JP6962007B2 - Driving control device for autonomous driving trolley, autonomous driving trolley - Google Patents

Description

The present invention relates to a traveling control device for an autonomous traveling vehicle and an autonomous traveling vehicle.

Conventionally, autonomous traveling trolleys that move autonomously in the surrounding environment have been known. For autonomous movement, an environmental map showing an area where an object (hereinafter referred to as an obstacle) exists and an area where an object (hereinafter referred to as an obstacle) exists in the moving space is required. While various environmental map acquisition methods have been devised, in recent years, SLAM (Simultaneus Localization and Mapping) has been attracting attention as a technique for estimating a self-position and creating an environmental map in real time while moving. .. For autonomous traveling trolleys using SLAM, an environmental map is created in advance using the terrain data obtained as a result of distance measurement by a distance sensor, and the environmental map and the distance data at each time point are map-matched during autonomous movement. Thereby, self-position estimation is performed (see, for example, Patent Document 1).

Japanese Unexamined Patent Publication No. 2014-219721 Japanese Unexamined Patent Publication No. 2008-83777

In the conventional autonomous driving trolley, it is required to run the autonomous traveling trolley not only indoors but also outdoors. However, in that case, the following problems can be considered.
First, in a wide environment such as the ground, there is a high possibility that no object exists at the sensing effective distance of the distance sensor. Therefore, the environmental map cannot be created normally, and it becomes difficult to estimate the self-position of the autonomous driving vehicle.
Secondly, even if there are peripheral objects in the effective sensing distance, it may not be possible to measure the distance accurately depending on the situation. For example, when LRF is used for the distance sensor and it is affected by the disturbance caused by strong sunlight.

Therefore, in order to expand the travelable range of the autonomous driving vehicle from indoors to outdoors, it is necessary to estimate the self-position using some kind of sensor other than the distance sensor.
An automatic guided vehicle capable of switching between a GPS receiver and a laser radar indoors and outdoors is known (see Patent Document 2). However, this automatic guided vehicle cannot solve the problem that GPS does not function sufficiently even when traveling outdoors, for example, in the vicinity of a building.

An object of the present invention is to enable an autonomous traveling vehicle to travel seamlessly indoors and outdoors.

Hereinafter, a plurality of aspects will be described as means for solving the problem. These aspects can be arbitrarily combined as needed.

The travel control device according to the seemingly present invention includes a first sensor that two-dimensionally or three-dimensionally acquires the distance from the autonomous driving vehicle to surrounding obstacles, and an absolute position of the autonomous driving vehicle on the ground. It is used for an autonomous driving vehicle that autonomously moves in a moving environment systematically or unplannedly, including a second sensor for acquiring the above and a traveling unit.
The travel control device includes a storage, an input / output device, and an arithmetic unit.
The storage stores a map of the mobile environment.
The input / output device inputs the data of the first sensor and the second sensor, and outputs a travel command to the travel unit.
The arithmetic unit performs the following operations.
◎ Using the data of the first sensor and the map of the moving environment, the self-position of the autonomous traveling trolley on the map of the moving environment is obtained as the first probability distribution.
◎ Using the data of the second sensor, the self-position of the autonomous driving bogie is obtained as the second probability distribution.
◎ A third probability distribution is created by synthesizing the first probability distribution and the second probability distribution.
◎ The peak position of the third probability distribution is determined as the self-position of the autonomous driving vehicle in the moving environment.
◎ Create a running command based on the determined self-position.
In this travel control device, the autonomous traveling trolley travels seamlessly indoors and outdoors. In particular, a third is a combination of the data of the first sensor, the first probability distribution obtained by using the map of the moving environment, and the second probability distribution obtained by using the data of the second sensor. Since the probability distribution of is created and the peak position is determined as the self-position, the most appropriate self-position estimation is possible according to the driving environment. For example, indoors, by emphasizing the first probability distribution, the self-position can be accurately grasped. Further, in an area where there are no obstacles such as buildings nearby outdoors, the self-position can be accurately grasped by emphasizing the second probability distribution. Further, in an area where there is an obstacle such as a building outdoors, the self-position can be accurately grasped by emphasizing the first probability distribution as compared with the case where there is no obstacle.

The arithmetic unit may change the coefficient for synthesizing the first probability distribution and the second probability distribution according to the detection amount of the first sensor.
In this travel control device, the above coefficient is changed according to the detection amount of the first sensor. Therefore, when the detection amount is large, the first probability distribution is strong and the second probability distribution is weak, and the third probability distribution is weak. On the contrary, when the detected amount is small, the first probability distribution can be weakly reflected and the second probability distribution can be strongly reflected in the third probability distribution.

The arithmetic unit may change the coefficient for synthesizing the first probability distribution and the second probability distribution according to the estimated self-position of the autonomous traveling vehicle on the map of the moving environment.
In this travel control device, the above coefficient is changed according to the estimated position of the autonomous driving vehicle, so that the first probability distribution and the second probability distribution are changed to the third probability distribution according to the accurate position and the surrounding conditions. Can be reflected.

