Indoor navigation system and path planning method for intelligent car

Abstract

The application provides a path planning method for indoor navigation of intelligent trolleys, and the method comprises the following steps: continuously taking pictures of a production site in real time by using a UAV, splicing the pictures by using a server, converting the spliced pictures into a two-dimensional grid map, and sending the two-dimensional grid map to each intelligent trolley in real time; the intelligent trolley calculates a first navigation route by using an improved ant colony algorithm, and if there is an obstacle in front of the intelligent trolley, the intelligent trolley calculates a second navigation route or a third navigation route to the nearest temporary parking position, then waits at the temporary parking position and calculates a fourth navigation route. The application also provides an indoor navigation system for intelligent trolleys, which comprises a UAV, a UAV charging platform, an intelligent trolley, a transportation channel and a fixed parking position. The application adopts the UAV to take pictures in a cycle, so that the generated map has fewer dead points and the probability of collision is reduced, and the intelligent trolley, a network, a server and the UAV form a complete system, so that multiple intelligent trolleys can transport goods at the same time.

Description

Technical Field

[0001] This invention relates to the field of intelligent vehicle navigation path planning technology, specifically to an intelligent vehicle indoor navigation system and path planning method. Background Technology

[0002] Material transportation of raw materials, semi-finished products, and finished products is a crucial part of factory production. Traditional transportation methods rely on conveyor belts or manual labor, while modern smart factories typically use intelligent vehicles. However, current intelligent transportation systems have several drawbacks: First, some rely solely on their own sensors for navigation and obstacle avoidance, resulting in limited detection capabilities, susceptibility to local optima, and an inability to generate globally optimal paths. Second, some systems use fixed cameras within the factory area to generate navigation maps for obstacle avoidance and navigation. However, these fixed cameras have blind spots due to their relatively fixed angles and positions, particularly at turning points, which can easily lead to collisions. Third, current path planning for intelligent vehicles is often geared towards individual vehicles, with limited application to multiple vehicles. Therefore, it is necessary to design an indoor navigation system and path planning method suitable for multiple intelligent vehicles to address these issues. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide an indoor navigation system and path planning method for intelligent vehicles, which can be used for multiple intelligent vehicles to simultaneously perform navigation path planning and obstacle avoidance in a production site setting.

[0004] To address the aforementioned technical problems, this invention provides a path planning method for indoor navigation of intelligent vehicles, comprising the following steps:

[0005] S1. Real-time map construction and processing: The drone takes real-time continuous photos of the production site and sends them to the server to stitch them into a complete indoor map, including the equipment, transportation channels, temporary parking spaces, fixed parking spaces and various intelligent vehicles on the site; then the indoor map is converted into a two-dimensional grid map and sent to each intelligent vehicle in real time.

[0006] S2, Route Planning and Navigation:

[0007] S2.1 After the intelligent vehicle receives the transportation instruction and the two-dimensional grid map, it uses the improved ant colony algorithm to calculate the navigation route between the current position and the target position. When there are multiple navigation routes, the route with the shortest distance is selected as the first navigation route. If the intelligent vehicle's lidar does not detect any obstacles ahead, the intelligent vehicle will continue to follow the first navigation route to reach the target position.

[0008] S2.2 If the LiDAR of the intelligent vehicle detects an obstacle ahead, it stops moving forward and, based on the latest two-dimensional mesh map sent by the server, generates a second navigation route to the target location using the improved ant colony algorithm, and then determines:

[0009] L2-L1>5L, where L2 is the length of the second navigation route, L1 is the remaining length of the first navigation route, and L is the maximum length of the intelligent vehicle. If not, the intelligent vehicle continues to move forward according to the second navigation route; if true, the intelligent vehicle uses the improved ant colony algorithm to generate a third navigation route to the nearest temporary stop and moves to the nearest temporary stop to wait.

