A lunar south pole rover autonomous path planning method for navigation benefit

Abstract

The application discloses a lunar south pole rover autonomous path planning method facing navigation advantages, comprising the following steps: extracting key nodes in a global path node planned by an A* algorithm, and then constructing a global auxiliary line by using the key nodes; fusing the global auxiliary line with a DWA algorithm by optimizing an evaluation function of the traditional DWA algorithm, wherein the optimized evaluation function comprises a distance and an included angle between the rover and the global auxiliary line; and further optimizing the obtained fusion path planning algorithm. In the application, the global auxiliary line is firstly constructed by the optimized A* algorithm, and then fused with the DWA algorithm, and then the evaluation function of the fusion path planning algorithm is optimized with the aim of improving the navigation accuracy of the rover, and finally the feasibility of the proposed path planning method and the adaptability to complex environments are verified through numerical simulation and semi-physical experiments, and the accuracy of rover positioning and mapping can be improved.

Description

Technical Field

[0001] This invention relates to the field of planetary rover path planning technology, specifically to an autonomous path planning method for a lunar south pole rover oriented towards navigation availability. Background Technology

[0002] As the closest celestial body to Earth, the Moon possesses unique scientific value and a unique geographical environment. To date, humanity has conducted over 120 lunar exploration missions. In recent years, to further develop and utilize the Moon, countries have focused their exploration efforts on the lunar polar regions, particularly the lunar south pole, which is rich in water ice. Compared to the near side of the Moon, the lunar south pole has a more complex terrain, and the working environment for rovers is much harsher. To ensure that rovers can better carry out their various exploration tasks, it is necessary to plan a safe and reasonable exploration path for them.

[0003] Since its inception in the 1970s, path planning algorithms have been a hot research topic in the field of robotics. Based on the extent of their environmental information cognition, mainstream path planning algorithms can be divided into global path planning algorithms and local path planning algorithms. Common global path planning algorithms include the A* algorithm, genetic algorithm, and fast random search tree algorithm; algorithms applied to local path planning include the artificial potential field method, the D* algorithm, and the DWA algorithm.

[0004] Existing global path planning algorithms can obtain globally optimal paths but lack the ability to avoid unknown obstacles. Local path planning algorithms, while possessing good local obstacle avoidance capabilities, are prone to getting stuck in local optima. Relying solely on a single path planning algorithm is insufficient to meet the demands of the complex lunar environment. Therefore, scientists have integrated global and local path planning algorithms and proposed a new path planning algorithm tailored to the actual lunar environment. Relevant literature has proposed: a rolling window local obstacle avoidance algorithm that incorporates global auxiliary lines constructed based on the A* algorithm, achieving the fusion of global and local maps and thus improving the success rate of lunar rover path planning; and a path planning algorithm more adaptable to the constraints of the complex lunar environment. This algorithm optimizes the evaluation function of the A* algorithm by weighted fusion of environmental factors such as lunar terrain, illumination, and communication, while simultaneously using an improved D* algorithm for local path planning, ensuring the rover's safe operation in environments with unknown obstacles.

[0005] Most existing literature focuses on path planning algorithms, emphasizing safety, distance, and success rate. Few studies use the rover's navigation accuracy as an optimization metric for path planning algorithms. Furthermore, due to the sparse and uniform terrain features, poor texture, and unstructured topography of the lunar south pole, sensors such as lidar or cameras struggle to acquire sufficient visual features. This affects the accuracy of Simultaneous Localization and Mapping (SLAM) algorithms, which rely on effective features, and further impacts the accuracy and success rate of the rover's path planning.

[0006] Therefore, it is urgent to study an autonomous path planning method for lunar south pole rovers oriented towards navigation availability. This is of great significance in addressing the problem that sensors such as lidar or cameras cannot obtain sufficient visual features on the lunar south pole, where the terrain features are sparse and monotonous, the texture is poor, and the terrain is unstructured. Summary of the Invention

[0007] The purpose of this invention is to provide an autonomous path planning method for a lunar south pole rover oriented towards navigation availability, in order to solve the problem mentioned in the background art that, due to the sparse and monotonous terrain features, poor texture, and unstructured terrain of the lunar south pole, sensors such as lidar or cameras have difficulty acquiring sufficient visual features, thereby affecting the accuracy of localization and mapping algorithms that rely on effective features.

