Moving body control system and moving body control method

Abstract

The present invention comprises: a movement destination setting unit that generates a movement route including a route shape and a control target; a parameter setting unit that sets a control parameter to be used in the movement route; and an information storage unit that stores a travel record in which a route shape along which a moving body has autonomously traveled in the past, the control target, and the control parameter are combined. A past block in which the route shape and the control target in the travel record are divided is generated, a past block set formed by similar past blocks in which the same control parameter is associated is generated, a current block obtained by dividing the route shape of the movement route and the control target is generated, the route shape and the control target of the current block and the past block set are compared, and a control parameter corresponding to a past block set that is similar to the current block is generated. This makes it possible to reduce the burden at the time of new introduction of the autonomous moving body into a site, and to quickly introduce the autonomous moving body into the site.

Description

Mobile object control system and mobile object control method

[0001] The present invention relates to a mobile object control system and a mobile object control method.

[0002] The introduction of autonomous vehicles is progressing to address driver shortages in construction sites, mines, transportation, and logistics, as well as in transportation and logistics. These autonomous vehicles require different driving conditions depending on the task at hand. For example, transporting people or precision machinery requires minimal vehicle sway to prevent car sickness and damage, while transporting soil and ore requires short round-trip times to increase the amount of material transported per hour. Different driving conditions require different control targets (e.g., trajectory targets expressed as nodes and branches, and target velocity distributions) for the autonomous vehicles. This necessitates the design of new control targets for each new task, resulting in significant engineering costs for the operator. Furthermore, different control targets require controller adjustments to achieve those targets. Adjusting the controller for each new control target further increases the burden on the operator. However, without such adjustments, a general-purpose configuration that can be applied to a variety of tasks is required, resulting in reduced control accuracy and limitations on the tasks that can be performed compared to task-specific adjustments.

[0003] A known prior art for addressing such a problem is, for example, that described in Patent Document 1. Patent Document 1 discloses a vehicle driving assistance device that outputs a target driving pattern serving as a reference for controlling vehicle speed in accordance with the vehicle's driving state, in which the device classifies the vehicle's driving state, including the presence or absence of a preceding vehicle and the presence or absence of a following vehicle, based on each time the vehicle travels through a section preset on a map, stores driving patterns that indicate changes in vehicle speed or inter-vehicle distance from the preceding vehicle within the section, and generates and outputs the target driving pattern based on a driving pattern that is classified as a driving state of the vehicle that is the same as the current driving state of the vehicle from among the stored driving patterns.

[0004] JP 2013-193671 A

[0005] In the above-mentioned conventional technology, a target driving pattern, which is the control target, is generated from driving patterns recorded during work performed by an autonomous mobile body in the past, and is therefore effective in reducing the burden of repeated operations at the same site or with the same vehicle (autonomous mobile body).

[0006] However, when an autonomous mobile unit is newly introduced to a site, it is likely to have unique conditions depending on the combination of the site conditions, the vehicle used, the tasks assigned, etc., and therefore it is difficult to obtain driving patterns for the same conditions as the site where the autonomous mobile unit is being introduced from past records at other sites. In other words, when an autonomous mobile unit is newly introduced, it is necessary to set new control targets, etc., which limits the burden reduction and may result in a long time required for introduction to the site.

[0007] The present invention has been made in consideration of the above, and aims to provide a mobile object control system and a mobile object control method that can reduce the burden when introducing a new autonomous mobile object to a site and enable the system to be introduced to the site quickly.

[0008] The present application includes multiple means for solving the above-mentioned problems. One example is a mobile body control system comprising: a destination setting unit that determines a destination for a mobile body and generates a travel route including a route shape to the destination and control targets; a parameter setting unit that sets control parameters to be used on the travel route set by the destination setting unit; and an information storage unit that stores a driving record that records at least combinations of route shapes, control targets, and control parameters along which the mobile body has traveled autonomously in the past; the parameter setting unit generates past blocks by dividing the route shape and control targets of the driving record, generates a past block set formed by similar past blocks to which the same control parameters can be assigned, generates a current block by dividing the route shape and control targets of the travel route, compares the route shape and control targets of the current block with the route shape and control targets of the past block set, and generates control parameters corresponding to the past block set similar to the current block.

[0009] According to the present invention, it is possible to reduce the burden when introducing a new autonomous moving body into a site, and to quickly introduce the body into the site.