[0010] S2.3 The intelligent vehicle located at the temporary parking position generates a fourth navigation route to the target location based on the latest two-dimensional grid map sent by the server and using the improved ant colony algorithm. The length of the fourth navigation route is L4. When L4-L1≤5L, the intelligent vehicle continues to move forward according to the fourth navigation route; otherwise, it continues to wait and generates a new fourth navigation route based on the updated two-dimensional grid map.

[0011] S2.4 Repeat steps 2.2 and 2.3 until the smart car reaches the target position.

[0012] As an improvement to the path planning method for indoor navigation of the intelligent vehicle of the present invention:

[0013] The two-dimensional grid map includes feasibility grids, no-entry grids, and grids representing the intelligent vehicle. The specific process for obtaining the grid map is as follows:

[0014] S1.1 The drone continuously takes photos along the transport channel according to a preset fixed route and transmits the photos to the server in real time.

[0015] S1.2 The server stitches the photos into a real-time map using a region registration method, and then obtains a two-dimensional grid map by modeling using a raster map method; then the grids in the two-dimensional grid map are mapped to coordinates using a numbering method.

[0016] S1.3 The smart car located in the fixed parking space takes a picture of the QR code on the fixed parking space with the camera under the car, identifies the coordinate information, and sends it to the server as the current location information of the smart car.

[0017] The current location information of the intelligent vehicles in the transportation channels and temporary parking spaces is included in the two-dimensional grid map;

[0018] S1.4 When the total number of intelligent vehicles on a transport channel and its adjacent temporary parking spaces exceeds the number of vehicles that can be parked in that temporary parking space, the transport channel is marked as a no-entry grid in the two-dimensional grid diagram.

[0019] As a further improvement to the path planning method for indoor navigation of the intelligent vehicle of the present invention:

[0020] The improved ant colony algorithm includes an improved state transition formula:

[0021]

[0022] in,

[0023]

[0024] τ ij (t) represents the pheromone value at time t between grid i and grid j, d ij (t) represents the Euclidean distance between grid i and grid j. Let A be the heuristic function, α be the pheromone heuristic factor, β be the expected heuristic factor, and χ be the target point distance weighting factor. k It is the set of the next grid cells that grid i can choose;

[0025] The global pheromone update formula is:

[0026]

[0027] in,

[0028] (1-ρ)τ ij (t) represents the pheromone evaporation term. Add an item for pheromones. As a pheromone reward and punishment item, L best For the shortest path length in this iteration, L worst For this iteration, the longest path length is given, ρ is the adaptive pheromone evaporation coefficient, κ is the adjustment coefficient, and Q is the pheromone intensity.

[0029] The present invention also provides an indoor navigation system for an intelligent vehicle for implementing a path planning method for indoor navigation of an intelligent vehicle, including a drone and a drone charging suction platform installed under the roof of the production site, and an intelligent vehicle, a transport channel and a fixed parking space installed on the ground of the production site, with a temporary parking space provided on the side of each transport channel; and a safety net installed between the drone charging suction platform and the ground.

[0030] Both the drones and the smart cars are connected to the server via wireless network.

[0031] As an improvement to the intelligent vehicle indoor navigation system of the present invention:

[0032] The drone charging platform includes a pair of charging bases, each of which includes a current detection device, a charging interface, and a photoelectric sensor.

[0033] The drone is attracted to the charging base (2) via an electromagnetic actuator module.

[0034] As a further improvement to the intelligent vehicle indoor navigation system of the present invention:

[0035] The width of the transport channel is at least 2.5 times the maximum width of the intelligent vehicle.

[0036] As a further improvement to the intelligent vehicle indoor navigation system of the present invention:

[0037] Each of the designated parking spaces has a QR code with location information affixed to the ground.

[0038] The beneficial effects of this invention are mainly reflected in:

[0039] 1. In this invention, when calculating navigation routes using an improved ant colony algorithm, the navigation path is calculated only at the beginning, and then the path is calculated again only when the current navigation smart car detects that there are other smart cars ahead. This reduces the computational load of the on-board control module, improves the transportation efficiency of the smart car, and reduces costs.

