JP7784979B2 - Mobile control device - Google Patents

Description

This disclosure relates to a control device for a moving object.

When an autonomously moving vehicle travels to a destination, it generates a route that will allow it to reach its destination from its current location while avoiding obstacles, based on a pre-stored environmental map used to determine whether it can pass through a specific space, and the positions of surrounding obstacles observed while moving, and then follows that route.

A common method for generating routes is to calculate an evaluation function (potential function) from the sum of an attractive function that monotonically decreases toward the destination and a repulsive function that monotonically increases toward an unmovable area or obstacle on the environmental map, and then generate a route based on that evaluation function. For example, there is a method for generating a route using Dijkstra's algorithm with the evaluation function as a weight, or a method for generating a route by calculating a route that minimizes the evaluation function using the steepest descent method, as proposed in Non-Patent Document 1 below. The latter method is also known as the artificial potential method. There is also a method that uses only a repulsive function as the evaluation function.

In artificial potential methods, there may be local solutions called local minima that do not reach the destination. Local minima occur when multiple repulsive functions overlap due to the close proximity of obstacles, or when it is difficult to solve using a monotonic attractive function, such as in an environment where a detour is required (such as a maze). One technique for suppressing the occurrence of local minima is to use a potential function with no local minima as the evaluation function. For example, Non-Patent Document 2 proposes a method that uses the Laplace equation, the solution of which has been proven to have no local minima, and Patent Document 1 below, for example, discloses a technique for generating a route using this method.

Furthermore, Patent Document 2 below discloses a technology that prevents multiple repulsive force functions from overlapping by increasing the gradient of the repulsive force function in a narrow space.

Japanese Patent Application Publication No. 8-22324 JP 2018-185633 A JP 2017-111473 A

Oussama Khatib, "Real-Time Obstacle Avoidance for Manipulators and Mobile Robots", Proceedings of the 1985 IEEE International Conference on Robotics & Automation, pp. 500-505, 1985. Keisuke Sato, "Robot Motion Planning Using Potential Fields Without Minima," Journal of the Robotics Society of Japan, 1993

The technology in Patent Document 1 uses a potential function based on the Laplace equation as an evaluation function, and generates a potential field that determines the guidance direction of a moving object by performing iterative calculations with the termination condition being that the change in the numerical solution is below a certain level. Because this takes time to generate the potential field, the technology is unable to respond in time to changes in the surrounding environment while the moving object is moving, which could result in the moving object colliding with an obstacle, making it difficult to use in dynamic environments. Non-Patent Document 2 uses the Gauss-Seidel method to solve the Laplace equation, taking advantage of the property that no local minima occur even if the termination condition is not met, and improves responsiveness by limiting the number of iterations. However, it is known that the computational cost increases exponentially as the range of the moving space increases.

This disclosure has been made to solve the above-mentioned problems, and aims to provide a mobile body control device that can prevent a mobile body from falling into a local minimum and is highly responsive to changes in the surrounding environment.

A mobile body control device according to the present disclosure is a mobile body control device that generates a potential field that determines a guidance direction for the mobile body on a plurality of nodes obtained by discretizing the space in which the mobile body moves, and includes an information transmission rate determination unit that determines an information transmission rate of the potential field for each node, a potential field update unit that updates the potential field in accordance with the information transmission rate based on observation information about the surroundings of the mobile body, and a command value calculation unit that calculates a command value for the mobile body based on a gradient vector of the potential field, wherein the information transmission rate of the potential field is the rate at which the observation information is transmitted to the potential field and the potential field is updated in the potential field update unit, and the information transmission rate determination unit determines the information transmission rate of each node in accordance with the position of the mobile body.

The mobile object control device according to the present disclosure can prevent a mobile object from falling into a local minimum and can control the mobile object to be highly responsive to changes in the surrounding environment, thereby making it possible to more reliably guide the mobile object to its destination.

1 is a block diagram showing an example of a configuration of a mobile object control device according to an embodiment; FIG. 1 is a schematic diagram illustrating an example of an environment in which a mobile object control device according to an embodiment operates. 1 is a block diagram showing an example of a system configuration of a mobile body equipped with a mobile body control device according to an embodiment. FIG. 2 is a schematic diagram illustrating an example of the configuration of a potential field storage unit according to the first embodiment. FIG. 10 is a block diagram showing an example of the configuration of a potential field update unit according to a second embodiment. FIG. 10 is a schematic diagram illustrating an example of a movement cost calculated by a movement cost calculation unit according to a second embodiment. FIG. 10 is a schematic diagram illustrating an example of a method for updating a potential value of each node in a potential value updating unit according to the second embodiment. FIG. 10 is a schematic diagram illustrating an example of an order in which the potential value of each node is updated by the potential value update unit according to the second embodiment. FIG. 10 is a block diagram showing an example of the configuration of a potential field update unit according to a modified example of the second embodiment. FIG. 11 is a schematic diagram illustrating an example of a method for calculating an estimated movement cost when updating a potential field in a modified example of the second embodiment. FIG. 11 is a schematic diagram illustrating an example of a process performed by an information transmission rate determination unit according to the third embodiment. FIG. 13 is a schematic diagram illustrating an example of a process performed by an information transmission rate determiner according to a first modified example of the third embodiment. FIG. 13 is a schematic diagram illustrating an example of processing by an information transmission rate determiner according to a second modification of the third embodiment. FIG. 10 is a block diagram showing an example of the configuration of a command value calculation unit according to a fourth embodiment. 13A and 13B are schematic diagrams illustrating an example of correction of a command value for a moving body in the fourth embodiment. FIG. 10 is a schematic diagram illustrating an example of processing by a command value initial value calculation unit according to the fourth embodiment. FIG. 2 is a diagram illustrating an example of a hardware configuration of a mobile object control device. FIG. 2 is a diagram illustrating an example of a hardware configuration of a mobile object control device.