[0010] 1 is a functional block diagram schematically showing the overall configuration of a mobile body control system according to a first embodiment; FIG. 2 is a diagram showing an autonomous mobile body operating in a mine, which is the site; FIG. 3 is a diagram showing a coordinate system set in a mine; FIG. 4 is a functional block diagram schematically showing the configuration of an autonomous mobile body; FIG. 5 is a diagram showing an example of a travel route set in a travel-permitted passage; FIG. 6 is a diagram showing how trajectory targets are set for a set travel route; FIG. 7 is a diagram showing how speed targets are set for a set travel route; FIG. 8 is a diagram showing how a set travel route and trajectory targets are divided; FIG. 9 is a diagram showing speed targets for each divided block; FIG. 10 is a diagram showing an example of block information for each divided block; FIG. 11 is a diagram showing an example of the relationship between block features and control parameters; FIG. 12 is a diagram schematically showing a method of searching for solution candidates; and FIG. 13 is a flowchart showing the processing content of a mobile body control system.

[0011] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0012] The mobile object control system in this embodiment controls mobile objects such as robots and vehicles capable of autonomous operation, calculates a movement plan (time series of coordinates, attitude, speed, etc.) for these mobile objects, and controls the mobile objects to follow this path plan. In the following explanation, an example will be given in which an autonomous vehicle operating in a mine is the autonomous mobile object to be controlled. Note that the application of the present invention is not limited to generating movement paths for mobile objects operating in a mine, but can also be applied to, for example, generating movement paths for transport vehicles in a port or generating paths for robots moving within a theme park.

[0013] First Embodiment A first embodiment of the present invention will be described with reference to FIGS.

[0014] FIG. 1 is a functional block diagram showing a schematic overall configuration of a mobile object control system according to this embodiment.

[0015] 1, the mobile object control system is generally composed of a mobile object position acquisition unit A101, a destination setting unit A102, a control information storage unit A103, a parameter setting unit A104, a control information storage unit A105, and a mobile object control unit A106. Note that, for simplicity of explanation, Fig. 1 illustrates an example in which there is one mobile object to be controlled, but the present invention is not limited to this and can also be applied to cases in which multiple mobile objects are controlled.

[0016] 2 is a diagram showing how autonomous mobile bodies are operating in a mine, which is a site. In FIG. 2, it is assumed that an autonomous mobile body B100 and a non-autonomous mobile body B102 are mixed and each of them moves along a travel-permitted path B101.

[0017] The autonomous moving body B100 is, for example, a dump truck capable of carrying loads such as earth and sand or ore. Note that the autonomous moving body B100 is not limited to so-called four-wheel vehicles such as dump trucks, and various types of autonomous moving bodies can be handled, for example, a differential two-wheel robot, an omni-wheel robot, a forklift, a towing vehicle, etc. Furthermore, the non-autonomous moving body B102 is, for example, a hydraulic excavator equipped with a loading function.

[0018] Examples of work performed in a mine include an autonomous mobile body B100 moving to a designated position so that a non-autonomous mobile body B102 can load cargo, and then discharging the loaded cargo in a designated area.

[0019] The target position to which the autonomous mobile body B100 should move is managed by a management server B104. The management server B104 may be installed within the mine or at another location outside the mine. The target position determined by the management server B104 is transmitted to the autonomous mobile body B100 via a wireless communication line B103 within the mine.

[0020] Fig. 3 is a diagram showing a coordinate system set in a mine, and Fig. 4 is a functional block diagram showing a schematic configuration of an autonomous mobile body.

[0021] 3, an XY coordinate system is set in the mine for the autonomous mobile body B100, with the position of the autonomous mobile body B100 represented by X-axis and Y-axis values ​​(x, y), and the orientation (for example, the forward direction of the autonomous mobile body B100) represented by the angle (θ) with the X-axis. The autonomous mobile body B100 is equipped with various sensors, such as an encoder that detects the rotation speed of the wheels (not shown), an inertial sensor (IMU: Inertial Measurement Unit) that acquires the rotational speed and acceleration of the vehicle body, LiDAR (Light Detection And Ranging), and GNSS (Global Navigation Satellite System).

[0022] In FIG. 4, the autonomous moving body B100 is composed of a recording unit M01, a communication unit M02, a sensor M03, a moving body control unit M04, a self-position calculation unit M05, and an actuator M06.

[0023] The recording unit M01 is a recording device of the calculation unit provided in the autonomous moving body B100, and corresponds to an HDD (hard disk drive), SSD (solid state drive), or memory. The recording unit records various control programs, parameters used in the control programs, and control targets distributed from the management server.

[0024] The communication unit M02 is a wireless communication device compatible with 5G, LTE (Long Term Evolution), etc., and can communicate with the management server B104 via a wireless communication line B103 within the mine.