[0040] 2. This invention uses drones to take cyclical photos of the entire factory area. The movement of the drones reduces the number of dead spots in the generated map, greatly reducing the probability of collisions with the intelligent vehicle and further ensuring that the intelligent vehicle can reach the target location smoothly. At the same time, the calculated two-dimensional grid map only needs to be constructed for the transportation channels and temporary parking spaces, which reduces the server load and improves the processing speed of the two-dimensional grid map.

[0041] 3. In order to further improve the path planning calculation speed, this invention improves the state transition probability and global pheromone update formula of the ant colony algorithm, making the algorithm converge faster.

[0042] 4. This invention uses intelligent vehicles, networks, servers, and drones to form a complete system, ensuring that multiple intelligent vehicles can transport goods simultaneously and efficiently. Attached Figure Description

[0043] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0044] Figure 1 This is a schematic diagram of the layout of the intelligent vehicle indoor navigation system of the present invention in the factory area;

[0045] Figure 2 for Figure 1 A schematic diagram of the charging dock;

[0046] Figure 3 This is a schematic diagram of the structure of the intelligent vehicle indoor navigation system of the present invention;

[0047] Figure 4 This is a block diagram of the intelligent vehicle of the present invention;

[0048] Figure 5 This is a block diagram of the UAV of the present invention;

[0049] Figure 6 This is a schematic diagram of a two-dimensional grid map of the first navigation route generated by the intelligent vehicle of the present invention when the route is unobstructed;

[0050] Figure 7 This is a schematic diagram of a two-dimensional grid representing the third navigation route of the intelligent vehicle of the present invention when it encounters an obstacle and navigates to a temporary parking spot;

[0051] Figure 8 This is a schematic diagram of a two-dimensional grid of the fourth navigation route of the intelligent vehicle of the present invention from the temporary parking spot to the target location;

[0052] Figure 9 This is a flowchart illustrating the path planning method for indoor navigation of an intelligent vehicle according to the present invention.

[0053] Figure 10 This is a flowchart illustrating the path planning process of the intelligent vehicle according to the present invention using the ant colony algorithm. Detailed Implementation

[0054] The present invention will be further described below with reference to specific embodiments, but the scope of protection of the present invention is not limited thereto:

[0055] Example 1: An indoor navigation system and path planning method for intelligent vehicles, including using drones to take real-time photos of the factory site, and then using a server to stitch together the photos of the indoor areas of the factory obtained by each drone into a complete real-time indoor factory map, thereby obtaining real-time information such as the location of equipment, transport channels, obstacles, and the location of each intelligent vehicle on the site. Each intelligent vehicle calculates the navigation route based on the real-time updated indoor factory map and completes the movement. If an obstacle (including objects on the transport channel or other intelligent vehicles) is encountered during the movement, the navigation route is recalculated based on the real-time updated indoor factory map.

[0056] The intelligent vehicle indoor navigation system includes multiple intelligent vehicles, a server, two drones, and a drone charging station, such as... Figure 1-5As shown, the factory ground 5 typically contains obstacles, primarily factory production equipment or stacked raw materials or semi-finished products. At least four fixed transport channels 6 are established between these obstacles as pathways for the intelligent vehicles. These transport channels 6 must be able to cover all production equipment to facilitate the transport of raw materials, semi-finished products, or finished products during the production process. A set of temporary parking spaces 7 (i.e., temporary parking spaces 7 cannot occupy the space of the transport channel 6) is located on the side of each transport channel 6 in the middle, allowing multiple intelligent vehicles to avoid each other when on the same transport channel 6. The width of the transport channel 6 is 2.5 times the maximum width of the intelligent vehicle, ensuring that two intelligent vehicles can move simultaneously and improving operational efficiency. Fixed parking spaces 8 equipped with charging devices are also provided on the factory ground 5 as the initial positions for the intelligent vehicles. These fixed parking spaces 8 are adjacent to the transport channels 6. When the factory is not in operation or needs charging, the intelligent vehicles return to the fixed parking spaces 8 to park and charge. Each fixed parking space 8 has a QR code with location information affixed to the ground.