An embodiment of a mobile object control device according to the present disclosure will be described below with reference to the accompanying drawings.

Figure 1 is a block diagram showing an example of the configuration of a mobile object control device S5 (hereinafter simply referred to as "control device S5") according to an embodiment. As shown in Figure 1, the control device S5 includes an initial potential field generation unit C1, a potential field storage unit C2, a potential field update unit C4, and an information transmission rate determination unit C5.

The initial potential field generation unit C1 generates an initial potential field that determines the guidance direction of the moving body from an environmental map. The potential field memory unit C2 stores the potential field generated by the initial potential field generation unit C1. The command value calculation unit C3 calculates a command value for the moving body from the potential field stored in the potential field memory unit C2. The potential field update unit C4 updates the potential field stored in the potential field memory unit C2. The information transmission rate determination unit C5 dynamically determines the information transmission rate of the potential field for each node. In other words, the information transmission rate determination unit C5 determines the information transmission rate of the potential field for each node.

Here, the information transmission speed refers to the speed at which environmental information is transmitted to the potential field in the potential field update unit C4 and the potential field is updated based on the environmental information. Furthermore, environmental information is information about the surroundings of the moving object, such as the location of obstacles, and is acquired by the observation device S2, which will be described later. The potential field stored in the potential field memory unit C2 can be defined in any coordinate system and any dimension, and may be a multi-layer or multi-story potential field.

In this example, the potential field memory unit C2 is incorporated into the control device S5, but it may also be provided outside the control device S5. By using differential equations whose solutions do not take local minima, such as the Laplace equation or the eikonal equation, as the potential functions used by the initial potential field generation unit C1 and the potential field update unit C4 when generating or updating the potential field, it is possible to guide the mobile object to its destination without it falling into a local minimum.

Figure 2 is a schematic diagram showing an example of an environment in which the control device S5 operates. The environment shown in Figure 2 includes a moving object E1 controlled by the control device S5, a destination E2 to which the moving object is heading, and a movement space E3 in which the moving object E1 operates, and has world coordinates E4 fixed to the movement space E3 and moving object coordinates E5 fixed to the moving object E1. The control device S5 operates with the purpose of guiding the moving object E1 to the destination E2.

Note that destination E2 may be given as multiple points or an area, and its position and shape may change over time. For example, if mobile object E1 is a cleaning robot and its task is to fill in the movement space E3, the task can be performed by giving the unexplored area as the destination E2 area and excluding the traversed area from the destination E2 area. Alternatively, destination E2 can be given as a point at certain intervals, making destination E2 the shape of a route, allowing mobile object E1 to follow the route.

There may be multiple mobile units E1 and multiple types of mobile units E1. The type of mobile unit E1 may be any type that is equipped with a drive mechanism capable of moving the mobile unit E1, such as a differential two-wheel mechanism, leg mechanism, or propeller mechanism. The moving space E3 may be defined in any dimension; for example, if the drive mechanism of the mobile unit E1 is a differential two-wheel mechanism, it may be a three-dimensional space defined by the x-axis and y-axis representing the position of the mobile unit and the θ-axis representing the posture.

Figure 3 is a block diagram showing an example of the system configuration of a mobile unit E1 equipped with a control device S5. As shown in Figure 3, the mobile unit E1 includes an environmental map distribution device S1, an observation device S2, a destination distribution device S3, a mobile unit position distribution device S4, a control device S5, and a drive device S6.

The environmental map distribution device S1 issues an environmental map. The observation device S2 observes the conditions around the mobile unit E1 and issues observation information that is the result of the observation. The destination distribution device S3 issues the destination E2 of the mobile unit E1. The mobile unit position distribution device S4 issues the position of the mobile unit E1. The control device S5 calculates command values for the mobile unit E1 using information issued by the environmental map distribution device S1, the observation device S2, the destination distribution device S3, and the mobile unit position distribution device S4. The drive device S6 controls the mobile unit E1 according to the command values calculated by the control device S5, thereby moving and guiding the mobile unit E1 to the destination E2.

The environmental map issued by the environmental map distribution device S1 may be created in advance using SLAM (Simultaneous Localization and Mapping), or it may be created online using information from the observation device S2. The observation device S2 does not necessarily have to be a device capable of acquiring distance, such as an RGB-D camera or LIDAR, but may also be composed of an RGB camera or an infrared thermography camera. Furthermore, the number of each of the environmental map distribution devices S1, observation devices S2, destination distribution devices S3, and mobile object position distribution devices S4 may be any number, and they do not have to be located on the mobile object E1; for example, they may be located in the mobile space E3 or outside it.

