Path planning method for unmanned sweeper to automatically search parking space

Abstract

The invention discloses a path planning method for an unmanned sweeper to automatically search for parking spaces. The method comprises the following steps: generating a channel topological graph of candidate parking spaces; calculating an optimal entry point of the vehicle entering the channel topological graph; calculating a path from the starting point of the vehicle to the optimal entry point of the channel topological graph; according to the topological relation of the channel topological graph, a next channel path node tonodeid is searched from a channel path node startnodeid corresponding to the optimal entry point, all channel path nodes are traversed, and a planned path passing through all candidate parking spaces is generated. According to the method, the candidate parking spaces can be searched in advance, the driving path is planned for the vehicle in advance, it is guaranteed that the vehicle accurately and rapidly enters the channel topological graph, the problem of blind searching of the parking spaces is avoided, meanwhile, the requirement of searching for the parking spaces in advance in some special scenes can be met, and the parking efficiency is effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of unmanned driving, and in particular to a path planning method for automatic searching of a parking space by an unmanned cleaning vehicle. BACKGROUND

[0002] Generally, an unmanned vehicle scans and identifies available parking spaces around the vehicle through the sensing components (cameras, radars, etc.) of the vehicle itself, and after obtaining candidate parking spaces, the vehicle is parked in a designated position according to the principle of proximity or the principle of convenient parking.

[0003] The above method can only be used for searching for a parking space while driving after entering a parking lot, and cannot meet the requirements in some special cases. For example, when an unmanned cleaning vehicle performs a cleaning task in an open space that includes multiple garbage can positions, charging stations, parking spaces, water replenishment stations, and supply stations; during the execution of the cleaning task, the unmanned cleaning vehicle inevitably encounters planning tasks such as searching for an idle garbage can to dump garbage when the garbage in the garbage can is full, searching for an idle charging position to charge when the power is insufficient, searching for an idle water replenishment station to replenish water when the water in the water tank is insufficient during wet cleaning in dry weather, and searching for an idle parking space to park nearby after completing the cleaning task.

[0004] In addition, some point-to-point navigation planning algorithms are disclosed in the prior art, for example, the invention patent with the application number 202310629106.7 and the name of the global path planning method, electronic device and storage medium, discloses a global path planning method, which constructs a global road topology structure by obtaining map data, and realizes the global planning path of the vehicle based on the starting point and the ending point of the vehicle. The technology of the invention can be used to provide an automatic driving function for the vehicle, and realizes point-to-point navigation planning, so it can be used for automatic navigation of the vehicle to the parking lot, but cannot meet the requirements of automatic search for parking spaces and parking.

[0005] Obviously, the prior art cannot search for parking spaces in advance, cannot simultaneously meet the requirements of automatic search for parking spaces and planning of the path to the parking space, and therefore cannot meet the requirements of the above task planning scenarios, and cannot realize precise parking, which affects the operation of the unmanned cleaning vehicle. In addition, the automatic parking of the prior art is realized by collecting data and calculating on site through the sensing components and control system on the vehicle, which cannot simultaneously compare and analyze multiple available parking spaces, and has problems such as multiple backward operations and low parking fluency, resulting in low parking efficiency.

[0006] Therefore, the prior art still needs to be improved. SUMMARY

[0007] In view of the deficiencies of the prior art described above, the purpose of the present application is to provide a path planning method for automatic searching of parking spaces by an unmanned cleaning vehicle, aiming to solve the problem in the prior art that automatic searching and finding of parking spaces by an unmanned cleaning vehicle and planning of paths to the parking spaces cannot be achieved, and at the same time solve the problem of low parking efficiency in the prior art.

[0008] The technical solution of the present application is as follows: a path planning method for automatic searching of parking spaces by an unmanned cleaning vehicle, the method comprising:

[0009] generating a channel topology graph of candidate parking spaces; calculating an optimal cut-in point of a vehicle entering the channel topology graph; calculating a path of the vehicle from a starting point to the optimal cut-in point of the channel topology graph; according to a topological relationship of the channel topology graph, finding a next channel path node to_node_id from a channel path node start_node_id corresponding to the optimal cut-in point, traversing all channel path nodes, and generating a planning path passing above all candidate parking spaces.