[0057] The drone charging station is fixedly connected to the roof of the factory building, located beneath it. One charging station is sufficient for all photography needs within the factory area; installing too many would increase the installation and maintenance burden. The drone charging station includes a pair of charging bases 2, allowing simultaneous charging of two drones or charging of a single drone. Each drone is fixedly connected to one charging base 2. Each charging base 2 includes a current detection device, a charging interface 201, and a photoelectric sensor 202. Figure 2 As shown, the charging interface 201 is used to connect with the drone and provide charging. The current detection device is used to detect the current at the charging interface 201 to determine the drone's charging status. The photoelectric sensor 202 is used to determine whether there is a drone on the charging base 2. If there is a strong current fluctuation at the charging interface 201 of one of the charging bases 2 of a drone charging platform, and the photoelectric sensor 202 detects a drone, it means that a drone is charging on this charging base 2. If there is no current fluctuation or a very weak current fluctuation at the charging interface 201 of one of the charging bases 2 of a drone charging platform, and the photoelectric sensor 202 detects a drone, it means that the drone on this charging base 2 has finished charging and can perform the photo-taking task. If the photoelectric sensor 202 of one of the charging bases 2 of a drone charging platform does not detect a drone, it means that the drone is performing the photo-taking task. By setting one drone charging platform to charge two drones in turn or simultaneously, it is ensured that at least one drone is always in the air taking pictures, thus ensuring that the server can obtain the latest situation of the ground transportation channel 6 in real time and send it to each intelligent vehicle.

[0058] The drone includes a drone network module, a data acquisition camera, an electromagnetic actuator module, an ultrasonic sensing module, a motor, a drone control module, and a rechargeable battery module, such as... Figure 5 As shown, the drone attaches to the underside of the charging base 2 via an electromagnetic actuator module, charges its rechargeable battery module using the charging interface 201, and connects to the server via a network module using Wi-Fi, 5G, or other wireless network technologies to exchange data and control commands. An ultrasonic sensing module is used to detect and determine the distance to the charging base 2. Each drone takes photos of the ground below along a fixed path, and the server stitches these photos together to create a complete indoor factory map. During the photo-taking process, if the drone control module detects that the rechargeable battery module's power level is below a preset value, it controls the drone to return to the charging base 2 for charging. The preset power level could be, for example, 10% of the total battery capacity, primarily to ensure the drone can take one last photo before returning to the charging base 2. Two drones and one drone charging platform are typically set up for each indoor factory area. If the factory area is large, the drone's range can be increased.

[0059] A safety net 4 is installed below the drone charging platform. When the drone malfunctions, it will fall onto the safety net 4 to prevent it from hitting the equipment or people below.

[0060] The intelligent vehicle includes a network module, an energy module, a power module, a drive and steering module, an under-vehicle camera, an audio module, indicator lights, a display screen, a lidar, and a control module. The network module sends and receives messages to the server via Wi-Fi, 5G, or other wireless networks. The energy module provides power to the intelligent vehicle. The power and drive / steering modules ensure the vehicle's forward movement and steering. The under-vehicle camera photographs the QR codes on fixed parking spaces to determine coordinates. The audio module provides voice prompts to alert factory workers and ensure their safety. Indicator lights display the vehicle's operating status: green indicates normal operation, red indicates a malfunction, and yellow indicates temporary parking. The display screen shows operational information. The lidar detects other intelligent vehicles ahead. Both drones and intelligent vehicles are existing technologies; for example, drones can be purchased or customized from Shenzhen DJI Innovations Technology Co., Ltd., and intelligent vehicles can be purchased or customized from Shenzhen Yabo Intelligent Technology Co., Ltd.