The specific processing flow of this embodiment will be explained using as an example a scenario in which a potential field held in a two-dimensional rectangular grid fixed to world coordinates E4 is updated based on the eikonal equation, and a differential two-wheel mobile object E1 is guided to a destination E2.

First, the environmental map distribution device S1 issues a two-dimensional occupancy grid map as an environmental map, which describes static information about the movement space E3 in which the moving body E1 can move, and the destination distribution device S3 issues the location of the moving body E1's destination E2.

When this information is input into the control device S5, the initial potential field generation unit C1 generates a potential field that monotonically decreases toward destination E2 while avoiding obstacles, based on the occupancy grid map and the position of destination E2, and the potential field storage unit C2 stores this as the initial potential field. Whether the information in the occupancy grid map indicates an obstacle can be determined, for example, by treating each node as an obstacle if its occupancy level is above a predetermined threshold.

Next, the command value calculation unit C3 calculates the target speed of the moving body E1, specifically the target values for the translational speed and rotational speed, from the potential field information stored in the potential field memory unit C2. The driving device S6 then uses speed tracking control to move the moving body E1 so that it tracks the target speed. Speed tracking control can use, for example, PID (Proportional-Integral-Differential) control.

Next, the information transmission rate determination unit C5 determines the information transmission rate for each node based on the information issued by the observation device S2 and the mobile unit position distribution device S4. The potential field update unit C4 dynamically adjusts the update period and discretization width for each node according to the information transmission rate, and updates the potential field stored in the potential field memory unit C2 based on the observation information issued by the observation device S2.

The control device S5 according to this embodiment can dynamically limit the information transmission speed of the potential field in the control loop of the moving body E1 for each node. This reduces the computational load on the control of the moving body E1, allowing emphasis to be placed on calculations in more important areas. As a result, the control cycle can be maintained at a constant level or below, ensuring real-time performance and enabling immediate response to dynamic changes in surrounding information.

The processing of the command value calculation unit C3 and the processing of the potential field update unit C4 are executed asynchronously, and the execution cycle of the command value calculation unit C3 is set to be sufficiently short relative to the movement speed of the moving body E1, for example, 100 ms. Furthermore, the eikonal equation can be solved efficiently by using the fast marching method for the processing of the initial potential field generation unit C1 and the potential field update unit C4. Here, the potential field is fixed to world coordinates E4, but it can also be fixed to another coordinate system, such as moving body coordinates E5 or the coordinate system that serves as the basis for the odometry of the moving body E1.

As a supplement, we will provide an overview of how to solve the eikonal equation using the Fast Marching Method. Equation (1) below is a two-dimensional eikonal equation.

In equation (1), U is the eikonal and n is the refractive index. The equation obtained by discretizing equation (1) with the grating width h is shown below in equation (2).

In formula (2), U _i,j (i=0, 1, 2, ...) (j=0, 1, 2, ...) is the potential value of the i-th and j-th nodes, U _H is U _i-1,j or U _i+1,j , and U _V is U _i,j-1 or U _i,j+1 .

The Fast Marching Method applies Dijkstra's algorithm to efficiently solve the equation (2) above. Specifically, destination E2 is constrained to U=0, and the potential value of each node is updated using the following equation (3) with U _H = min(U _i-1,j , U _i+1,j ) and U _V = min(U _i,j-1 , U _i,j+1 ). Note that min(a,b) is a function that returns the smaller of a and b.

The Fast Marching Method is known to enable efficient information transmission by updating the potential values of each node in ascending order using a sorting algorithm. However, in order to sort the potential values of each node, all nodes must be updated synchronously. Therefore, in this embodiment, this method can only be used in the processing of the initial potential field generation unit C1 and the processing of the potential field update unit C4 within areas with the same update period.

Example 1
Example 1 shows a specific example of the potential field storage unit C2 included in the control device S5 according to the embodiment. Fig. 4 is a schematic diagram showing an example of the configuration of the potential field storage unit C2 according to Example 1. The potential field storage unit C2 shown in Fig. 4 has a function of storing a potential field on nodes N1 obtained by discretizing the movement space E3, and asynchronously updating the potential value of each node by dividing it into, for example, high-speed information transmission N2, medium-speed information transmission N3, and low-speed information transmission N4 according to the information transmission speed determination unit C5.

Note that discretization can be performed using division using a rectangular grid, hexagonal grid, or Voronoi diagram. However, it is not necessary to discretize the movement space; the potential field can also be maintained using continuous functions such as NURBS surfaces or neural networks. The potential field storage unit C2 in this embodiment also has the function of calculating the potential value of any coordinate by interpolating from the potential values of surrounding nodes.

Example 2
Example 2 shows a specific example of a potential field update unit C4 provided in a control device S5 according to an embodiment. Fig. 5 is a block diagram showing an example of the configuration of the potential field update unit C4 according to Example 2. The potential field update unit C4 shown in Fig. 5 includes a movement cost calculation unit U1, an update period calculation unit U2, an update discretization width calculation unit U3, and a potential value update unit U4.