[0008] To achieve the above objectives, the present invention provides the following technical solution: an autonomous path planning method for a lunar Antarctic rover oriented towards navigation availability, specifically including the following steps;

[0009] Step 1: Extract key nodes from the global path nodes planned by the A* algorithm, and then use the key nodes to construct a global auxiliary line as a guide for the DWA algorithm.

[0010] Step 2: By optimizing the evaluation function of the traditional DWA algorithm, the global auxiliary line is integrated with the DWA algorithm. This includes optimizing the evaluation function to include the distance and angle between the patrol and the global auxiliary line.

[0011] Step 3: Using the fusion path planning algorithm obtained in Step 2, draw a fan-shaped region with a radius of R, where the angle between the simulated trajectory endpoint of the DWA algorithm at the current moment and the heading angle of the rovers is α. Then, use the ratio of the area ratio of obstacles within the fan-shaped region to the distance between the nearest obstacle to the trajectory endpoint as an evaluation index for further optimization.

[0012] Preferably, extracting key nodes in step 1 specifically includes:

[0013] A1. Set both the starting node and the target node as key nodes. Starting from the starting node, traverse all path nodes until the target node is reached. Find the key node closest to the current node, and then check whether there are obstacles in the rectangular area enclosed by the two nodes.

[0014] If it does not exist, discard the current node;

[0015] If it exists, then further check whether the line segment connecting the two passes through an obstacle;

[0016] If the node is passed, discard the current node and select the previous node as the key node; otherwise, discard the current node and continue the above operation for the next node.

[0017] B1. Perform collinearity checks on the extracted key nodes to ensure that the three key nodes are not collinear;

[0018] C1. Extract the remaining key nodes and connect them sequentially to construct global auxiliary lines.

[0019] Preferably, the fusion of global auxiliary lines and the DWA algorithm in step 2 specifically includes the following steps:

[0020] A2. Considering the factors affecting the speed sampling space range of the patrol vehicle, including its own maximum and minimum speed factors, motor performance factors, and patrol vehicle safety factors, optimization is performed using the evaluation function of the DWA algorithm. The speed sampling space of the patrol vehicle will simultaneously satisfy the above factors, specifically manifested as follows:

[0021] V r =V m ∩V d ∩V a ;

[0022] B2. The distance and angle between the patrol device and the global auxiliary line are used as new evaluation indicators to optimize the evaluation function of the traditional DWA algorithm. The optimized evaluation function is expressed as follows:

[0023] G e =α·H s +β·L o +γ·V+λL s

[0024] In the formula, H s To simulate the angle difference between the endpoint of the trajectory and the nearest global auxiliary line, L s This is the distance between the endpoint of the simulated trajectory and the nearest global auxiliary line.

[0025] Preferably, in step 3, α and R can be determined based on the ranging range of the lidar, and the optimized navigation availability evaluation function is expressed as:

[0026]

[0027] In the formula, d is the minimum distance between the end point of the simulated trajectory and the obstacle, and δ min λ is the set minimum distance threshold. in d represents the percentage of the area of ​​obstacles that the rovers can observe within the fan-shaped region. in This is the minimum distance between the end point of the trajectory and the obstacle within the fan-shaped area;

[0028] Combining the optimized evaluation function from step B2, the final evaluation function can be expressed as follows:

[0029] G e =α·H s +β·N+γ·V+λL s .

[0030] Compared with the prior art, the beneficial effects of the present invention are:

[0031] In this invention, a global auxiliary line is first constructed using an optimized A* algorithm and then fused with the DWA algorithm. Then, with the goal of improving the navigation accuracy of the rover, the evaluation function of the fused path planning algorithm is optimized. Finally, the feasibility and adaptability to complex environments of the proposed path planning method are verified through numerical simulation and semi-physical experiments, and it can improve the accuracy of the rover's positioning and mapping. Attached Figure Description

[0032] Figure 1 This is a flowchart illustrating the overall process of the present invention.