[0061] The method for path planning in indoor navigation of an intelligent vehicle using the intelligent vehicle indoor navigation system of the present invention, such as... Figure 9 As shown, specifically:

[0062] 1. Real-time map construction and processing

[0063] 1.1 The drone continuously takes photos along the predetermined fixed route of the ground transport channel 6, and transmits the photos to the server in real time through the drone network module. The photos include the production equipment, transport channel 6, temporary parking spaces 7, fixed parking spaces 8, and various intelligent vehicles on the factory ground. The position of each drone hovering for photos is at a fixed height. For example, when the drone performs a photo-taking task, it is set to fly downwards for 2 seconds after leaving the charging base 2, and then take N photos along the fixed path (i.e., each transport channel 6), and then send the N pictures to the server for processing. The distance between the drone and the safety net 4 when taking photos is 0.5-0.8m, which can be ensured by the distance detection of the charging base 2 by the drone's ultrasonic module.

[0064] 1.2 The server stitches the received photos into a complete real-time factory map using traditional region registration methods. Then, it performs environmental map modeling and conversion using a raster map method to obtain a two-dimensional grid map. White grids in the 2D grid map represent feasibility grids (i.e., transport lane 6, temporary parking spaces 7, and fixed parking spaces 8), gray grids represent restricted areas for the intelligent vehicles (e.g., obstacles, including production equipment or raw materials on the ground, and temporarily impassable transport lane 6), and black grids represent intelligent vehicles moving in transport lane 6 or stopped in fixed parking spaces 8. Since the positions of production equipment on the factory ground 5 are generally fixed, the server only needs to rebuild transport lane 6, temporary parking spaces 7, and fixed parking spaces 8 each time it reconstructs the 2D grid map, reducing server load and improving the processing speed of the 2D grid map.

[0065] Coordinate mapping is performed on the grid cells in the 2D mesh diagram to convert the 2D mesh diagram into coordinate information that the intelligent vehicle can recognize. Generally, a numbering method is used for coordinate mapping. All grid cells in the factory area are numbered from left to right and from top to bottom, starting with number 1. The mapping relationship between the number and the coordinate system is as follows:

[0066]

[0067]

[0068] Where g is the number of each grid cell, n is the number of columns in the grid cell, m is the number of rows in the grid cell, δ is the side length of each grid cell, mod is the modulo operation, ceil is the floor operation, and δ is usually set to 1.

[0069] 1.3 The smart car located in the fixed parking space 8 takes a picture of the QR code on the fixed parking space 8 using the camera under the car, identifies the coordinate information, sends it to the server as the current location information of the smart car, and waits to receive transportation instructions;

[0070] The current location information of the smart cars on transport lane 6 and temporary parking space 7 is included in the two-dimensional grid map that is updated and sent by the server in real time.

[0071] 1.4 The number of temporary parking spaces 7 on each transport channel 6 can be A, the number of transport channels 6 is B, and the maximum number of intelligent vehicles that the entire factory area can accommodate is A*B. When the total number of intelligent vehicles running on a transport channel 6 and parked on the temporary parking spaces 7 on the side of this transport channel 6 is greater than A, the transport channel 6 is processed into a gray grid in the two-dimensional grid diagram, and this transport channel 6 is temporarily blocked from passage to prevent other intelligent vehicles from navigating to this transport channel 6.

[0072] 1.5 The server sends the constructed real-time two-dimensional mesh map containing the coordinate information of all smart cars to all smart cars; steps 1.1-1.5 are repeated continuously so that each smart car can obtain a real-time updated two-dimensional mesh map.

[0073] 2. Route planning and navigation

[0074] 2.1 After receiving the transportation instructions (including target location information) and a two-dimensional grid map, the intelligent vehicle calculates the navigation route between its current location and the target location using an improved ant colony algorithm; the ant colony algorithm process is an existing technology, such as... Figure 10 As shown, the improved ant colony algorithm of this invention improves the state transition formula and the global pheromone update formula. Specifically, the state transition formula for the intelligent vehicle selecting the target position grid j from the current grid i is:

[0075]

[0076] in,

[0077]

[0078] τ ij (t) represents the pheromone value at time t between grid i and grid j, d ij (t) represents the Euclidean distance between grid i and grid j. Let be the heuristic function, α be the pheromone heuristic factor, β be the expected heuristic factor, χ be the target point distance weight factor, and Ak be the set of next grid cells that grid i can choose.