[0010] The path planning method for automatic searching of parking spaces by an unmanned cleaning vehicle in the present application can generate a channel topology graph of candidate parking spaces of a predicted parking site, for a vehicle not on the channel topology graph, the optimal cut-in point of the vehicle entering the channel topology graph is further calculated, so as to ensure that the vehicle travels to the channel topology graph at a short distance and in a short time, thereby ensuring that the travel path of the vehicle to the channel topology graph is an optimal travel path, and effectively improving the parking efficiency; after the path calculation of the vehicle from the starting point to the optimal cut-in point of the channel topology graph is completed, the travel path of the vehicle to each candidate parking space after entering the channel topology graph is further calculated, specifically by traversing the paths of each parking space in the channel topology graph according to a topological relationship search order, and finally planning a planning path of the vehicle to above each candidate parking space, to provide an accurate travel route for the vehicle; for a vehicle located in the channel topology graph, the last step of traversing the paths of each parking space according to the topological relationship search order is directly performed, and a planning path of the vehicle to above each candidate parking space is planned, so that the candidate parking space can be quickly searched, and the quick and accurate parking of the vehicle can be ensured. Obviously, for the unmanned cleaning vehicle encountering situations such as finding an idle garbage can for garbage dumping when the cleaning task is performed and the garbage in the garbage can is full, finding an idle charging position for charging when the electric quantity is insufficient, finding an idle water replenishing station for water replenishing when the water quantity in the water tank is insufficient during wet cleaning in dry weather, and finding an idle parking space for parking near the cleaning terminal after the cleaning task is performed, the present application can plan a travel path in advance, so as to ensure that the candidate parking space can be accurately and quickly found, and the vehicle can be parked in the candidate parking space, thereby avoiding blind searching for a parking space and effectively improving the parking searching efficiency and accuracy.

[0011] In an embodiment, the step of generating the channel topology graph of the candidate parking spaces comprises:

[0012] Checking whether the space above each empty parking space meets the parking space requirement.

[0013] Taking the empty parking space meeting the parking space requirement as a candidate parking space, and performing parking space clustering and merging.

[0014] According to the result of the parking space clustering and merging, performing parking space traversal sorting on the candidate parking spaces meeting the requirement. In this way, the candidate parking spaces in the same cluster can be sequentially arranged on a channel in a certain direction.

[0015] According to the result of the parking space traversal sorting, calculating and generating channel path nodes. The gradient descent smoothing algorithm is used to generate the channel path nodes.

[0016] According to the generated channel path nodes, calculating the connection edges between the channels, and generating a channel topology graph. The channel topology graph is a channel topology graph above the candidate parking spaces.

[0017] In an embodiment, the process of calculating the connection edges between the channels according to the generated channel path nodes comprises:

[0018] S100: Arbitrarily selecting a channel path node from the set of channel path nodes as a starting channel path node from_node.

[0019] S101: Judging whether the head and tail of all channel path nodes have been connected. If yes, directly ending, indicating that the head and tail of all channel path nodes have been connected, and the generation of the channel connection edges is completed.

[0020] S102: If not, traversing all channel path nodes other than the starting channel path node from_node, and calculating the distance dist and the heading difference diff_theta between the pre-parking point s_point of the tail of the from_node and the pre-parking point e_point of the head of the other node.

[0021] S103: Calculating the cost value, and judging whether the tail of the other node has been connected. Specifically, the cost = dist + 10.0 x diff_theta.

[0022] S104: If not, selecting the other node with the minimum cost value as the next channel path node to_node.

[0023] S105: Calculating and generating the planning connection path from the from_node to the to_node.

[0024] S106: generating a connection edge.

[0025] After step S106 is completed, return to step S101 and repeat steps S101-S106. After the loop calculation is completed, the process of generating the connection edge of the channel is completed, and the other_node with the minimum cost value is selected as the next channel path node to_node, effectively ensuring that the generated connection edge is the shortest distance and the channel has high smoothness.

[0026] In an embodiment, if the tail of the other_node is connected, a first penalty value is added to the cost value, which is used as a new cost value, and the other_node with the minimum cost value in the new cost value is selected as the to_node of the next channel path node, and steps S105-S106 are repeated. The purpose of adding the first penalty value is that for the other_node whose tail is connected, after adding the first penalty value, the cost value is larger, and it will not be selected again, so that the selected next other_node is not connected at both ends (head and tail). Therefore, the other_node whose tail is not connected can be connected preferentially. In addition, it should be noted that the first penalty value is an empirical value, which can be adjusted according to the actual demand and the adjustment experience to achieve global optimization.

