastic Band (TEB), etc. Compared with other local path planning algorithms, the dynamic window method takes into account the dynamic characteristics and inertia of the photovoltaic cleaning vehicle, generates a feasible arc trajectory by selecting a suitable speed, and can add various constraints as needed. Because its algorithm principle is simple and reliable, and its application range is wide, the dynamic window method is selected for local path planning in this study.

[0069] In this embodiment, the local path planning adopts the dynamic window method, specifically: the linear velocity v and angular velocity w of the cleaning trolley are used to simulate the travel path; first, multiple groups of (v, w) data are sampled in the speed space of the cleaning trolley, and then the travel trajectory of the cleaning trolley in a unit time under a certain set of linear velocity and angular velocity is derived according to the dynamic model of the trolley; secondly, a scoring function is used to evaluate the derived trajectory, and the evaluation criteria include the speed of the trolley, the distance to the obstacle, and the angle between the trolley's movement direction and the target site; finally, the optimal travel trajectory is selected as the final travel path of the cleaning trolley in unit time.

[0070] The solution process of the DWA algorithm is as follows:

[0071] (1) Kinematic model of cleaning trolley: The cleaning trolley experimental device of this embodiment adopts a crawler-type mobile chassis, and its driving mode is a two-wheel differential drive, that is, it can only move forward and backward and rotate left and right. Therefore, the following is established: Figure 8 The two-wheel differential cleaning trolley model shown.

[0072] Assume that in a very short sampling time interval Δ s Within Δ, the trajectory of the photovoltaic cleaning vehicle can be considered as a straight line. s The displacement increment of the photovoltaic cleaning vehicle in the global coordinate system at this moment is as follows:

[0073]

[0074] Photovoltaic cleaning vehicle at (s+Δ s The relationship between the position and rotation angle at time ) and time s is as follows:

[0075]

[0076] (2) Speed ​​space sampling: Before sampling, the operating speed range of the cleaning trolley must be confirmed to ensure that it can operate smoothly and safely at this speed. The main factors affecting the trolley speed include the hardware configuration of the photovoltaic cleaning trolley and the operating environment restrictions. The maximum values ​​of the linear speed and angular velocity need to be determined based on its hardware performance:

[0077] V l ={(v,w)|v∈[v min ,v max ],w∈[v min ,v max ]}

[0078] When the photovoltaic cleaning car changes speed, its acceleration per unit time will be affected by the motor. The motor's constraint on the car's acceleration causes the car's speed change per unit time to have a certain range. Therefore, based on the limitation of the car's acceleration, the speed sampling space at this time is:

[0079] V a

[0080] ={(v,w)|v∈[v c -a vmax ·Δ s ,v c +a vmin ·Δ s ],w∈[w c -a wmax ·Δ s ,w c +a wmin ·Δ s ]}

[0081] Among them, v c and w c They represent the linear velocity and angular velocity of the photovoltaic cleaning vehicle at the current moment respectively; a vmax and a vmin They represent the maximum and minimum values ​​of the car's linear acceleration respectively; a wmax and a wmin They represent the maximum and minimum angular acceleration of the car respectively.

[0082] Based on the safety considerations of the photovoltaic cleaning car, in order to prevent the car from colliding with obstacles during the cleaning operation, it is necessary to ensure that the car can stop or turn within a certain distance before contacting the obstacle under the maximum deceleration. Based on the physical formula of displacement and acceleration, it can be deduced that the speed of the car needs to meet the following constraints:

[0083]

[0084] In the formula, d(v,w) represents the distance between the photovoltaic cleaning car and the nearest obstacle in the map. Only when the speed of the car meets the above conditions can it be guaranteed that the car will not collide with the obstacle.

[0085] The sampling speed of the photovoltaic cleaning car must meet the above constraints, that is, the speed set of the car:

[0086] V=V l ∩V a ∩V z

[0087] After calculating the travel paths of the photovoltaic cleaning vehicle at different operating speeds, it is necessary to select the best path from these travel paths. The trajectory evaluation function will evaluate all predicted paths in the speed sampling space, and then select the speed corresponding to the path with the best score as the next speed state. The mathematical formula of the trajectory evaluation function is as follows:

[0088] F(v,w)=σ(α·heading(v,w)+β·dist(v,w)+γ·velocity(v,w))

[0089] In the formula, α, β, and γ represent the weight coefficients of the corresponding functions respectively. Heading(v,w) means that the photovoltaic cleaning car reaches the end of the path at the currently set sampling speed (v,w). At this moment, the angle Δθ between the car's direction and the target is inversely proportional to the score. Dist(v,w) represents the distance Δl between the car and the nearest obstacle on the current path. The distance is proportional to the score. Assuming that there are no obstacles on this path, the value is set to a constant. Velocity(v,w) represents the linear speed v of the cleaning car. c , which is proportional to the score.

[0090] The simulation experiment of this embodiment is also based on the MATLAB platform. The simulation range is set to 60m×60m. The starting point and the end point are set to the diagonal state to increase the path length. In order to more intuitively show the local path planning effect, the obstacle position is preferably arranged near the diagonal straight line. The simulation parameters are adjusted according to the speed setting of the photovoltaic intelligent cleaning car: the maximum linear speed is 5m / s, the linear acceleration is 0.2m / s2, the maximum rotation speed is 20rad / s, and the rotation acceleration is 50rad / s2. The parameters of the evaluation function are set to: α=0.05, β=0.1, γ=0.1, and the simulation process is as follows: Fig. 9 shown.