[0079] After all ants have completed this iterative cycle, the global pheromone update formula is:

[0080]

[0081] in,

[0082] (1-ρ)τ ij (t) represents the pheromone evaporation term. Add an item for pheromones. As a pheromone reward and punishment item, L best For the shortest path length in this iteration, L worst For this iteration, the longest path length is given, ρ is the adaptive pheromone evaporation coefficient, κ is the adjustment coefficient, and Q is the pheromone intensity.

[0083] There may be multiple navigation routes. The shortest route is selected as the first navigation route. The vehicle moves along the first navigation route. During the movement, the intelligent vehicle no longer performs path planning calculations, but it still continuously receives real-time two-dimensional grid maps sent by the server. The LiDAR performs real-time obstacle detection. If the intelligent vehicle's LiDAR does not detect any other intelligent vehicles blocking the way, the intelligent vehicle will continue to follow the first navigation route until it reaches the target location.

[0084] 2.2 If the LiDAR of the intelligent vehicle detects another intelligent vehicle (obstacle) blocking its path, the navigation intelligent vehicle with a later departure task allocation time retrieves the latest 2D mesh map sent by the server and uses the improved ant colony algorithm to generate a second navigation route to the target location. The length of the second navigation route is calculated as L2, and the remaining length of the untraveled first navigation route is calculated as L1. The maximum length of the intelligent vehicle is L. If L2-L1>5L (not true), it indicates that the second navigation route is a better route, and the intelligent vehicle continues to move forward according to the second navigation route. During the journey, the LiDAR continues to detect obstacles in real time. If the LiDAR of the intelligent vehicle does not detect any other intelligent vehicles blocking its path, the intelligent vehicle will continue to move forward according to the second navigation route to reach the target location. The navigation intelligent vehicle with an earlier departure task allocation time continues to move forward according to the original navigation route.

[0085] When L2-L1>5L, it means that the current intelligent vehicle has selected a navigation route that is very far away and is very different from the first navigation route. At this time, the current intelligent vehicle uses the improved ant colony algorithm to generate a third navigation route to the nearest empty temporary parking space 7, and navigates to the nearest empty temporary parking space 7 to wait.

[0086] 2.3 The intelligent vehicle located at temporary stop 7 continuously receives real-time 2D grid maps from the server and uses the improved ant colony algorithm to generate a fourth navigation route to the target location. The length of the fourth navigation route is L4. If L4-L1≤5L, the intelligent vehicle at temporary stop 7 continues to move according to the fourth navigation route; otherwise, it waits at temporary stop 7 for the 2D grid map to be generated again for a new fourth navigation route. Figure 6-8 As shown;

[0087] 2.4 Repeat steps 2.2 and 2.3 until the smart car reaches the target position.

[0088] 2.5 When the factory is not in operation, i.e. after get off work hours, all the smart cars return to their designated parking spaces 8 to standby and charge.

[0089] Finally, it should be noted that the above examples are merely some specific embodiments of the present invention. Obviously, the present invention is not limited to the above embodiments and many variations are possible. All variations that can be directly derived or conceived by those skilled in the art from the disclosure of the present invention should be considered within the scope of protection of the present invention.