[0027] In an embodiment, the RS curve and / or the quintic polynomial are used in step S105 to calculate the planned connection path from the from_node to the to_node.

[0028] In an embodiment, the basis for candidate parking space clustering and merging is:

[0029] If the distance between the pre-parking points above the candidate parking space is less than the parking distance threshold, the heading difference between the pre-parking points is less than the parking heading threshold, and the straight line connection has no collision, the candidate parking space is clustered and merged into the same parking set.

[0030] In an embodiment, the step of calculating the optimal cut-in point of the vehicle entering the channel topology graph is:

[0031] S200: traversing all channel path nodes node on the channel topology graph.

[0032] S201: determining whether the channel topology graph is bidirectional. If not, the calculation process is directly ended.

[0033] S202: If yes, calculate the head node distance dist_to_head of the vehicle's starting point to the head node of the node, and calculate the heading difference diff_theta_to_head of the vehicle's starting point to the head node of the node.

[0034] S203: Calculate the head cost head_cost of the vehicle's starting point to the head. Specifically, the head_cost = dist_to_head + 7.5 x diff_theta_to_head.

[0035] S204: Determine whether the vehicle's starting point and the head of the node can be directly connected.

[0036] S205: If no, calculate the tail node distance dist_to_tail of the vehicle's starting point to the tail node of the node, and calculate the heading difference diff_theta_to_tail of the vehicle's starting point to the tail node of the node.

[0037] S206: Calculate the tail cost tail_cost of the vehicle's starting point to the tail. Specifically, the tail_cost = dist_to_tail + 7.5 x diff_theta_to_tail.

[0038] S207: Determine whether the vehicle's starting point and the tail of the node can be directly connected.

[0039] S208: If no, select the target point with the smallest tail_cost, and take the target point as the optimal cut-in point.

[0040] Obviously, after steps S201-S208, the optimal cut-in point from the vehicle's starting point to the channel topology graph can be calculated. In this process, each group of parallel candidate parking spaces formed by traversing and sorting in the channel topology graph is compared, and finally the target point with the smallest tail_cost is selected as the optimal cut-in point, effectively ensuring that the cut-in point from the vehicle's starting point into the channel topology graph is the closest, the path to the optimal cut-in point is smooth in the later navigation, effectively saving the parking time of the vehicle, solving the problem of low parking flow in the prior art, and effectively improving the low parking efficiency.

[0041] In an embodiment, if the result of step S204 is no, the head_cost minus a second penalty value is calculated, and the result is taken as a new head_cost, and steps S205-S208 are repeated. The second penalty value is also an experience value, which can be adjusted according to the actual demand and the adjustment experience to achieve global optimization. After the original head_cost minus the second penalty value, the value is smaller, which can ensure that the node directly at the head of the node (the starting point) is preferentially selected.

[0042] In an embodiment, if the result of step S207 is no, the tail_cost minus a second penalty value is calculated, and the result is taken as a new tail_cost, and step S208 is repeated. Similarly, after the second penalty value is subtracted, it can be ensured that the node directly at the tail of the node is preferentially selected.

[0043] In an embodiment, a navigation planning algorithm is used to calculate the path of the vehicle from the starting point to the optimal entry point of the channel topology graph. Specifically, a point-to-point navigation planning algorithm is used to calculate an optimal driving path from the starting point of the vehicle to the channel topology graph, reducing driving distance and time, and reducing energy consumption and cost.

[0044] In summary: the path planning method for automatic searching of a parking space by an unmanned cleaning vehicle proposed in this paper has the following effects:

[0045] 1. The candidate parking space can be searched in advance, and the driving path can be planned in advance to ensure that the vehicle accurately and quickly enters the channel topology graph, avoiding the problem of blind searching for a parking space, and meeting the demand of searching for a parking space in advance in some special scenarios, thereby effectively improving the parking efficiency.

[0046] 2. After entering the channel topology graph, each candidate parking space can be traversed, and a planning path of the vehicle to each candidate parking space above can be planned to provide an accurate driving route for the vehicle. The vehicle can select the optimal candidate parking space according to the demand, which can effectively improve the smoothness of the automatic parking process.