The movement cost calculation unit U1 calculates the movement cost, which indicates the difficulty of passing through each node, from information issued by the environmental map distribution device S1 and the observation device S2. The update period calculation unit U2 calculates the potential update period from the information transmission rate set by the information transmission rate determination unit C5. The update discretization width calculation unit U3 calculates the update discretization width from the information transmission rate. The potential value update unit U4 calculates and issues the potential value of each node from the movement cost of each node calculated by the movement cost calculation unit U1, the update period calculated by the update period calculation unit U2, and the update discretization width calculated by the update period calculation unit U2.

The information transmission speed of each node is determined by the update period and update discretization width, and the relationship shown in equation (4) below holds true.

In equation (4), v _I is the information transmission rate, ΔT _U is the update period, and Δe _U is the update discretization width. The update period calculation unit U2 and the update discretization width calculation unit U3 determine the update period and the update discretization width, respectively, based on the relational expression in equation (4). Since the relational expression in equation (4) is redundant, for example, in areas where the resolution cannot be made coarse, the update period can be determined by fixing the update discretization width. Furthermore, to eliminate redundancy, the potential field update unit C4 may be configured to include only either the update period calculation unit U2 or the update discretization width calculation unit U3.

Figure 6 is a schematic diagram showing an example of a movement cost calculated by the movement cost calculation unit U1. In the example of Figure 6, the maximum movement cost is assigned to the range from 0 distance from the obstacle to the radius of the mobile unit E1, and for ranges farther away, the movement cost is calculated using a function that exponentially decays according to the distance from the obstacle. By assigning the maximum movement cost to the range from 0 distance from the obstacle to the radius of the mobile unit E1, the mobile unit E1 can be treated as a mass point, thereby reducing calculation costs.

The movement cost calculated by the movement cost calculation unit U1 is applied as a constraint or weighting factor to the differential equation used in the potential value update unit U4. For example, if the differential equation is the two-dimensional eikonal equation, formula (1), the calculated movement cost can be used as the refractive index n to reflect this in the potential field.

Figure 7 is a schematic diagram showing an example of how the potential value update unit U4 updates the potential value of each node N1. As shown in Figure 7, the potential value update unit U4 updates the potential value of each node N1 using the potential interpolation values of the potential field memory unit C2 located at positions (±Δx, 0) and (0, ±Δy) relative to each node N1 on the x-axis and y-axis.

Figure 8 is a schematic diagram showing an example of the order in which the potential values of each node are updated by the potential value update unit U4. Within a specific region with the same update period, environmental information can be transmitted efficiently by determining the order in which the nodes update their potential fields using a search algorithm in each update loop. For example, within a region centered on moving body E1 as shown in Figure 8, surrounding environmental information can be transmitted efficiently to moving body E1.

The search algorithm can be Dijkstra's algorithm, as in the Fast Marching Method described above, but by using a breadth-first search, which has lower computational costs, transmission efficiency can be improved without increasing the amount of calculation. For example, if the road to destination E2 is blocked and a long detour is required, requiring a major change to the potential field, the former Dijkstra's algorithm can be used, and if the change in potential is small, the latter breadth-first search can be used, allowing for efficient information transmission.

When the node update order is determined using a search algorithm, the information transmission speed toward the center within a specific area will be equal to or greater than the value set by the information transmission speed determination unit C5. Specifically, when a search algorithm is used, the information transmission speed in the direction of the arrow in Figure 8 will be equal to or greater than the value set by the information transmission speed determination unit C5, as the speed transmitted by the search algorithm is added to the original transmission speed. On the other hand, the information transmission speed in the direction perpendicular to the arrow direction in Figure 8 will be the speed set by the information transmission speed determination unit C5.

Therefore, it is advisable to determine the order in which the potential field of each node is updated using a search algorithm that uses nodes located on the boundary of a specific area as the initial node, as shown in Figure 8. In this case, gradual information transmission from the boundary of the area centered on mobile unit E1 allows surrounding information to be transmitted to mobile unit E1 more quickly, thereby improving responsiveness to the surrounding environment.

However, if a parallel computing device such as a GPU (Graphics Processing Unit) can be used to update each node, it may be preferable to perform parallel calculations for each node rather than using a search algorithm within areas with the same update period.

<Modification of Example 2>
Fig. 9 is a block diagram showing an example of the configuration of a potential field update unit C4 according to a modified example of Example 2. The configuration of the potential field update unit C4 in Fig. 9 is obtained by adding an estimated movement cost calculation unit U5 to the configuration in Fig. 5 .

The estimated movement cost calculation unit U5 calculates an estimated movement cost, which is an estimate of the cost of movement to destination E2. The potential value update unit U4 calculates and issues a potential value for each node based on the movement cost of each node calculated by the movement cost calculation unit U1, the update period calculated by the update period calculation unit U2, the update discretization width calculated by the update period calculation unit U2, and the estimated movement cost calculated by the estimated movement cost calculation unit U5. By using the estimated movement cost to destination E2, the potential value update unit U4 can quickly form a potential field with a gradient toward destination E2 without waiting for information about destination E2 to be transmitted.