[0033] Figure 2 This is a schematic diagram of the key nodes in this invention;

[0034] Figure 3 This is a schematic diagram of obstacle detection in this invention;

[0035] Figure 4 A schematic diagram of the path planning results for introducing unknown obstacles;

[0036] Figure 5 A schematic diagram of path planning results in a complex scenario;

[0037] Figure 6 This is a schematic diagram of path planning results in a scenario with sparse environmental features.

[0038] Figure 7 A point cloud map and a schematic diagram of the patrol vehicle's trajectory;

[0039] Figure 8 This diagram illustrates the positioning errors of different methods in different scenarios.

[0040] Figure 9 A schematic diagram of a Monte Carlo simulation;

[0041] Figure 10 A schematic diagram showing the composition of the patrol device simulation prototype and its sensors;

[0042] Figure 11 A schematic diagram a of obstacle arrangement for path planning in a semi-physical simulation test of an experimental scenario;

[0043] Figure 12 Schematic diagram b of obstacle arrangement for path planning semi-physical simulation test of experimental scenario;

[0044] Figure 13 A raster map of the experimental scene built for cartographer;

[0045] Figure 14 This is a schematic diagram of the Nokov optical 3D motion capture system.

[0046] Figure 15 This is a schematic diagram illustrating the navigation results of the present invention in a real-world environment;

[0047] Figure 16 The image shows the positioning error results of the LOAM algorithm on the X-axis.

[0048] Figure 17 This is a graph showing the positioning error results of the LOAM algorithm on the Y-axis. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] Example 1, please refer to Figure 1 This invention provides a technical solution for an autonomous path planning method for a lunar Antarctic rover oriented towards navigation availability:

[0051] An autonomous path planning method for a lunar Antarctic rover, oriented towards navigation availability, specifically includes:

[0052] Step 1: Extract key nodes from the global path nodes planned by the A* algorithm, and then use the key nodes to construct a global auxiliary line as a guide for the DWA algorithm.

[0053] Traditional A* algorithm-planned paths often suffer from redundancy and numerous inflection points, which are detrimental to the actual movement and control of the patrol vehicle. Therefore, key nodes are further extracted from the path nodes output by the A* algorithm. The process is as follows:

[0054] (1) Both the starting node and the target node are critical nodes;

[0055] (2) Starting from the starting node, traverse all path nodes until the target node is reached. First, find the key node closest to the current node, then check if there is an obstacle within the rectangular area enclosed by the two nodes. If not, discard the current node; if so, further check if the line segment connecting the two nodes passes through an obstacle. If it does, discard the current node and select the previous node as the key node; otherwise, discard the current node and continue the above operation for the next node. Figure 2 For example, there is an obstacle in the rectangular area (gray area) enclosed by node 5 and the nearest key node S, and the connecting line segment passes through the obstacle. Therefore, node 4 is selected as the key node, and the traversal continues from node 6. The above operation is repeated, and the final key nodes are S, 4, and G.

[0056] in, Figure 2 In the diagram, S and G are the center nodes of the starting grid and the target grid, respectively. 1-7 are the path grid nodes. The black area represents obstacles, the gray area represents the rectangular area formed by the current grid and the key grid, and the dashed line represents the constructed global auxiliary line.

[0057] (3) Perform collinearity judgment on the extracted key nodes to ensure that no three key nodes are on the same straight line;

[0058] (4) Extract the remaining key nodes and connect them sequentially to construct global auxiliary lines, such as... Figure 2 As shown by the dashed line in the image.

[0059] Step 2: By optimizing the evaluation function of the traditional DWA algorithm, the global auxiliary line is integrated with the DWA algorithm. This includes optimizing the evaluation function to include the distance and angle between the patrol and the global auxiliary line.

[0060] The traditional DWA algorithm first samples the velocity space of the rovers, then uses the sampled velocities to simulate the trajectory, and finally obtains the optimal trajectory through an evaluation function.

[0061] The trajectory calculation formula for a 3-DOF rover is:

[0062]

[0063] In the formula, v and ω are the velocity and angular velocity of the rover at the current moment, respectively, and θ tΔt is the heading angle, and Δt is the simulation time interval.

[0064] Considering that the factors affecting the speed sampling range of the patrol device are limited by its own performance and the environment, these factors include its own maximum and minimum speed factors, motor performance factors, and patrol device safety factors:

[0065] (1) The rover is limited by its own maximum and minimum speed:

[0066] V m ={v∈[v min ,v max ],ω∈[ω min ,ω max ]}

[0067] In the formula, v min ,v max These are the minimum and maximum speeds of the rover, ω and ω', respectively. min ,ω max These are the minimum and maximum angular velocities of the rover, respectively.