[0047] 3. The vehicle enters the channel topology graph through the optimal entry point, and the path of the vehicle from the starting point to the optimal entry point of the channel topology graph is calculated in a point-to-point manner, which ensures the shortest driving distance and the fastest parking speed, reduces energy consumption and saves cost, and effectively improves the parking efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0048] The application will be further described below with reference to the accompanying drawings and embodiments. In the drawings:

[0049] Figure 1 The flowchart of steps S1-S4 of the application is shown in the drawings.

[0050] Figure 2 Flowchart for steps S10-S14 of the present application;

[0051] Figure 3 Schematic diagram of the channel topology map of the present application;

[0052] Figure 4 Schematic diagram of the planning path from the starting point of the vehicle to above the candidate parking space generated by the present application. DETAILED DESCRIPTION

[0053] In order to make the objectives, technical solutions and effects of the present application clearer and more explicit, the present application is further described in detail below. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application. The embodiments of the present application are described below with reference to the accompanying drawings.

[0054] Please refer to Figures 1-4 The present application provides a path planning method for an unmanned cleaning vehicle to automatically search for a parking space, which can realize the planning of the driving path of a vehicle located in a predicted parking lot during the parking process, so as to achieve fast and high-precision parking. At the same time, the planning of the optimal driving path for a vehicle not in the predicted parking lot to the predicted parking lot can also be realized, so as to achieve fast and efficient and high-precision parking. Specifically, the present application can generate a channel topology map of candidate parking spaces in a predicted parking lot. For a vehicle not on the channel topology map, the optimal entry point into the channel topology map is further calculated to ensure that the vehicle travels a short distance and for a short time to the channel topology map, and an optimal driving path for the vehicle to reach the channel topology map is calculated to effectively shorten the time for the vehicle to reach the predicted parking lot. The driving path for the vehicle to reach each candidate parking space after entering the channel topology map is further calculated, specifically by searching the paths of each parking space in the channel topology map in the order of the topological relationship, and finally planning a planning path for the vehicle to reach above each candidate parking space. For a vehicle located in the channel topology map, the last step of searching the paths of each parking space in the order of the topological relationship is directly performed to plan a planning path for the vehicle to reach above each candidate parking space. Obviously, for the two different cases, the present application can provide accurate driving routes for the vehicle to enter the candidate parking space, effectively improving the smoothness and parking efficiency of the automatic parking process.

[0055] Specifically, in the present embodiment, the path planning method for an unmanned cleaning vehicle to automatically search for a parking space in the present application includes the following steps:

[0056] S1: generating a channel topology map of candidate parking spaces.

[0057] It should be noted that the lane topology graph of the candidate parking space is generated according to the map data of the high-definition map, therefore, the map data of the predicted parking lot needs to be acquired first, and the acquired map data includes: Lane road and Edge connection edge information, Openspace open space boundary information, Obstacle obstacle boundary and ParkingSpace parking space boundary information.

[0058] Specifically, in the embodiment, the high-definition map (HD Map) refers to a digital map constructed by collecting, processing and analyzing city roads, buildings, traffic signs and other information based on laser radar, high-resolution camera, satellite navigation and other technical means. It can provide detailed and high-precision map data of the predicted parking lot where the vehicle will park, and the map data is comprehensive and accurate, which can ensure the accuracy of the later calculation.

[0059] Specifically, the step of generating the lane topology graph of the candidate parking space includes:

[0060] S10: Verify whether the upper part of each empty parking space meets the parking space requirement.

[0061] Specifically, in the embodiment, the boundary point information and the surrounding environment information of the acquired parking space are analyzed, and the size of the vehicle frame is combined for joint determination. Specifically, generally, if the vehicle is located directly above the empty parking space, the horizontal directly above 50 cm is collision-free (it is indicated that there is no obstacle, and the empty parking space can accommodate the vehicle), it is considered that the empty parking space has parking space and meets the parking space requirement. In addition, it should be noted that the upper part of the empty parking space is selected to verify whether it meets the parking space requirement instead of directly verifying the empty parking space, because whether there is an obstacle in the idle parking space can be quickly captured by some perception data or laser radar data, so that it can be directly judged whether the empty parking space can be used as a candidate parking space. If there is an obstacle, there is no need to perform other calculations and judgments. Obviously, the accuracy of the result of the calculated candidate parking space is higher.