[0025] The sensor M03 corresponds to the sensors provided on the autonomous moving body B100, such as the encoder, IMU, LiDAR, and GNSS described above.

[0026] The mobile body control unit M04 has a function of performing calculations related to the traveling control of the mobile body, and is executed by a controller (not shown) provided in the autonomous mobile body B100. The calculations performed by the mobile body control unit M04 will be described in detail later.

[0027] The self-position calculation unit M05 is a function that calculates the autonomous mobile body B100's own position (x, y coordinates on a two-dimensional plane) and orientation using a combination of multiple sensors attached to the autonomous mobile body B100. The functions of the self-position calculation unit M05 are executed in the controller, similar to the mobile body control unit M04. The self-position calculation unit M05 can be realized by a technology known as SLAM (Simultaneous Localization And Mapping), and therefore a detailed description thereof will be omitted.

[0028] The actuator M06 corresponds to the traveling motor, steering mechanism, and braking device of the autonomous moving body B100.

[0029] The autonomous moving body B100 receives control targets and control parameters from the management server B104 via the communication unit M02, records them in the recording unit M01, and controls its own operation in accordance with this data. Note that if the communication cycle is sufficiently fast, the data may not be stored in the recording unit M01 but may be used directly by the moving body control unit M04 (shown by the dotted line in FIG. 4).

[0030] Here, the mobile object control system will be described in detail.

[0031] The mobile object control system is a system that manages the operation of the autonomous mobile object B100 within the mine based on information acquired by the autonomous mobile object B100, and much of the processing is executed by the management server B104.

[0032] As described above, the mobile object control system is composed of a mobile object position acquisition unit A101, a destination setting unit A102, a control information storage unit A103, a parameter setting unit A104, a control information storage unit A105, and a mobile object control unit A106.

[0033] The mobile object position acquisition unit A101 is composed of a sensor A101a and a self-position calculation unit A101b. Each function of the mobile object position acquisition unit A101 corresponds to the sensor M03 and the self-position calculation unit M05 shown in FIG.

[0034] The control information memory unit A103 stores map information, which is a database of the terrain such as the curvature, gradient, and road width of the travel-permitted passage B101; vehicle information, which is a database of vehicle specifications such as the maximum speed, maximum acceleration, and minimum turning radius of the autonomous mobile body B100; and block information, which is a database of blocks composed of the terrain and control targets and the control parameters corresponding to those blocks.

[0035] The destination setting unit A102 calculates the target position of the autonomous mobile body B100 based on the current position of the autonomous mobile body B100 calculated by the mobile body position acquisition unit A101 and the work being done within the mine, sets a movement route on the travel-permitted passage B101 from the current position to the target position, and generates a control target.

[0036] For example, the tasks to be assigned are loading, dumping, maintenance, etc., and the corresponding target locations are loading areas, dumping areas, parking areas, etc. The assignment of tasks within the mine is realized, for example, by linking with an FMS (Fleet Management System) not shown. Note that a detailed description of the processing contents of the FMS will be omitted.

[0037] FIG. 5 is a diagram showing an example of a travel route set in a travel-permitted passage.

[0038] As shown in Figure 5, when setting a travel route, a travel route from the current position to the target position is searched for on the travel-permitted passage B101. Methods such as the Dijkstra algorithm and the A* algorithm have been proposed for route search, and these methods can be used. Note that detailed explanations of the Dijkstra algorithm and the A* algorithm will be omitted here.

[0039] Fig. 6 is a diagram showing how trajectory targets are set for a set movement route, and Fig. 7 is a diagram showing how speed targets are set for a set movement route.

[0040] As shown in Figures 6 and 7, the generation of control targets involves generating trajectory targets and speed targets to be followed by the autonomous mobile body B100. Methods such as the Timed Elastic Band Algorithm (TEB) and Model Predictive Trajectory Optimization have been proposed as generation methods that can be optimized for different control objectives (efficiency-oriented, energy-saving-oriented, safety-oriented, etc.) based on the terrain and vehicle specifications, and these can be realized. Note that a detailed explanation of these methods will be omitted.

[0041] The parameter setting unit A104 has a block search unit A104a and a parameter generation unit A104b.

[0042] Fig. 8 is a diagram showing how the set movement route and trajectory targets are divided, and Fig. 9 is a diagram showing how the speed targets are divided into blocks.

[0043] 8, the block search unit A104a divides the travel route and the control target to generate blocks, searches the obtained blocks from the block information, and obtains the corresponding control parameters. The parameter generation unit A104b generates control parameters suitable for the control target based on the block topography, vehicle specifications, and the control target.