[0068] (2) The patrol device is limited by the performance of the motor, resulting in a maximum acceleration or deceleration limit:

[0069]

[0070] In the formula, These are the maximum acceleration and maximum deceleration of the rover, respectively. These are the maximum angular acceleration and maximum angular deceleration of the rover, respectively.

[0071] (3) Based on the safety considerations of the patrol vehicle, in order to allow it to stop before encountering an obstacle, the speed of the patrol vehicle is limited:

[0072]

[0073] In the formula, dist(v,ω) is the minimum distance between the end point of the simulated trajectory of the patrol vehicle and the obstacle.

[0074] The velocity sampling space of the rover will simultaneously satisfy the above three conditions, specifically manifested as follows:

[0075] V r =V m ∩V d ∩V a

[0076] Next, the evaluation function is used to select the optimal trajectory. The evaluation function of the traditional DWA algorithm is:

[0077] G=α·H t +β·L o+γ·V

[0078] In the formula, H t To simulate the difference in azimuth between the endpoint of the trajectory and the target point, L o V represents the minimum distance between the end point of the simulated trajectory and the obstacle, while V is the velocity of the rovers at the current moment.

[0079] To avoid any one factor having too much weight in the evaluation function, H t L o. V has been normalized, and α, β, and γ are weighting coefficients.

[0080] Then, global auxiliary lines are integrated with the DWA algorithm. The traditional DWA algorithm uses only a single target point as guidance, and its planning effect depends on the selection of weight coefficients. In complex environments with many obstacles, the rovers are prone to planning failure or getting trapped in local optima. Therefore, global auxiliary lines are introduced to guide the rovers.

[0081] By using the distance and angle between the patrol device and the global auxiliary line as new evaluation metrics, the evaluation function of the traditional DWA algorithm is optimized. The optimized evaluation function is expressed as follows:

[0082] G e =α·H s +β·L o +γ·V+λL s

[0083] In the formula, H s To simulate the angle difference between the endpoint of the trajectory and the nearest global auxiliary line, L s This is the distance between the endpoint of the simulated trajectory and the nearest global auxiliary line.

[0084] Step 3: Using the fusion path planning algorithm obtained in Step 2, draw a fan-shaped region with a radius of R, where the angle between the simulated trajectory endpoint of the DWA algorithm at the current moment and the heading angle of the rovers is α. Then, use the ratio of the area ratio of obstacles within the fan-shaped region to the distance between the nearest obstacle to the trajectory endpoint as an evaluation index for further optimization.

[0085] The accuracy of positioning and mapping ensures the success rate and effectiveness of path planning. Although combining global path planning algorithms with local path planning algorithms can improve the success rate of existing path planning algorithms to a certain extent, in the more complex and unknown environment of the lunar south pole, the rover is very likely to fail to obtain the correct positioning information, resulting in planning failure.

[0086] The rover relies heavily on data from its onboard sensors for autonomous navigation. However, the sparse and textured environment of the lunar south pole limits the effectiveness of visual sensors such as cameras or lidar, making it difficult to acquire sufficient visual feature information and affecting the accuracy of navigation algorithms that depend on feature registration. Furthermore, the high similarity among some visual features on the lunar surface introduces significant ambiguity into the rover's feature registration and localization estimation. In the lunar south pole, significant visual information or effective navigational markers can only be obtained through obstacles with significant undulations, such as rocks, craters, and slopes. Therefore, the rover's planned path must fully utilize obstacle information. In addition, to enable the lidar to obtain more accurate ranging information, the rover should move as close to obstacles as possible, while ensuring safety.