This modified example is particularly effective when it is desired to retransmit information about destination E2 without generating an initial potential using the initial potential field generation unit C1, for example when the movement path to destination E2 becomes passable again after being blocked by an obstacle. The potential field acting as an attractive force generated by the estimated movement cost calculation unit U5 is added to the potential field acting as a repulsive force generated by the movement cost calculation unit U1, allowing the moving body E1 to reach destination E2 more smoothly.

Figure 10 is a schematic diagram showing an example of a method for calculating estimated movement costs when updating a potential field, and shows an example in which the estimated movement cost is calculated as a value corresponding to the straight-line distance (Euclidean distance) from destination E2. For example, the potential value update unit U4 can easily update the potential field by adding the movement cost calculated by movement cost calculation unit U1 and the estimated movement cost calculated by estimated movement cost calculation unit U5 multiplied by a small coefficient as the new movement cost.

In this modified example, based on a differential equation that can assign weights to the interior of space, the sum of the repulsive force function from obstacles calculated from observation information and the attractive force function toward destination E2 is assigned as a weight. This makes it possible to approximately determine a potential field with a gradient toward destination E2 without waiting for information about destination E2 to be transmitted through potential updates, thereby improving responsiveness to surrounding information.

Example 3
Example 3 shows a specific example of the information transmission rate determiner C5 included in the control device S5 according to the embodiment. FIG. 11 is a schematic diagram showing an example of the processing of the information transmission rate determiner C5 according to Example 3. In this example, a high-speed information transmission area A1 and a low-speed information transmission area A2 surrounding the high-speed information transmission area A1 are defined with the moving object E1 at the center, and an information transmission rate is defined for each area (the information transmission rate in the high-speed information transmission area A1 is set to be faster than the information transmission rate in the low-speed information transmission area A2). Thus, in this example, the information transmission rate of each node dynamically changes depending on the position of the moving object E1.

In this way, by defining a high-speed information transmission area A1 near moving body E1 and defining a low-speed information transmission area A2 surrounding it, it is possible to focus on updating the potential field in the area close to moving body E1, making it possible to respond immediately, for example, if a person suddenly appears near moving body E1.

Figure 11 shows an example in which two information transmission areas with different information transmission speeds are defined, but the number of information transmission areas may be three or more, and the information transmission speed for each area is set so that the closer the area is to mobile unit E1, the faster the information transmission speed. Also, it is not necessary to define the information transmission speed for each area; for example, the information transmission speed may be defined as a continuous function centered on mobile unit E1, with the information transmission speed decreasing the further away from mobile unit E1 it is.

<First Modification of Third Embodiment>
FIG. 12 shows a first modified example of the third embodiment, and is a schematic diagram illustrating an example of the processing of the information transmission rate determination unit C5. In this modified example, the information transmission rate of each node is calculated based on the potential update amount around the node (i.e., the amount of change in potential due to the update of the potential field). FIG. 12 shows an example of calculating the information transmission rate using a sigmoid function. Note that, because the potential field will no longer be updated if the information transmission rate becomes zero, a minimum transmission rate is set as the lower limit of the information transmission rate in FIG. 12.

According to this modified example, the information transmission speed of each node increases as the amount of potential updates in the surrounding area increases. Therefore, the information transmission speed increases in nodes in dynamic environments where environmental changes occur frequently, improving responsiveness in dynamic environments.

<Second Modification of the Third Embodiment>
13 is a schematic diagram showing an example of the processing of the information transmission rate determination unit C5, which is a second modification of the third embodiment. In this modification, the information transmission rate of each node is increased to the maximum transmission rate when an environmental change occurs at that node, and then exponentially decays over time. In this modification, the information transmission rate is also increased in nodes in dynamic environments where environmental changes occur frequently, thereby improving responsiveness in dynamic environments.

Note that the nodes that increase their transmission speed when an environmental change occurs may be limited to the node where the environmental change occurred, or may also increase their information transmission speed for nodes where no environmental change occurred.

Example 4
Example 4 shows a specific example of a command value calculation unit C3 included in a control device S5 according to an embodiment. Fig. 14 is a block diagram showing an example of the configuration of the command value calculation unit C3 according to Example 4. The command value calculation unit C3 includes a gradient vector calculation unit M1, an initial command value calculation unit M2, an evaluation value calculation unit M3, a command value correction unit M4, and a convergence determination unit M5.

The gradient vector calculation unit M1 calculates the gradient vector of the potential field at the position of the moving body E1 (hereinafter referred to as the "potential gradient vector") from the potential field. The command value initial value calculation unit M2 calculates the initial value of the command value for the moving body E1 from the potential gradient vector at the position of the moving body E1 calculated by the gradient vector calculation unit M1. The evaluation value calculation unit M3 calculates the evaluation value of the command value for the moving body E1. The command value modification unit M4 modifies the command value for the moving body E1 in a direction that increases the evaluation value. The convergence determination unit M5 determines the command value that maximizes the evaluation value by having the evaluation value calculation unit M3 and the command value modification unit M4 perform iterative calculations until the convergence condition is met. The evaluation value is calculated based on the potential field, and in this embodiment, the evaluation value is the integral of the movement path of the potential. However, the evaluation value is not limited to this.