[0044] The blocks may be divided, for example, at points where the changes in terrain, such as gradient or road width, relative to the distance traveled become constant, making it easier to search for blocks with a common tendency in terrain changes, or at points where the changes in trajectory targets or speed targets relative to the distance traveled become constant, making it easier to search for blocks with a common tendency in control targets. In sections where the changes in trajectory targets or speed targets are constant, the state of the autonomous mobile body B100 becomes steady, making it easier to switch control parameters between blocks.

[0045] Fig. 10 is a diagram showing an example of block information for each divided block, and Fig. 11 is a diagram showing an example of the relationship between block characteristics and control parameters.

[0046] As shown in Fig. 10, the block information stored in the control information storage unit A103 records the route shape (topography), start target, speed target, etc. as characteristics of each divided block. The route shape (topography) stored in the control information storage unit A103 is represented, for example, by the distance traveled, the road width at the initial position, the road width change rate, the curvature at the initial position, the curvature change rate, the gradient at the initial position, and the gradient change rate. The trajectory target is represented, for example, by the initial state and the state change rate. The speed target is represented, for example, by the initial speed and the speed change rate.

[0047] As shown in Fig. 11, control parameters are associated based on the features shown in Fig. 10. Note that, for ease of explanation, Fig. 11 illustrates an example in which two features of a block are associated with a control parameter, but the present invention is not limited to this, and for example, more than two types of features may be associated with a control parameter.

[0048] In the block search, if the characteristics of a newly obtained new block fall within the range corresponding to a certain control parameter P1 (as in the case of new block A in FIG. 11 ), control parameter P1 is associated with new block A. On the other hand, if the characteristics of the new block do not fall within the range corresponding to any control parameter (as in the case of new block B in FIG. 11 ), the existing block with the closest characteristics (smallest absolute difference in numerical values) may be associated with the corresponding control parameter, or a corresponding control parameter may be generated as described below.

[0049] The range of the control parameters may be the range of variation of the block features within which the tracking accuracy of the block with respect to the control target is within an acceptable range when the same control parameters are used, or, when control parameters are generated for each block, the range of variation of the control parameters within which the tracking accuracy of the block with respect to the control target is within an acceptable range may be determined, and blocks with overlapping ranges may be collected, and the range corresponding to the control parameter averaged over the overlapping range may be set from a range based on a confidence interval by calculating a normal distribution for each feature of each collected block. Here, the range of the tracking accuracy with respect to the block's control target indicates the accuracy with which the autonomous mobile body B100 can complete the assigned task, such as the accuracy with which the autonomous mobile body B100 can travel without deviating from the travel-permitted path B101 or the accuracy with which the autonomous mobile body B100 can achieve the estimated arrival time.

[0050] Because the block information stores control parameters and the range of block characteristics corresponding to those parameters, this correspondence can be used to divide the movement route and control targets. First, the terrain, trajectory target, and speed target at the current location are used as initial values ​​to search for corresponding control parameter candidates. Next, changes in the terrain, trajectory target, and speed target as the block moves in the direction of the path are tracked, and the block is divided into one block until a single control parameter candidate is found. The same process is repeated using the terrain, trajectory target, and speed target at the divided point as initial values, and the block is further divided. If no corresponding control parameter is found in the search using the initial values, the block is moved in the direction of the path until a corresponding control parameter appears, and the block is divided into one block until the corresponding control parameter is found. Since no control parameters are assigned to this block, the existing block with the closest characteristics (smallest absolute difference in numerical values) can be associated with the corresponding control parameters, or corresponding control parameters can be generated as described below. This method minimizes the number of divisions required for assigning control parameters and minimizes the number of blocks for which control parameters need to be generated.

[0051] The above method makes it possible to assign control parameters to the entire movement route. Here, if the control parameters differ between blocks, switching may cause discontinuity in the control input. For example, if PID control is used, multiplying the state quantity accumulated using the integral component parameters of the previous block by the integral component parameters of the new block will cause the control input to become discontinuous, resulting in a decrease in tracking accuracy until the state quantity is newly accumulated. Therefore, in order to continuously change the control input, a buffer section is provided that linearly connects the control parameters before and after the boundary between blocks. The length of the buffer section is set within a range such that the decrease in tracking accuracy in the buffer section does not fall below the accuracy at which the autonomous mobile body B100 can complete the assigned task.