[0062] S11: The empty parking space meeting the parking space requirement is taken as a candidate parking space, and the parking space clustering and merging is performed.

[0063] Specifically, in the embodiment, the basis for the parking space clustering and merging of the candidate parking space is specifically:

[0064] If the distance between the pre-parking points above the candidate parking spaces is less than the parking distance threshold, the heading difference between the pre-parking points is less than the parking heading threshold, and the straight line connection has no collision, the clustering is merged into the same parking space set. It should be noted that the parking distance threshold and the heading threshold are set according to actual needs. In addition, the straight line connection without collision generally refers to candidate parking spaces on the same horizontal line, for example, parallel and side-by-side candidate parking spaces.

[0065] S12: According to the result of the parking space clustering and merging, the candidate parking spaces meeting the requirements are sorted by parking space traversal.

[0066] Specifically, in this embodiment, for the parking space set {parking_space_1, parking_space_2, …, parking_space_n} clustered and merged into one category, the (x, y) coordinates of the pre-parking points above the candidate parking spaces are sorted in descending order of x and then in descending order of y. In this way, the candidate parking spaces clustered and merged into one category can be ordered in a certain direction on a channel. For details, please refer to the attached drawings. Figure 3 .

[0067] S13: According to the result of the parking space traversal sorting, the channel path nodes are calculated and generated.

[0068] Specifically, the main direction is confirmed according to the pre-parking points on the candidate parking space sequence after the traversal sorting. After confirming the main direction, the corresponding candidate parking space sequence is strung along the main direction. Finally, the gradient descent smoothing algorithm is used to generate the channel path nodes.

[0069] Specifically, the gradient descent smoothing algorithm is a common optimization algorithm mainly used to solve the minimum value of the objective function. It iteratively updates the parameters along the negative gradient direction to gradually approach the minimum value of the objective function. In this embodiment, the mathematical calculation model of the gradient descent smoothing algorithm is the existing structure of "preliminary path + optimization smoothing". For details, please refer to the process of calculating and generating the channel path nodes of the road by using the gradient descent smoothing algorithm in the prior art.

[0070] S14: According to the generated channel path nodes, the connection edges between the channels are calculated, and the channel topology graph is generated.

[0071] Specifically, in this embodiment, the specific process of calculating the connection edges between the channels according to the generated channel path nodes is as follows:

[0072] S100: Arbitrarily select a channel path node from the set of channel path nodes as a starting channel path node from_node.

[0073] S101: Determine whether the head and tail of all channel path nodes have been connected.

[0074] If the result of S101 is yes, it is directly ended, which means that the head and tail of all the channel path nodes have been connected and the generation of the connection edge of the channel is completed.

[0075] S102: If no, all the channel path nodes other than the starting channel path node from_node are traversed and the distance dist and the difference in heading diff_theta between the pre-parking point s_point of the tail of the from_node and the pre-parking point e_point of the head of the other_node are calculated.

[0076] S103: The cost value is calculated and it is judged whether the tail of the other_node has been connected. Specifically, the cost = dist + 10.0 x diff_theta.

[0077] S104: If no, the other_node with the minimum cost value is selected as the next channel path node to_node.

[0078] S105: The planning connection path from the from_node to the to_node is calculated.

[0079] Specifically, in the embodiment, the RS curve and / or the quintic polynomial are used to calculate the planning connection path from the from_node to the to_node in the step S105. Generally, the RS curve is preferred, and when the RS curve cannot be used to calculate the planning connection path from the from_node to the to_node, the quintic polynomial can be further used to calculate. Other related calculation methods which can be used to calculate the planning connection path from the from_node to the to_node in the prior art can also be selected according to the requirements.

[0080] S106: The connection edge connection_edge is generated.

[0081] After the step S106 is completed, the step S101 is returned to and the steps S101-S106 are repeated. After the loop calculation is completed, the process of generating the connection edge of the channel is completed, the other_node with the minimum cost value is selected as the next channel path node to_node, which effectively ensures that the generated connection edge is the shortest and the flow of the channel is high.