[0091] In the figure, the rectangular box represents the obstacle, the long line represents the trajectory of the simulated car, and the short ray beam shows the local path planning results of different speed combinations. It can be seen from the figure that the DWA algorithm can effectively complete the local path planning in the map, which proves the practicality of the DWA algorithm in obstacle avoidance.

[0092] Therefore, the present invention adopts the above-mentioned photovoltaic panel cleaning vehicle laser radar SLAM autonomous navigation system, so that the cleaning vehicle can perceive the environment in real time, locate its own position, and build an accurate map to plan the path and avoid obstacles.

[0093] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A photovoltaic panel cleaning vehicle laser radar SLAM autonomous navigation system, characterized in that: include: Map construction module, which completes SLAM map construction through Gmapping algorithm; The positioning module uses adaptive Monte Carlo positioning based on particle filter algorithm to realize the positioning of the cleaning trolley itself; The path planning module includes global path planning and local path planning; global path planning is to perform global path planning based on the global cost map according to the specified target point; local path planning is to adjust the local planned path based on the local cost map and the obstacle information during the movement of the cleaning car to avoid obstacles.

2. According to the photovoltaic panel cleaning vehicle laser radar SLAM autonomous navigation system of claim 1, it is characterized in that: The specific operation process of the Gmapping algorithm is as follows: S1. Initialize particles. In the initial stage, initialize N particles so that the weight of each particle is initialized to the average weight, as follows S2, sampling, using the odometer model as the proposed distribution of the cleaning car for sampling; S3, calculate the particle weight, when generating a new generation of sample particles, the sensor observation data Z s Integrate into the proposed distribution as follows S4, resampling, is the process of re-collecting particles based on particle weights and determining whether to resample. In actual calculation, N eff for Among them, N eff represents the particle weight dispersion measure; N represents the number of particles; Indicates the particle weight difference; when N is small, the particle weight difference is large, and the sampled particle distribution is also large from the actual distribution. At this time, resampling is performed to replace the sampled particles in S2; S5, update the map according to the particle set The optimal particle is selected according to the particle weight in , the trajectory of the cleaning vehicle is determined, the map is updated in real time, and the constructed map is stored to complete the map construction.

3. The photovoltaic panel cleaning vehicle laser radar SLAM autonomous navigation system according to claim 2 is characterized in that: In the process of calculating particle weights in S3, the long-term change mean of particle weights ω is also tracked slow and the short-term mean ω fast , according to the preset long-term change rate α slow and the short-term rate of change α fast , for slow and ω fast Maintain and update according to the following formula:

4. The photovoltaic panel cleaning vehicle laser radar SLAM autonomous navigation system according to claim 2 is characterized in that: In the particle resampling stage of S4, random sampling is performed on any particle according to the above formula to increase random particles, as follows 5. The photovoltaic panel cleaning vehicle laser radar SLAM autonomous navigation system according to claim 1 is characterized in that: Monte Carlo positioning uses KLD sampling, and the expression is as follows: Among them, M T represents the size of the particle set; k represents the degree to which the particle set covers the global map; α represents the error threshold between the approximate probability distribution and the true distribution; β represents a point on the normal distribution.

6. The photovoltaic panel cleaning vehicle laser radar SLAM autonomous navigation system according to claim 1 is characterized in that: Global path planning adopts A * algorithm, and add self-adjusting weighting coefficients weights, the mathematical formula is: f(n)=g(n)+w(n)·h(n) Among them, n represents the location identifier of the current scene search; f(n) represents the minimum cost from the initial location of the cleaning car to the target location; g(n) represents the actual minimum cost of the cleaning car from the initial location to the current location n; h(n) is a heuristic function, which represents the estimated cost of the cleaning car from the current location n to the target location; w(n) represents the self-adjusting weight coefficient weights, and w(n)≥1.

7. The photovoltaic panel cleaning vehicle laser radar SLAM autonomous navigation system according to claim 1 is characterized in that: In the global path planning, the path smoothing design is used to realize the smoothness of the cleaning vehicle's travel path, specifically: Assume there is a series of control sites P0, P1, ..., P n , the planned path passes through the first point P0 and the last point P n , its mathematical expression formula is as follows: in, represents the binomial coefficient; t represents the parameter, and t≥0; P i represents the control site; the value of n is proportional to the smoothness of the path.

8. The photovoltaic panel cleaning vehicle laser radar SLAM autonomous navigation system according to claim 1, characterized in that: The local path planning adopts the dynamic window method, which is as follows: the linear velocity v and angular velocity w of the cleaning trolley are used to simulate the travel path; first, multiple groups of (v, w) data are sampled in the velocity space of the cleaning trolley, and then the travel trajectory of the cleaning trolley in unit time under a certain set of linear velocity and angular velocity is derived according to the dynamic model of the trolley; secondly, a scoring function is used to evaluate the derived trajectory, and the evaluation criteria include the speed of the trolley, the distance to the obstacle, and the angle between the trolley's movement direction and the target location; finally, the optimal travel trajectory is selected as the final travel path of the cleaning trolley in unit time.

Images