[0087] Based on the above, with the goal of improving navigation accuracy, the obstacle distance term in the evaluation function of the fusion path planning algorithm is further optimized to obtain the navigation availability evaluation function. This allows the rover to acquire significant visual information without colliding with obstacles. The optimization method is as follows:

[0088] In the DWA algorithm, a sector-shaped region with radius R (α and R are determined based on the ranging range of the lidar) is plotted at the end point of the simulated trajectory at the current moment, with an angle α between the end point and the heading angle of the rover. Figure 3 As shown in the figure, the black curve represents the simulated trajectory generated by the DWA algorithm at the current moment, and the dashed area represents a fan-shaped region with a radius of R at which the heading angle between the end point of the simulated trajectory and the rover is α. Then, the ratio of the area of ​​obstacles within the fan-shaped region to the distance between the nearest obstacle to the end point of the trajectory is used as the evaluation index. The optimized navigation availability evaluation function is expressed as:

[0089]

[0090] In the formula, d is the minimum distance between the end point of the simulated trajectory and the obstacle, and δ min λ is the set minimum distance threshold. in d represents the percentage of the area of ​​obstacles that the rovers can observe within the fan-shaped region. in This is the minimum distance between the end point of the trajectory and the obstacle within the fan-shaped area;

[0091] Finally, the evaluation function is optimized by combining global auxiliary lines and the DWA algorithm. The final evaluation function can be expressed as follows:

[0092] G e =α·H s +β·N+γ·V+λL s

[0093] Example 2:

[0094] To verify the effectiveness of the path planning method proposed in this invention, the following settings were configured: Figure 4-6 The three simulation scenarios shown have the following settings: the radius of the sector region is R = 4, and the included angle is α = 60°.

[0095] First, the local obstacle avoidance capability of the rovers was verified. Two unknown obstacles (objects not detected by the lidar or objects that suddenly appeared) were added to the constructed global map. Then, path planning was performed using the optimized A* algorithm and the present invention, respectively. The target point was located at (10, 9). The weight parameters of the evaluation function of the present invention are as follows:

[0096] α=2, β=0.2, γ=1, λ=2,

[0097] The final simulation results are as follows Figure 4 As shown in the figure. Simulation results show that the global path planned using the optimized A* algorithm cannot avoid unknown obstacles, while the path planning method proposed in this invention can detect new obstacles in real time and avoid them.

[0098] To verify the applicability of the path planning method of the present invention to complex environments, the following were set up: Figure 5 In the complex scenario shown, the same target point (10, 9) is first selected. Then, path planning is performed using the DWA algorithm and the present invention, respectively. The simulation results are as follows: Figure 5 (1) and Figure 5 As shown in (2). Figure 5 (1) The result of path planning using only the DWA algorithm. Figure 5 (2) As the result of path planning using this invention, the weight parameters of the evaluation function of this invention are as follows:

[0099] α=5, β=0.1, γ=1, λ=5,

[0100] It can be seen that relying solely on the DWA algorithm for path planning in complex environments can easily lead to local optima and path planning failure. However, by introducing global auxiliary lines, the rovers successfully reached the target point.

[0101] To verify the application of the path planning algorithm proposed in this invention in scenarios with sparse environmental features, the following settings are configured: Figure 6 In the scenario shown, the target point is selected as (7, 6). Path planning is performed using both the fusion algorithm and this invention. The weight parameters of the evaluation functions for both algorithms are...

[0102] α=3, β=0.5, γ=0.1, λ=5,

[0103] The final simulation results are as follows Figure 6 (1) and Figure 6 As shown in (2), Figure 6 (1) The simulation results are for planning using only the fusion algorithm. Figure 6 (2) The result of using the present invention for path planning;

[0104] Figure 4-6 The black areas represent known obstacles, the gray areas represent unknown obstacles, the asterisks indicate the target's location, the solid lines represent the path trajectory finally planned by the DWA algorithm, and the dashed lines represent the global auxiliary lines constructed by the optimized A* algorithm. It can be seen that the path planned by the fusion algorithm, optimized for improved navigation accuracy, makes better use of obstacle information and gets closer to the obstacles.

[0105] The simulation results show that the path planning method proposed in this invention is feasible. In complex environments, compared with a single path planning algorithm, the proposed path planning method is more reliable and can make fuller use of significant obstacle information.