Note that the gradient vector calculation unit M1 may be implemented as a function of the potential field memory unit C2, rather than being implemented in the command value calculation unit C3.

Figure 15 is a schematic diagram showing an example of correcting a command value for a moving body E1. Figure 15 shows the moving body E1, the destination E2, an obstacle O1 that the moving body E1 needs to avoid, a potential field O2 stored in a potential field memory unit C2, an initial predicted trajectory O3 which is the predicted trajectory of the moving body E1 when using the initial command value calculated by the initial command value calculation unit M2, and a corrected predicted trajectory O4 which is the predicted trajectory of the moving body E1 when using a command value corrected by iterative calculation. The predicted trajectory of the moving body E1 can be obtained, for example, by calculating the trajectory of the moving body E1 assuming that it operates with the same command value for a predetermined period of time, based on a control model of the moving body E1.

Figure 16 is a schematic diagram showing an example of the processing of the command value initial value calculation unit M2. Figure 16 shows the initial predicted trajectory O3 of the moving body E1 calculated from the potential gradient vector, the maximum forward predicted trajectory O3-1, which is the predicted trajectory of the moving body E1 when the maximum forward speed is used as the command value, and the maximum backward predicted trajectory O3-2, which is the predicted trajectory of the moving body E1 when the maximum backward speed is used as the command value. The command value initial value calculation unit M2 calculates not only the initial predicted trajectory O3, but also the maximum forward predicted trajectory O3-1 and the maximum backward predicted trajectory O3-2 as initial values (initial trajectories) of the command value used in the iterative calculation. By adding the maximum forward predicted trajectory O3-1 and the maximum backward predicted trajectory O3-2 to the initial trajectory candidates, command values can be output stably regardless of the convergence of the potential field.

One method for determining which of the initial predicted trajectory O3, maximum forward predicted trajectory O3-1, and maximum backward predicted trajectory O3-2 to use as the initial trajectory is to modify the command values for each of the initial trajectory candidates and select the one with the highest evaluation value after convergence as the initial trajectory. Furthermore, the maximum forward predicted trajectory O3-1 and maximum backward predicted trajectory O3-2 do not necessarily need to be added; to reduce calculation costs, only the initial predicted trajectory O3 based on the potential gradient vector may be used as the initial trajectory.

The specific processing flow performed by the command value calculation unit C3 in this embodiment is described below. Here, we show an example in which the speed command value for a moving object E1 having a differential two-wheel mechanism is calculated based on a potential field defined in two dimensions.

First, the gradient vector calculation unit M1 calculates the potential gradient vector at the position of the mobile unit E1 from the potential field information stored in the potential field memory unit C2 and the position information of the mobile unit E1 issued by the mobile unit position distribution device S4. The potential gradient vector is calculated using the following formula (5):

In equation (5), _vN is a potential gradient vector, U is a potential value, the moving direction axis of the moving body coordinate E5 is the x component, and the lateral axis is the y component.

Next, the command value initial value calculation unit M2 calculates the initial value of the speed command value for the moving object E1 based on the potential gradient vector calculated by the gradient vector calculation unit M1. The speed command value is calculated, for example, using the following formula (6):

In equation (6), v _N [0] is the first component of the potential gradient vector, v _N [1] is the second component of the potential gradient vector, u _x is the initial value of the velocity command value in the forward direction (x-axis direction in the moving body coordinate E5), u _xmax is the maximum translational velocity, u _θ is the initial value of the velocity command value in the turning direction, and u _θmax is the maximum turning velocity.

When the moving body E1 moves forward, the y-axis component of the potential gradient vector is used as the turning speed, and when the moving body E1 moves backward, it turns at the maximum value in the direction of the y-axis component of the potential gradient vector. As described above, the command value initial value calculation unit M2 uses the maximum forward predicted trajectory O3-1 and the maximum backward predicted trajectory O3-2 as candidates for initial values for the iterative calculation, in addition to the initial predicted trajectory O3 based on the speed command value.

Next, the evaluation value calculation unit M3 evaluates the initial value of each speed command value. One evaluation method is to use an evaluation function in which the integral value of the potential on each predicted trajectory becomes the penalty.

Next, the command value correction unit M4 corrects the speed command value in a direction that increases the evaluation value calculated by the evaluation value calculation unit M3. The speed command value can be corrected using, for example, the hill-climbing method.

Next, the convergence determination unit M5 determines whether the corrected speed command value has converged. A convergence condition can be, for example, that the amount of correction made by the command value correction unit M4 is equal to or less than a predetermined threshold value.

In this embodiment, three initial value candidates for the iterative calculation were used: the initial predicted trajectory O3 based on the potential gradient vector, the maximum forward predicted trajectory O3-1, and the maximum backward predicted trajectory O3-2. However, other initial value candidates, such as the maximum turning predicted trajectory, may also be added. Furthermore, in the specific example described above, the attitude of the moving body E1 is not evaluated. However, the turning operation of the moving body E1 can also be evaluated by adding, for example, the magnitude of the angle between the potential gradient vector and the x-axis of the moving body coordinate system to the evaluation function.