[0052] The control parameters may be generated using, for example, Bayesian optimization (BO). BO is an algorithm that efficiently searches for parameters based on a specific evaluation index c by repeating simulations and experiments so as to maximize or minimize the evaluation index c. In this embodiment, the evaluation index c indicates the degree of achievement of a control objective (such as tracking accuracy for a control target) that should be achieved within a block in order for the autonomous mobile body B100 to complete an assigned task. For example, the evaluation index c may be the time tend required for the autonomous mobile body B100 to reach the target position r.

[0053] Here, a control parameter adjustment method using BO will be explained using the example of the problem of minimizing the time tend to reach the target position r. A vector ρ is a variable that summarizes the control parameters to be adjusted. Since the evaluation index c is the time tend to reach the target position r, an optimization problem can be given as a problem of finding the optimal ρ* that minimizes this value. Generally, this optimization problem cannot be solved directly. In BO, several solution candidates ρi are given, and the values ​​of the evaluation index c that are actually obtained are confirmed, while the function shape of the evaluation index c is estimated and the next solution candidate ρi is calculated.

[0054] FIG. 12 is a diagram showing a schematic diagram of a method for searching for solution candidates.

[0055] In FIG. 12, the black dots represent evaluation values ​​obtained from the solution candidates ρi (i = 1, ..., 6) obtained so far. Using the solution candidates obtained so far, it is possible to calculate the estimated value of the function shape (solid line) and its variance (dashed line). The variance takes a large value in an area where no solution candidate ρi has been found so far. The next solution candidate ρ7 (see "△" in FIG. 12) is determined taking into account the estimated value of the function shape and the variance. By performing the above processing, it is possible to search for control parameters very efficiently compared to the case where the solution candidates ρi are verified brute-force.

[0056] Here, even in different blocks, if the evaluation index c calculated using the same control parameter is equivalent, that is, if the control objective can be achieved equally in each block, the block and the control parameter may be stored in correspondence with each other in the block information. In this case, there may be multiple ranges corresponding to a certain control parameter, as in the case of control parameter P1 in Figure 11.

[0057] The control information storage unit A105 acquires the blocks on the movement route and the corresponding control parameters from the parameter setting unit A104 and stores them.

[0058] The mobile body control unit A106 calculates a control input to the actuator M06 that causes the position and orientation of the autonomous mobile body B100 acquired by the self-position calculation unit A101b to follow the control target. The control parameters used to calculate the control input are obtained from the control information storage unit A105, and are control parameters corresponding to the block corresponding to the current position acquired by the self-position calculation unit A101b. Control methods such as PID control and model predictive control (MPC) have been proposed for calculating the control input, and these can be used to realize the control. Note that detailed explanations of these methods will be omitted.

[0059] When the travel route is long or when the blocks are divided into many smaller blocks, the number of blocks and corresponding control parameters stored in the control information storage unit A105 becomes enormous, which is expected to cause a shortage of storage capacity in the recording device (recording unit M01) installed in the autonomous mobile unit B100. Therefore, it is desirable that the blocks on the travel route and the corresponding control parameters be managed in the recording device of the management server B104. In this implementation form, based on the position of the autonomous mobile unit B100 acquired by the self-position calculation unit A101b, the management server B104 delivers to the autonomous mobile unit B100 the blocks and control parameters that are expected to be requested by the mobile unit control unit A106, and the autonomous mobile unit B100 travels while receiving the necessary information.

[0060] FIG. 13 is a flowchart showing the processing contents of the mobile object control system.

[0061] 13, the mobile object control system first checks the current position of the autonomous mobile object B100 (step S100). This process is realized by the function of the mobile object position acquisition unit A101.

[0062] Next, based on the acquired position of the autonomous mobile body B100, it is determined whether there is a task that can be assigned to the autonomous mobile body B100 (step S110). If the determination result in step S110 is NO, that is, if it is determined that there is no task that can be assigned to the autonomous mobile body B100, there is no load to be released or loaded on the autonomous mobile body B100, and there is no need to move the autonomous mobile body B100, so the process returns to step S100.

[0063] Also, if the determination result in step S100 is YES, that is, if it is determined that there is a task that can be assigned to the autonomous mobile body B100, then a target position r corresponding to the assigned task is determined (step S120).

[0064] Note that even if the autonomous moving body B100 has no cargo to be dumped or loaded, that is, even if the determination result in step S110 is NO, if the autonomous moving body B100 staying at the location would interfere with the movement of other moving bodies, a task to move the autonomous moving body B100 to another location may be generated. Similarly, if the autonomous moving body B100 is a battery-powered vehicle, a task to move to a charging station may be generated if the remaining battery power is low, and if the autonomous moving body B100 is a gasoline-powered vehicle, a task to move to a refueling station may be generated if the remaining gasoline power is low.