[0082] Specifically, in the embodiment, if the tail of the other_node has been connected, i.e., the result of the determination in step S103 is YES, the first penalty value is added to the cost value, and the new cost value is used as the new cost value, and in the embodiment, the first penalty value is 1000, i.e., new cost = original cost + 1000; the other_node with the minimum cost value in the new cost value is selected as the to_node of the next channel path node, and steps S105-S106 are repeated. The purpose of adding the first penalty value is that for the other_node whose tail has been connected, after the first penalty value is added, the cost value is larger, and the other_node will not be selected again, so that the selected next other_node is not connected through both ends (head and tail), and thus, the other_node whose tail has not been connected can be connected preferentially.

[0083] Obviously, after the calculation in steps S10-S14, the channel topology graph can be obtained, which is the channel topology graph above the candidate parking space.

[0084] For the vehicle not on the channel topology graph, the optimal cut-in point on the channel topology graph needs to be further planned for the vehicle to ensure that the distance from the starting point of the vehicle to the channel topology graph is the shortest, and the path from the starting point of the vehicle to the optimal cut-in point is smooth.

[0085] S2: Calculate the optimal cut-in point of the vehicle into the channel topology graph.

[0086] Specifically, in the embodiment, the step of calculating the optimal cut-in point of the vehicle into the channel topology graph is as follows:

[0087] S200: Traverse all channel path nodes node on the channel topology graph.

[0088] S201: Determine whether the channel topology graph is bidirectional.

[0089] If not, the calculation process is directly ended. It should be noted that the horizontal direction candidate parking space is generally on the right side of the driving direction, and at this time, the channel topology graph is unidirectional.

[0090] S202: If YES, calculate the distance dist_to_head from the starting point of the vehicle to the head node of the node, and calculate the heading difference diff_theta_to_head between the starting point of the vehicle and the head node of the node.

[0091] S203: Calculate the generation value head_cost of the starting point of the vehicle to the head. Specifically, the head_cost = dist_to_head + 7.5 x diff_theta_to_head.

[0092] S204: Determine whether the start point of the vehicle can be directly connected to the head of the node.

[0093] Specifically, in the present embodiment, if the determination result of step S204 is no, the result of head_cost minus a second penalty value is calculated, and the calculated result is taken as a new head_cost, and steps S205-S208 are repeated. The second penalty value is also an experience value, which can be adjusted according to the actual demand and the adjustment experience to achieve global optimization. In the present embodiment, the second penalty value is 100, i.e., new head_cost = original head_cost - 100. Obviously, after the original head_cost is subtracted by the second penalty value, the value becomes smaller, which can ensure that the node directly walking to the head of the node (the start point) is preferentially selected in subsequent selection.

[0094] S205: If no, calculate the distance dist_to_tail from the start point of the vehicle to the tail node of the node, and calculate the heading difference diff_theta_to_tail from the start point of the vehicle to the tail node of the node.

[0095] S206: Calculate the cost value tail_cost from the start point of the vehicle to the tail. Specifically, tail_cost = dist_to_tail + 7.5 x diff_theta_to_tail.

[0096] S207: Determine whether the start point of the vehicle can be directly connected to the tail of the node.

[0097] Specifically, in the present embodiment, if the determination result of step S207 is no, the result of tail_cost minus a second penalty value is calculated, and the calculated result is taken as a new tail_cost, and step S208 is repeated. Specifically, new tail_cost = original tail_cost - 100, and for the same reason, after the second penalty value is subtracted, the node directly walking to the tail of the node is preferentially selected.

[0098] S208: If no, select the target point with the smallest cost value tail_cost, and take the target point as the optimal cut-in point.

[0099] After selecting the target point with the minimum generation value tail_cost as the optimal cut-in point, it can be further determined whether the main direction is from the head of the channel path node to the tail or from the tail to the tail. Specifically, the vehicle driving direction is defined as the main direction, there are generally multiple groups of side-by-side subsequent parking spaces, according to the current position of the vehicle, the optimal cut-in point is selected, generally the selected one is the group of parallel candidate parking spaces (after clustering and merging) corresponding to the position closest to the vehicle, in this embodiment, the left end of the selected group of parallel candidate parking spaces is defined as the head and the right end is defined as the tail. Generally, the optimal cut-in point is at the leftmost or rightmost of the selected group of parallel candidate parking spaces.