Claims

1. A path planning method for indoor navigation of an intelligent vehicle, characterized in that... Includes the following steps: S1. Construction and processing of real-time maps: The drone takes real-time continuous photos of the production site and sends them to the server to stitch them together into a complete indoor map, including the equipment, transportation channels (6), temporary parking spaces (7), fixed parking spaces (8) and various smart cars on the site; then the indoor map is converted into a two-dimensional grid map and sent to each smart car in real time. S2, Route planning and navigation, includes the following steps: S2.1 After the intelligent vehicle receives the transportation instruction and the two-dimensional grid map, it uses the improved ant colony algorithm to calculate the navigation route between the current position and the target position. When there are multiple navigation routes, the route with the shortest distance is selected as the first navigation route. If the intelligent vehicle's lidar does not detect any obstacles ahead, the intelligent vehicle will continue to follow the first navigation route to reach the target position. S2.2 If the LiDAR of the intelligent vehicle detects an obstacle ahead, it stops moving forward and, based on the latest two-dimensional mesh map sent by the server, generates a second navigation route to the target location using the improved ant colony algorithm, and then determines: ,in, The length of the second navigation route. L is the remaining length of the first navigation route, and L is the maximum length of the smart car. If it is not true, the smart car continues to move forward according to the second navigation route. If it is true, the smart car uses the improved ant colony algorithm to generate a third navigation route to the nearest temporary stop (7) and moves to the nearest temporary stop (7) to wait. S2.3 The intelligent vehicle located at the temporary parking spot (7) generates a fourth navigation route to the target location based on the latest two-dimensional grid map sent by the server and using an improved ant colony algorithm. The length of the fourth navigation route is... ,when The intelligent car continues to move forward according to the fourth navigation route; otherwise, it continues to wait and generates a new fourth navigation route according to the updated two-dimensional grid map. S2.4 Repeat steps 2.2 and 2.3 until the smart car reaches the target position.

2. The path planning method for indoor navigation of an intelligent vehicle according to claim 1, characterized in that: The two-dimensional grid map includes feasibility grids, no-entry grids, and grids representing the intelligent vehicle. The specific process for obtaining the grid map is as follows: S1.1 The drone takes photos continuously along the transport channel (6) according to the preset fixed route and transmits the photos to the server in real time. S1.2 The server stitches the photos into a real-time map using a region registration method, and then obtains a two-dimensional grid map by modeling using a raster map method; then the grids in the two-dimensional grid map are mapped to coordinates using a numbering method. S1.3 The smart car located in the fixed parking space (8) takes a picture of the QR code on the fixed parking space (8) through the under-vehicle camera, sets up coordinate information and sends it to the server as the current location information of the smart car; The current location information of the intelligent vehicles on the transport channel (6) and the temporary parking space (7) is included in the two-dimensional grid diagram; S1.4 When the total number of intelligent vehicles on a transport channel (6) and its adjacent temporary parking spaces (7) exceeds the number of vehicles that can be parked in the temporary parking spaces (7), the transport channel (6) is marked as a no-entry grid in the two-dimensional grid diagram.

3. The path planning method for indoor navigation of an intelligent vehicle according to claim 2, characterized in that: The improved ant colony algorithm includes an improved state transition formula: （2） in, , ， For time In the grid With grid The pheromone values ​​between them For grid With grid The Euclidean distance between them For heuristic functions, As a pheromone-inspired factor, As the expected heuristic factor, The distance to the target point is the weighting factor. It is a grid The set of next grid cells that can be selected; The global pheromone update formula is: (3) in, , , For pheromone volatiles, Add an item for pheromones. For pheromone rewards and penalties, This is the shortest path length for this iteration. This is the longest path length for this iteration. For adaptive pheromone evaporation coefficient, To adjust the coefficient, The intensity of the pheromone.

4. An indoor navigation system for an intelligent vehicle used to implement the path planning method for indoor navigation of an intelligent vehicle as described in any one of claims 1-3, characterized in that: The facility includes a drone and a drone charging suction platform installed under the roof of the production site. On the ground (5) of the production site, there are smart cars, transport channels (6) and fixed parking spaces (8). Temporary parking spaces (7) are provided on the side of each transport channel (6). A safety net (4) is installed between the drone charging suction platform and the ground. Both the drones and the smart cars are connected to the server via wireless network.

5. The intelligent vehicle indoor navigation system according to claim 4, characterized in that: The drone charging platform includes a pair of charging bases (2), each charging base (2) including a current detection device, a charging interface (201) and a photoelectric sensor (202); The drone is attracted to the charging base (2) via an electromagnetic actuator module.

6. The intelligent vehicle indoor navigation system according to claim 5, characterized in that: The width of the transport channel (6) is at least 2.5 times the maximum width of the intelligent vehicle.

7. The intelligent vehicle indoor navigation system according to claim 6, characterized in that: Each of the fixed parking spaces (8) has a QR code with location information affixed to the ground.

Images