[0106] Example 3:

[0107] Numerical simulations were used to verify the improvement in navigation accuracy achieved by this invention. Two sets of simulation scenarios were selected to simulate the rover's exploration and exploration activities at the lunar south pole (e.g., Figure 7 (1) and Figure 7 (2) shown) Figure 7 (1) The point cloud map of scene 1 and the driving trajectory of the patrol vehicle; Figure 7 (2) The point cloud map of scene 2 and the driving trajectory of the patrol vehicle; Figure 7 The dashed line represents the state value estimated by the radar point cloud as the rover travels along the path planned in this invention, while the solid line represents the true trajectory value of the rover. The rover is equipped with a lidar, which has an effective detection range of 50 meters, a horizontal field of view of 360°, and a vertical field of view of ±15°.

[0108] To increase the diversity and credibility of the simulation, an ICP (Iterative Closest Point) reobservation algorithm is introduced to verify whether navigation accuracy can be further improved when the rover travels along the path planned by the proposed algorithm during the reobservation process.

[0109] The following four simulations were set up: (1) the rover travels along a randomly selected path; (2) the rover travels along the path planned in this invention; (3) under the same conditions as (1), ICP re-observation is performed every 20 seconds; (4) under the same conditions as (2), ICP re-observation is performed every 20 seconds. The navigation accuracy was further evaluated by recording the positioning error of each simulation.

[0110] The single simulation results based on scenario 1 and scenario 2 are as follows: Figure 8 (1) and Figure 8 As shown in (2), Figure 8 (1) The positioning error of different methods in scenario 1; Figure 8 (2) The positioning error of different methods in scenario 2. It can be seen from the single simulation results that when the patrol vehicle walks along the path planned by the present invention, the positioning error can be reduced to a certain extent. The positioning error results of simulation (3) and simulation (4) are similar because the selected point cloud map is noise-free.

[0111] To further verify the validity of the simulation results, a Monte Carlo simulation was performed. Sixty different target points were randomly selected in both Scene 1 and Scene 2. The final simulation results are as follows: Figure 9 (1) and Figure 9 As shown in (2), Figure 9 (1) A schematic diagram of the Monte Carlo simulation results of the positioning error MAE (Mean Absolute Error) in scenario 1; Figure 9 (2) A schematic diagram of the Monte Carlo simulation results of the positioning error MAE in scenario 2;

[0112] Tables 1 and 2 record the cumulative average absolute positioning error for different simulations in two sets of scenarios.

[0113]

[0114] As can be seen from the simulation results in Tables 1 and 2, the results of the single simulation are consistent with those of the Monte Carlo simulation. Under different scenarios and by selecting different target points, the navigation accuracy of the rover is improved to a certain extent when it travels along the path planned by this invention. This demonstrates the applicability of this invention and its effectiveness in suppressing positioning errors.

[0115] Example 4:

[0116] A semi-physical experiment was set up to verify the improvement effect of this invention on the navigation accuracy of the rover in practical applications. The rover and indoor scenario used in the semi-physical experiment are as follows: Figure 10 , Figure 11 and Figure 12 As shown. The main components of the robot include: one SLAMTEC 2D LiDAR, one VLP16 LiDAR, three reflective markers, and two controllers equipped with the ROSMelodic system based on Ubuntu 18.04.

[0117] First, the patrol robot is controlled by the keyboard to move around, while the mapping node of the Cartographer algorithm is activated to build a map of the actual scene. The constructed global raster map is as follows: Figure 13As shown. LOAM (Lidar Odometry and Mapping in Real Time) is a SLAM algorithm that relies on registration between feature point clouds for localization and mapping. To verify whether this invention can help the rover acquire sufficient visual features, thereby further improving the accuracy of localization and mapping, the localization accuracy of the LOAM algorithm is introduced as a new evaluation metric. Simultaneously, in the selected experimental environment, the Nokov optical 3D motion capture system (with 12 motion capture lenses) produced by Beijing Metrology Technology Co., Ltd. was used to obtain the rover's ground truth localization value, such as... Figure 14 As shown.

[0118] The following three sets of experiments were set up: (1) the rover was controlled to walk using a keyboard; (2) the rover walked autonomously along the path planned by the fusion path planning algorithm; and (3) the rover walked autonomously along the path planned by this invention. While the rover was moving, the LOAM algorithm and the Nokov optical 3D motion capture system were activated to estimate the rover's position. The kinematic parameters of the rover during the navigation experiment are shown in Table 3. Figure 15 The navigation results of experiment (3) are presented. Figure 15 (1) A schematic diagram of the position of the rover at time 1; Figure 15 (2) A schematic diagram of the roving device at the target location.