In this embodiment, it is assumed that a more ideal movement trajectory exists near the potential gradient vector, and by evaluating the predicted trajectory up to a certain time ahead and fine-tuning the command value, it is possible to efficiently calculate a command value that is appropriate for the environment. This makes it possible to smoothly guide the moving object E1 even in severe environments such as small spaces.

Note that while this embodiment focuses on potential fields generated using differential equations that require iterative calculations, this embodiment may also be applied to artificial potential methods.

<Hardware configuration example>
17 and 18 are diagrams illustrating examples of the hardware configuration of the control device S5. The functions of the components of the control device S5 illustrated in FIG. 1 are realized, for example, by a processing circuit 100 illustrated in FIG. 17. That is, the control device S5 includes a processing circuit 100 for determining the information transmission speed of the potential field for each node, updating the potential field in accordance with the information transmission speed based on observation information about the surroundings of the mobile object, and calculating a command value for the mobile object based on the gradient vector of the potential field. The processing circuit 100 may be dedicated hardware, or may be configured using a processor (also referred to as a central processing unit (CPU), processing device, arithmetic unit, microprocessor, microcomputer, or DSP (Digital Signal Processor)) that executes a program stored in memory.

When the processing circuit 100 is dedicated hardware, the processing circuit 100 may be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a combination of these. The functions of each component of the control device S5 may be realized by a separate processing circuit, or these functions may be realized together by a single processing circuit.

Figure 18 shows an example of the hardware configuration of the control device S5 when the processing circuit 100 is configured using a processor 101 that executes a program. In this case, the functions of the components of the control device S5 are realized by software, etc. (software, firmware, or a combination of software and firmware). The software, etc. is written as a program and stored in memory 102. The processor 101 realizes the functions of each part by reading and executing the program stored in memory 102. In other words, the control device S5 includes memory 102 for storing a program that, when executed by the processor 101, results in the following processes: determining the information transmission speed of the potential field for each node; updating the potential field in accordance with the information transmission speed based on observation information about the surroundings of the mobile object; and calculating a command value for the mobile object based on the gradient vector of the potential field. In other words, this program can be said to cause a computer to execute the procedures and methods of operation of the components of the control device S5.

Here, memory 102 may be, for example, non-volatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable Read Only Memory), EEPROM (Electrically Erasable Programmable Read Only Memory), HDD (Hard Disk Drive), magnetic disk, flexible disk, optical disk, compact disk, mini disk, DVD (Digital Versatile Disc) and its drive device, or any other storage medium that will be used in the future.

The above describes a configuration in which the functions of the components of the control device S5 are realized either by hardware or software, etc. However, this is not limited to this, and the configuration may also be such that some of the components of the control device S5 are realized by dedicated hardware, and other components are realized by software, etc. For example, the functions of some components can be realized by the processing circuit 100 as dedicated hardware, and the functions of other components can be realized by the processing circuit 100 as the processor 101 reading and executing a program stored in memory 102.

As described above, the control device S5 can realize each of the above-mentioned functions through hardware, software, etc., or a combination of these.

In addition, it is possible to freely combine each embodiment, and to modify or omit each embodiment as appropriate.

<Additional Notes>
Various aspects of the present disclosure are summarized below as appendices.

(Appendix 1)
A mobile object control device that generates a potential field that determines a guiding direction of a mobile object on a plurality of nodes obtained by discretizing a space in which the mobile object moves,
an information transmission rate determination unit that determines an information transmission rate of the potential field for each of the nodes;
a potential field update unit that updates the potential field in accordance with the information transmission speed based on observation information about the surroundings of the moving object;
a command value calculation unit that calculates a command value for the moving object based on a gradient vector of the potential field;
Equipped with
the information transmission rate determination unit determines the information transmission rate of each of the nodes in accordance with a position of the mobile object;
Mobile control device.

(Appendix 2)
A mobile object control device that generates a potential field that determines a guiding direction of a mobile object on a plurality of nodes obtained by discretizing a space in which the mobile object moves,
an information transmission rate determination unit that determines an information transmission rate of the potential field for each of the nodes;
a potential field update unit that updates the potential field in accordance with the information transmission speed based on observation information about the surroundings of the moving object;
a command value calculation unit that calculates a command value for the moving object based on a gradient vector of the potential field;
Equipped with
the information transmission rate determination unit determines the information transmission rate of each of the nodes based on an update amount of the potential field in each of the nodes.
Mobile control device.

(Appendix 3)
A mobile object control device that generates a potential field that determines a guiding direction of a mobile object on a plurality of nodes obtained by discretizing a space in which the mobile object moves,
an information transmission rate determination unit that determines an information transmission rate of the potential field for each of the nodes;
a potential field update unit that updates the potential field in accordance with the information transmission speed based on observation information about the surroundings of the moving object;
a command value calculation unit that calculates a command value for the moving object based on a gradient vector of the potential field;
Equipped with
the information transmission rate determination unit determines the information transmission rate of each of the nodes based on a timing of an environmental change in each of the nodes.
Mobile control device.