[0065] After the processing of step S120 is completed, a movement route of the autonomous moving body B100 is set (step S130), and a control target is generated (step S140). Note that the processing of steps S110, S120, S130, and S140 is realized by the function of the destination setting unit A102 using information stored in the control information storage unit A103.

[0066] After the processing of step S140 is completed, the movement route and the control target are divided into blocks (step S150), and the blocks on the movement route are searched for using block information to determine whether corresponding control parameters can be obtained from the block information for all blocks on the movement route (step S160). The processing of steps S150 and S160 is realized by the function of the block search unit A104a using information stored in the control information storage unit A103.

[0067] If the determination result in step S160 is NO, that is, if there is a block among the searched blocks for which the corresponding control parameters do not exist in the block information, then control parameters are generated for the blocks for which the corresponding control parameters do not exist in the block information (step S161). Note that this process is realized by the function of the parameter generation unit A104b using the information stored in the control information storage unit A103.

[0068] If the determination result in step S160 is YES, or if the processing of step S161 is completed, the control parameters of the block corresponding to the current position X of the autonomous mobile body B100 are subsequently set (step S170). This processing is realized by the function of the control information storage unit A105. Note that if all blocks and control parameters are recorded in the management server B104, the necessary information is provided to the autonomous mobile body B100 at the timing of the processing of step S170.

[0069] Next, the control input to the actuator M06 is calculated using the control parameters set in step S170 (step S180), and the process ends. Note that the process of step S180 is realized by the function of the moving body control unit A106.

[0070] The above-described processing of steps S100 to S180 is repeated for each control cycle. By performing the processing for each control cycle, a control target is generated based on the current position even if the current position deviates from the control target, and therefore an appropriate control target is provided. Furthermore, if the current position does not deviate from the control target, it is sufficient to perform the processing of steps S179 and S180 for each control cycle, thereby reducing calculation costs.

[0071] In this embodiment configured as described above, the burden of introducing a new autonomous mobile body to a site can be reduced, and the mobile body can be introduced quickly. In other words, if a system designer prepares a database of blocks and their corresponding control parameters in advance, the operator does not need to perform control design appropriate for the control objectives, and can optimally control the mobile body for each task. In particular, if previously generated blocks can be reused (if they can be expressed by rearranging existing blocks or if the same task is assigned to the same position as before), the control parameter generation process is unnecessary, and the same execution results can be obtained while reducing calculation costs.

[0072] Second Embodiment A second embodiment of the present invention will be described with reference to FIG.

[0073] This embodiment is a mobile object control system that can cope with changes in vehicle performance and disturbances that are difficult to record in advance in block information. In this embodiment, the same reference numerals are used to designate the same components as in the first embodiment, and descriptions thereof will be omitted where appropriate.

[0074] In mines, obstacles such as spilled rocks during transportation can fall onto the travel-permitted path B101. Furthermore, because autonomous vehicles used in mines have a large mass, braking systems using friction brakes for emergency use heat up after a single use, reducing braking performance until the brakes cool down, limiting continuous use. Such obstacles and changes in vehicle performance occur exceptionally, making it difficult to record them as block information. Therefore, in this embodiment, model predictive control (MPC) is introduced into the mobile unit control unit A106 of the mobile unit control system in the first embodiment. This enables the generation of control inputs that can track control targets even in the presence of disturbances or performance changes, such as obstacles or reduced vehicle braking performance that are not recorded in the block information, thereby enabling the vehicle to more reliably accomplish its assigned tasks in the mine.

[0075] FIG. 1 is a functional block diagram showing a schematic overall configuration of a mobile object control system according to this embodiment.

[0076] In FIG. 1, the mobile object control system is roughly composed of a mobile object position acquisition unit A101, a destination setting unit A102, a control information storage unit A103, a parameter setting unit A104, a control information storage unit A105, and a mobile object control unit A106.

[0077] The mobile object control unit A106 has an evaluation function setting unit A106a and a prediction control unit A106b.

[0078] The evaluation function setting unit A106a constructs an evaluation function by combining several evaluation functions and parameters related to the evaluation functions. For example, as shown in the following (Equation 1) to (Equation 3), a vector rk that summarizes the target position (xrk, yrk) and target orientation θrk of the autonomous mobile body B100 at each time is defined, and an evaluation function J can be constructed using a function (the first term in the following (Equation 1)) that evaluates the deviation ek = Xk - rk between the current state (xk, yk, θk) of the autonomous mobile body B100 and the target state, and a function (the second term in the following (Equation 1)) that evaluates the magnitude of the control input uk.