[0100] Obviously, after steps S201-S208, it can be determined that the optimal cut-in point from the starting point of the vehicle to the channel topology graph, in the process, each group of parallel candidate parking spaces formed after traversing and sorting in the channel topology graph are compared, and finally the target point with the minimum generation value tail_cost is selected as the optimal cut-in point, which effectively ensures that the cut-in point from the starting point of the vehicle on the channel topology graph is the closest, the path to the optimal cut-in point in the later navigation is smooth, effectively saves the parking time of the vehicle, solves the problem of low parking flow in the prior art, and effectively improves the low parking efficiency.

[0101] S3: calculating the path from the starting point of the vehicle to the optimal cut-in point of the channel topology graph.

[0102] It should be noted that the starting point of the vehicle is generally the current position coordinate of the vehicle.

[0103] The navigation planning algorithm is used to calculate the path from the starting point of the vehicle to the optimal cut-in point of the channel topology graph. Specifically, it is a point-to-point navigation planning algorithm, for example, the point-to-point navigation planning algorithm disclosed in the prior art mentioned in the background art (using bidirectional Hybrid A* algorithm to alternately search from the starting point and the end point, until the two points meet), by using the coordinates of the starting point of the vehicle and the optimal cut-in point (equivalent to the end point), using Hybrid A* algorithm, Egoplanning, TEB, etc. Planning algorithm, a shortest path from the starting point to the optimal cut-in point is planned, which can significantly improve the traffic efficiency and safety of the path, reduce the driving distance and time, and reduce the energy consumption and cost.

[0104] S4: according to the topological relationship of the channel topology graph, starting from the channel path node start_node_id corresponding to the optimal cut-in point, finding the next channel path node to_node_id, traversing all channel path nodes, and generating a planning path above all candidate parking spaces.

[0105] Specifically, after planning the planning path above all the candidate parking spaces, the vehicle can select the optimal candidate parking space (the shortest driving distance, the most convenient parking, etc.) according to the specific needs, and then the vehicle can be parked, and the later parking process can be automatically completed by the vehicle through the mixed A* algorithm or other algorithms.

[0106] Specifically, in this embodiment, when the unmanned cleaning vehicle encounters a full garbage bin during the execution of the cleaning task and needs to find an idle garbage bin to dump garbage, the method in the present application is used to directly search for a garbage parking space and plan a path to enter the garbage dumping parking space, so as to realize rapid garbage dumping; when the vehicle needs to find an idle charging space to charge due to insufficient power, the method in the present application is used to directly search for a charging space and plan a path to enter the charging space in the charging station area, so as to realize timely charging of the vehicle; when the vehicle needs to find an idle water refilling station to refill water due to insufficient water in the water tank during wet cleaning in dry weather, the method in the present application is used to directly search for a water refilling parking space and plan a path to enter the water refilling parking space in the water refilling area, so as to realize timely water refilling of the vehicle; when the vehicle does not want to stay at the cleaning end point after completing the cleaning task and needs to find an idle parking space to park nearby, the method in the present application is used to directly search for a parking space and plan a path to enter the nearest parking lot to park, so as to realize rapid parking of the vehicle; for details, please refer to Figure 4 Veicle is the current position of the vehicle, and the predicted parking lot has a water refilling parking space (Water), a charging parking space (Charge), and a parking parking space (Park); therefore, the path planning method for automatically searching for a parking space of the unmanned cleaning vehicle in the present application can be applied to the search for a parking space in a special scenario and the planning of a path during parking, so as to realize rapid parking of the vehicle and meet the needs of charging, water refilling, garbage dumping, etc. In addition, the path planning method for automatically searching for a parking space of the unmanned cleaning vehicle in the present application can also be applied to the search for a parking space in an underground parking lot, a shopping mall, a park, etc. according to the use needs, so as to meet different needs.

[0107] It should be understood that the application of the present application is not limited to the above examples, and those skilled in the art can make improvements or changes according to the above description, and all these improvements and changes shall belong to the protection scope of the appended claims of the present application.

Claims

1. A path planning method for an unmanned road sweeper to automatically search for parking spaces, characterized in that: The method comprises: Generate a channel topology map of candidate parking spaces; Calculating the optimal entry point for the vehicle to enter the channel topology map; Calculating a path from the starting point of the vehicle to the optimal entry point of the channel topology graph; According to the topological relationship of the channel topology graph, the next channel path node to_node_id is searched starting from the channel path node start_node_id corresponding to the optimal entry point, all channel path nodes are traversed, and a planned path passing through all candidate parking spaces is generated.