(Appendix 4)
the potential field update unit updates the potential field by assigning, as the weight of a differential equation that can assign weights to the interior of space, a sum of a repulsive function calculated from the observation information around the moving object and an attractive function toward the destination of the moving object.
4. A mobile object control device according to any one of claims 1 to 3.

(Appendix 5)
the information transmission rate determination unit defines a plurality of areas centered on the position of the mobile unit, and determines the information transmission rate of each of the nodes for each of the areas.
2. A mobile object control device according to claim 1.

(Appendix 6)
the potential field update unit determines an order of updating the potential field of each of the nodes within the specific region by a search algorithm in which the node located on the boundary of the specific region is set as an initial node.
4. A mobile object control device according to any one of claims 1 to 3.

(Appendix 7)
the command value calculation unit sets a command value for the moving body calculated from the gradient vector of the potential field as an initial value, and corrects the command value based on an evaluation value of the command value calculated based on the potential field.
4. A mobile object control device according to any one of claims 1 to 3.

C1 initial potential field generation unit, C2 potential field storage unit, C3 command value calculation unit, C4 potential field update unit, C5 information transmission speed determination unit, E1 mobile body, E2 destination, E3 movement space, E4 world coordinates, E5 mobile body coordinates, S1 environmental map distribution device, S2 observation device, S3 destination distribution device, S4 mobile body position distribution device, S5 control device, S6 drive device, N1 node, N2 high-speed information transmission, N3 medium-speed information transmission, N4 low-speed information transmission, U1 movement cost calculation unit, U2 update period calculation unit, U3 update discretization width calculation unit, U4 potential value update unit, U5 estimated movement cost calculation unit, A1 high-speed information transmission area, A2 low-speed information transmission area, M1 gradient vector calculation unit, M2 command value initial value calculation unit, M3 evaluation value calculation unit, M4 command value correction unit, M5 convergence determination unit, O1 Obstacle, O2 potential field, O3 initial predicted trajectory, O3-1 maximum forward predicted trajectory, O3-2 maximum backward predicted trajectory, O4 corrected predicted trajectory, 100 processing circuit, 101 processor, 102 memory.

Claims (7)

A mobile object control device that generates a potential field that determines a guiding direction of a mobile object on a plurality of nodes obtained by discretizing a space in which the mobile object moves,
an information transmission rate determination unit that determines an information transmission rate of the potential field for each of the nodes;
a potential field update unit that updates the potential field in accordance with the information transmission speed based on observation information about the surroundings of the moving object;
a command value calculation unit that calculates a command value for the moving object based on a gradient vector of the potential field;
Equipped with
the information transmission speed of the potential field is a speed at which the potential field update unit transmits the observation information to the potential field and updates the potential field,
the information transmission rate determination unit determines the information transmission rate of each of the nodes in accordance with a position of the mobile object;
Mobile control device. A mobile object control device that generates a potential field that determines a guiding direction of a mobile object on a plurality of nodes obtained by discretizing a space in which the mobile object moves,
an information transmission rate determination unit that determines an information transmission rate of the potential field for each of the nodes;
a potential field update unit that updates the potential field in accordance with the information transmission speed based on observation information about the surroundings of the moving object;
a command value calculation unit that calculates a command value for the moving object based on a gradient vector of the potential field;
Equipped with
the information transmission speed of the potential field is a speed at which the potential field update unit transmits the observation information to the potential field and updates the potential field,
the information transmission rate determination unit determines the information transmission rate of each of the nodes based on an update amount of the potential field in each of the nodes.
Mobile control device. A mobile object control device that generates a potential field that determines a guiding direction of a mobile object on a plurality of nodes obtained by discretizing a space in which the mobile object moves,
an information transmission rate determination unit that determines an information transmission rate of the potential field for each of the nodes;
a potential field update unit that updates the potential field in accordance with the information transmission speed based on observation information about the surroundings of the moving object;
a command value calculation unit that calculates a command value for the moving object based on a gradient vector of the potential field;
Equipped with
the information transmission speed of the potential field is a speed at which the potential field update unit transmits the observation information to the potential field and updates the potential field,
the information transmission rate determination unit determines the information transmission rate of each of the nodes based on a timing of an environmental change in each of the nodes.
Mobile control device. the potential field update unit updates the potential field by assigning, as the weight of a differential equation that can assign weights to the interior of space, a sum of a repulsive function calculated from the observation information around the moving object and an attractive function toward the destination of the moving object.
The mobile object control device according to any one of claims 1 to 3. the information transmission rate determination unit defines a plurality of areas centered on the position of the mobile unit, and determines the information transmission rate of each of the nodes for each of the areas.
The mobile object control device according to claim 1 . the potential field update unit determines an order of updating the potential field of each of the nodes within the specific region by a search algorithm in which the node located on the boundary of the specific region is set as an initial node.
The mobile object control device according to any one of claims 1 to 3. the command value calculation unit sets a command value for the moving body calculated from the gradient vector of the potential field as an initial value, and corrects the command value based on an evaluation value of the command value calculated based on the potential field.
The mobile object control device according to any one of claims 1 to 3.