[0079]

[0080]

[0081]

[0082] The symbol T in the above (Equation 1) is a symbol representing transposition. Furthermore, the matrix Q is a weighting parameter related to the deviation ek, the matrix R is a weighting parameter related to the input uk, and the variable N is a parameter representing the evaluation interval (prediction interval). In this way, the evaluation function setting unit A106a includes multiple parameters, and these parameters are recorded in the control information storage unit A105.

[0083] Note that the above (Equation 1) shows one example of the evaluation function J, and is not limited to this. For example, an evaluation function in which only the final step N is evaluated separately (terminal cost) may be used, as shown in the following (Equation 4).

[0084]

[0085] The prediction control unit A106b calculates the control input uk in accordance with the concept of MPC so as to reduce the value of the evaluation function J set by the evaluation function setting unit A106a.

[0086] In MPC, in order to evaluate the evaluation function J up to the interval of the evaluation function (behavior from the current time to the next N steps), it is necessary to predict the future behavior of the autonomous mobile body B100. The behavior prediction of the mobile body up to the next N steps can be calculated recursively by inputting the control input u found to minimize the above (Equation 1) (or (Equation 4)) into the equation of motion of the autonomous mobile body B100.

[0087] MPC has the advantage of being able to explicitly handle constraints. For example, in the case of the autonomous vehicle shown in Figure 3, the upper and lower limits of the traveling speed v are determined by the rotational speed limit of the traveling motor and the braking limit of the braking device. By incorporating the operating characteristics of such actuators as constraints, it becomes possible to generate control inputs that the autonomous vehicle can actually execute. MPC that takes such constraints into account can be expressed as in the following (Equation 5) and (Equation 6).

[0088]

[0089]

[0090] In the above (Equation 6), the symbol [s.t.] is an abbreviation for [subject to], and means that the optimization problem is solved under the conditions following [s.t.]. The subscripts max and min correspond to upper and lower limits, respectively. Generally, to facilitate design, the weight parameter matrices Q and R are each given as a diagonal matrix. For example, Q can be expressed using the x-coordinate weight Qx, the y-coordinate weight Qy, and the orientation weight Qθ.

[0091] The control information storage unit A105 stores blocks on the movement route and weight matrices Qi and Ri (i is an index assigned to the block) as corresponding control parameters.

[0092] In the mobile object control unit A106, the evaluation function setting unit A106a acquires control parameters from the control information storage unit A105 based on the block containing the predicted position calculated by the prediction control unit A106b, and sets an evaluation function.

[0093] As described above, introducing MPC and reflecting it in the constraint conditions makes it possible to respond to disturbances and performance changes. However, in an environment where there is an obstacle on the trajectory target, the situation may deviate from the situation assumed in the block, making it impossible to achieve the trajectory target. In this case, the control parameters may become inappropriate. To address this issue, there are two methods: one is to regenerate the control target for the entire movement route based on the current position and reset the control parameters, and the other is to set a buffer section between the current position and the position where the situation returns to the assumed situation. The latter method in particular has the advantage of reducing calculation costs because it does not require recalculation of the entire movement route.

[0094] Here, a method for setting the buffer zone will be described. Because the MPC can predict state variables occurring ahead and calculate control inputs, the control inputs can be adjusted to improve tracking accuracy ahead after returning to the expected situation, rather than tracking accuracy near the current position that deviates from the expected situation. To achieve this, the control parameters for the buffer zone may be set to general-purpose control parameters that emphasize stabilizing driving, or weighting parameters may be reduced by a predetermined percentage to reduce the influence of the buffer zone on the entire prediction zone. The length of the buffer zone may be set to a position in the prediction zone from the current position to a position where the tracking error with respect to the control target falls below a predetermined percentage, or may be set to a position from the current position to the start position of the next block.

[0095] The other configurations are the same as those of the first embodiment.

[0096] The present embodiment configured as above can also achieve the same effects as the first embodiment.

[0097] Furthermore, by introducing MPC into the mobile unit control unit A106, it is possible to generate control inputs that can track the control target even in the presence of disturbances or performance changes such as exceptional obstacles or reduced vehicle braking performance that are not recorded in the block information, making it possible to more reliably accomplish the tasks assigned to the autonomous mobile unit B100 within the mine.