2. A path planning method for an unmanned road sweeper to automatically search for parking spaces according to claim 1, characterized in that: The steps of generating a channel topology map of candidate parking spaces include: Check whether the parking space requirements are met above each vacant parking space; The vacant parking spaces that meet the parking space requirements are regarded as candidate parking spaces, and the parking spaces are clustered and merged; According to the results of parking space clustering and merging, the candidate parking spaces that meet the requirements are traversed and sorted; Calculate and generate channel path nodes based on the results of parking space traversal and sorting; The connection edges between channels are calculated based on the generated channel path nodes, and a channel topology graph is generated.

3. A path planning method for an unmanned road sweeper to automatically search for parking spaces according to claim 2, characterized in that: The process of calculating the connection edges between channels based on the generated channel path nodes is as follows: S100: randomly selecting a channel path node from the set of channel path nodes as the starting channel path node from_node; S101: Determine whether the beginning and end of all channel path nodes are connected; S102: If not, traverse all channel path nodes other_node except the starting channel path node from_node, and calculate the distance dist and the heading difference diff_theta between the pre-parking point s_point at the tail of the from_node and the pre-parking point e_point at the head of the other_node; S103: Calculate the cost value and determine whether the tail of other_node is connected; S104: If not, select other_node with the smallest cost value as the next channel path node to_node; S105: Calculate and generate a planned connection path from from_node to to_node; S106: Generate connection edge connection_edge; After completing step S106, the process returns to step S101 and repeats steps S101 to S106.

4. A path planning method for an unmanned road sweeper to automatically search for parking spaces according to claim 3, characterized in that: If the tail of other_node is connected, the first penalty value is added to the cost value and used as the new cost value. The other_node with the smallest cost value in the new cost value is selected as the to_node of the next channel path node, and steps S105~S106 are repeated.

5. A path planning method for an unmanned road sweeper to automatically search for parking spaces according to claim 3, characterized in that: In step S105 , the planned connection path from from_node to to_node is calculated using the RS curve and / or the quintic polynomial.

6. A path planning method for an unmanned road sweeper to automatically search for parking spaces according to claim 2, characterized in that: The basis for clustering and merging candidate parking spaces is: If the distance between the pre-parking points above the candidate parking spaces is less than the parking spacing threshold, the heading difference between the pre-parking points is less than the parking heading threshold, and there is no collision in the straight line connection, they are clustered and merged into the same parking space set.

7. A path planning method for an unmanned road sweeper to automatically search for parking spaces according to claim 3, characterized in that: The steps for calculating the optimal entry point for a vehicle to enter the channel topology are: S200: traverse all channel path nodes on the channel topology graph; S201: Determine whether the channel topology is bidirectional; S202: If yes, calculate the distance dist_to_head from the starting point of the vehicle to the head node of node, and calculate the heading difference diff_theta_to_head between the starting point of the vehicle and the head node of node; S203: Calculate the cost value head_cost from the starting point to the head of the vehicle; S204: Determine whether the starting point of the vehicle and the head of the node can be directly connected; S205: If not, calculate the distance dist_to_tail from the starting point of the vehicle to the tail node of node, and calculate the heading difference diff_theta_to_tail between the starting point of the vehicle and the tail node of node; S206: Calculate the cost value tail_cost from the starting point to the tail of the vehicle; S207: Determine whether the starting point of the vehicle and the tail of the node can be directly connected; S208: If not, select the target point with the minimum cost value tail_cost and use the target point as the optimal entry point.

8. A path planning method for an unmanned road sweeper to automatically search for parking spaces according to claim 7, characterized in that: If the judgment result of step S204 is no, the result of subtracting the second penalty value from the head_cost is calculated, and the calculated result is used as the new head_cost, and steps S205 to S208 are repeated.

9. A path planning method for an unmanned road sweeper to automatically search for parking spaces according to claim 7, characterized in that: If the judgment result of step S207 is no, the result of subtracting the second penalty value from the tail_cost is calculated, and the calculated result is used as the new tail_cost, and step S208 is repeated.

10. The path planning method for an unmanned road sweeper to automatically search for parking spaces according to claim 7, wherein a navigation planning algorithm is used to calculate a path from the starting point of the vehicle to the optimal entry point of the channel topology map.

Images