[0119]

[0120] In the actual experiment, as the rovers moved along the paths of each experiment, the positioning errors of the LOAM algorithm on the X and Y axes were as follows: Figure 16 and Figure 17 As shown (where the position information collected by the Nokov optical 3D motion capture system has been mapped and aligned to the coordinate system of the LOAM algorithm). Table 4 shows the positioning errors (MAE) of the LOAM algorithm on the X and Y axes in each experiment.

[0121]

[0122] The experimental results above show that when the rovers travel along the path planned in this invention, the positioning accuracy of the LOAM algorithm is improved in both the X and Y axes. This is mainly because the rovers can acquire more sufficient and more accurate salient visual features during the movement.

[0123] In summary, based on the environmental characteristics of the lunar south pole, this invention discloses a novel path planning method oriented towards navigation availability. First, a global auxiliary line is constructed using an optimized A* algorithm and then fused with the DWA algorithm. Next, with the goal of improving the rover's navigation accuracy, the evaluation function of the fused path planning algorithm is optimized. Finally, numerical simulation and semi-physical experiments verify the feasibility of the proposed path planning method and its adaptability to complex environments, and demonstrate that it can improve the accuracy of the rover's positioning and mapping.

[0124] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for autonomous path planning for a lunar south pole rover with navigation benefits, characterized in that: Specifically comprising the following steps: Step 1, extracting key nodes in the global path nodes planned by the A* algorithm, and then constructing a global auxiliary line using the key nodes as a guide for the DWA algorithm; Step 2, fusing the global auxiliary line with the DWA algorithm by optimizing the evaluation function of the traditional DWA algorithm, wherein the optimized evaluation function includes the distance and the included angle between the rover and the global auxiliary line; Step 3, drawing a sector area with an included angle of α and a radius of R using the heading angle between the end point of the simulation trajectory of the DWA algorithm at the current time and the rover as the angle, and then using the ratio of the area proportion of the obstacles in the sector area to the distance between the closest obstacle to the end point of the trajectory as an evaluation index for further optimization; Wherein, the fusion of the global auxiliary line and the DWA algorithm in step 2 specifically comprises the following steps: A2, considering the factors affecting the velocity sampling space range of the rover, including the maximum and minimum velocity factors, motor performance factors and safety factors of the rover, respectively optimizing the evaluation function of the DWA algorithm, the velocity sampling space of the rover will meet the above factors at the same time, and the specific form is: V r = V m ∩ V d ∩ V a ; B2, taking the distance and the included angle between the rover and the global auxiliary line as new evaluation indexes, optimizing the evaluation function of the traditional DWA algorithm, and the optimized evaluation function is represented as: G e = a · H s + b · L o + g · V + lL s where H s is the angle difference between the end point of the trajectory and the closest global auxiliary line to it, L s is the distance between the end point of the trajectory and the closest global auxiliary line to it. In step 3, α and R can be determined according to the ranging range of the laser radar, and the optimized navigation advantage evaluation function is represented as: where d is the minimum distance between the end point of the simulated trajectory and the obstacle, δ min is the set minimum distance threshold, λ in is the area ratio of the obstacle that the inspector can observe within the fan-shaped region, d in is the minimum distance between the end point of the trajectory and the obstacle within the fan-shaped region; Combined with the optimized evaluation function in step B2, the final evaluation function can be represented as, G e = a · H s + β · N + γ · V + λ L s .

2. The method according to claim 1, wherein the method is characterized in that: The extraction of key nodes in step 1 specifically includes: A1, setting the starting node and the target node as key nodes, starting from the starting node, traversing all path nodes until reaching the target node, finding the closest key node to the current node, and then detecting whether there are obstacles in the rectangular area surrounded by the two nodes; If not, discard the current node; If so, further detect whether the line segment connecting the two nodes passes through an obstacle; If so, discard the current node, select the previous node as the key node, otherwise, discard the current node, continue the above operation on the next node; B1, collinearity judgment is performed on the extracted key nodes to ensure that the three key nodes are not collinear; C1, extracting the remaining key nodes and connecting them in turn to construct a global auxiliary line.

Images