[0098] <Notes> The present invention is not limited to the above-described embodiments, and includes various modifications and combinations of embodiments within the scope of the gist of the present invention. Furthermore, the present invention is not limited to those having all of the configurations described in the above-described embodiments, and includes those in which some of the configurations are changed or deleted. Furthermore, the above-described configurations, functions, etc. may be realized in part or in whole by designing them as, for example, integrated circuits. Furthermore, the above-described configurations, functions, etc. may be realized in software by a processor interpreting and executing a program that realizes each function.

[0099] A101...mobile body position acquisition unit, A101a...sensor, A101b...self-position calculation unit, A102...destination setting unit, A103...control information storage unit, A104...parameter setting unit, A104a...block search unit, A104b...parameter generation unit, A105...control information storage unit, A106...mobile body control unit, A106a...evaluation function setting unit, A106b...prediction control unit, B100...autonomous mobile body, B101...travel-permitted passage, B102...non-autonomous mobile body, B103...wireless communication line, B104...management server, M01...recording unit, M02...communication unit, M03...sensor, M04...mobile body control unit, M04...direct mobile body control unit, M05...self-position calculation unit, M06...actuator

Claims

1. A mobile object control system comprising: a destination setting unit which determines a destination for a mobile object, and generates a movement route including a path shape to the destination and control targets; a parameter setting unit which sets control parameters to be used on the movement route set by the destination setting unit; and an information storage unit which stores a driving record which records at least a combination of a path shape, a control target, and a control parameter along which the mobile object has traveled autonomously in the past, wherein the parameter setting unit generates past blocks by dividing the path shape and control target of the driving record, generates a past block set formed of similar past blocks to which the same control parameters can be assigned, generates a current block by dividing the path shape and control target of the movement route, compares the path shape and control target of the current block with the path shape and control target of the past block set, and generates control parameters corresponding to the past block set similar to the current block.

2. A mobile object control system according to claim 1, wherein the parameter setting unit generates blocks divided at points where the rate of change of the route shape with respect to the distance traveled becomes constant.

3. A mobile object control system according to claim 1, wherein the parameter setting unit generates blocks divided at points where the rate of change of the control target with respect to the travel distance becomes constant.

4. A mobile object control system according to claim 1, wherein the parameter setting unit generates blocks divided at points where the blocks become included in a single past block set.

5. A mobile object control system according to claim 1, characterized in that the parameter setting unit provides a buffer section before and after the boundary of each current block, and generates control parameters that smoothly connect the control parameters generated for the previous and next current blocks in the buffer section.

6. A mobile object control system according to claim 1, characterized in that the parameter setting unit calculates an evaluation index indicating the degree of achievement of a control goal, and generates a past block set that brings together past blocks that have the same evaluation index with the same control parameters.

7. A mobile body control system as claimed in claim 1, further comprising: a self-position calculation unit which calculates a self-position of the mobile body from the detection results of sensors which detect state quantities of actuators provided on the mobile body; and a mobile body control unit which controls the behavior of the mobile body on the movement route set by the movement destination setting unit, wherein the mobile body control unit comprises: an evaluation function setting unit which constructs an evaluation function defined by one or more functions expressed using the state quantities of the mobile body and command values ​​to the actuators and weight parameters corresponding to those functions; and a prediction control unit which has a function of predicting the behavior of the mobile body up to a prediction step number and which calculates a control input at each prediction step so that the value of the evaluation function determined by the evaluation function setting unit is reduced from the previous step, wherein the parameter setting unit is a weight parameter used by the evaluation function setting unit, and the parameter setting unit provides the weight parameter to the evaluation function setting unit based on the current position of the mobile body calculated by the self-position calculation unit and the predicted position of the mobile body calculated by the prediction control unit.

8. A mobile body control system as described in claim 7, wherein the parameter setting unit provides a buffer zone ahead of the current position of the mobile body calculated by the self-position calculation unit, and sets a weighting parameter for the buffer zone that prioritizes the tracking accuracy from the end of the buffer zone to the end of the predicted position of the mobile body calculated by the prediction control unit over the tracking accuracy in the buffer zone.

9. A method for controlling a mobile body, comprising the steps of: determining a destination for the mobile body, and generating a mobile route including a route shape to the destination and control targets; and setting control parameters to be used on the set mobile route, wherein the step of setting the control parameters comprises the steps of: generating past blocks by dividing the route shape and control targets of a driving record that records at least a combination of the route shape, control targets, and control parameters traveled autonomously by the mobile body in the past; generating a past block set formed of similar past blocks to which the same control parameters can be assigned; generating a current block by dividing the route shape and control targets of the mobile route; and comparing the route shape and control targets of the current block with the route shape and control targets of the past block set, and generating control parameters corresponding to the past block set similar to the current block.

Images