The autonomous driving trolley runs seamlessly between the indoor environment and the outdoor environment.
A coefficient may be set that more strongly reflects the second probability distribution when the autonomous vehicle is in an outdoor environment.
In this travel control device, the second probability distribution is more strongly reflected outdoors than indoors, so that the self-position estimation while traveling outdoors becomes more accurate.

The first sensor is a laser range sensor that two-dimensionally acquires the distance from the autonomous driving vehicle to an obstacle, and a map of the moving environment and the first probability distribution are created by SLAM (Simultaneus Localization and Mapping). It may be.
Since SLAM is adopted in this travel control device, the self-position estimation by the first probability distribution becomes more accurate.

The second sensor is a receiver of a satellite positioning system (Navigation Satellite System), and an absolute position obtained by RTK (Real Time Kinematic) positioning may be input to the input / output device.
Since the satellite positioning system is adopted in this travel control device, the self-position estimation by the second probability distribution becomes more accurate.

The autonomous driving vehicle according to another aspect of the present invention autonomously moves in a moving environment in a planned or unplanned manner. The autonomous traveling vehicle has a first sensor, a second sensor, a traveling unit, and the traveling control device described above.
Autonomous trolleys run seamlessly indoors and outdoors. In particular, a third is a combination of the data of the first sensor, the first probability distribution obtained by using the map of the moving environment, and the second probability distribution obtained by using the data of the second sensor. Since the probability distribution of is created and the peak position is determined as the self-position, the most appropriate self-position estimation is possible according to the driving environment. For example, indoors, by emphasizing the first probability distribution, the self-position can be accurately grasped. Further, in an area where there are no obstacles such as buildings nearby outdoors, the self-position can be accurately grasped by emphasizing the second probability distribution. Further, in an area where there is an obstacle such as a building outdoors, the self-position can be accurately grasped by emphasizing the first probability distribution as compared with the case where there is no obstacle.

In the present invention, the autonomous traveling vehicle can travel seamlessly indoors and outdoors.

The schematic block diagram which shows the structure of the autonomous driving bogie of 1st Embodiment. The flowchart which shows the outline of the operation of the autonomous driving bogie. A flowchart showing the operation of the autonomous driving trolley during the reproduction driving mode. Flowchart of self-position estimation operation. The schematic plan view which shows the running state of the autonomous driving bogie. A graph showing the SLAM probability distribution. A graph showing a GNSS probability distribution. The graph which shows the SLAM / GNSS probability distribution. A graph showing the SLAM probability distribution. A graph showing a GNSS probability distribution. The graph which shows the SLAM / GNSS probability distribution.

1. 1. 1st Embodiment (1) Overall configuration of autonomous driving trolley The autonomous driving trolley 1 of this embodiment is self-position estimation by SLAM using a distance sensor and map matching, and GNSS (Global Navigation Satellite System). By combining this with the positioning results of the above, we have realized a technology to expand the travelable range of the autonomous driving vehicle 1 from indoors to outdoors. As a result, the autonomous driving vehicle 1 can seamlessly travel between the indoor environment and the outdoor environment.

Hereinafter, the overall configuration of the autonomous driving vehicle 1 according to the first embodiment of the present invention will be described with reference to FIG. FIG. 1 is a schematic block diagram showing a configuration of an autonomous traveling bogie according to the first embodiment.
The autonomous traveling carriage 1 has a traveling unit 3. The traveling unit 3 is provided in the main body of the autonomous traveling vehicle 1, and drives the autonomous traveling vehicle 1 to travel.

The autonomous driving vehicle 1 has a distance sensor 5. The distance sensor 5 is a sensor for two-dimensionally or three-dimensionally acquiring the distance from the autonomous driving vehicle 1 to the surrounding obstacles. The distance sensor 5 is a front laser range sensor and a rear laser range sensor, which are provided in front of and behind the traveling direction of the autonomous traveling vehicle 1, respectively, to acquire the distance two-dimensionally.
The laser range sensor, for example, irradiates a target object such as an obstacle with a laser beam pulse-oscillated by a laser oscillator, and receives the reflected light reflected from the target object by a laser receiver to reduce the distance to the target object. It is a laser range finder (LRF: Laser Range Finder) to calculate. These can scan the laser beam in a fan shape at a predetermined angle by using a rotating mirror to irradiate the laser beam.
Further, as the distance sensor 5, a stereo camera or a TOF (Time Of Flight) camera can be adopted as the one that acquires the distance three-dimensionally.

The autonomous driving vehicle 1 has a control unit 7 (an example of an arithmetic unit). The control unit 7 includes a CPU (Central Processing Unit), a storage device as storage (RAM (Random Access Memory), a ROM (Read Only Memory), an HDD (Hard Disk Drive), an SSD (Solid State Drive), or the like. It is a computer equipped with (configured) and various interfaces. A part or all of the functions of each component of the control unit 7, which will be described later, may be realized by executing a predetermined program stored in the above-mentioned storage device.
Alternatively, a part or all of the functions of each component of the control unit 7 may be realized by hardware such as a custom IC.

The autonomous traveling vehicle 1 includes an operation unit 9. The operation unit 9 receives an instruction input by the operator's manual operation. The operation unit 9 is a user interface operated by the operator when the autonomous traveling vehicle 1 is manually operated, and includes a throttle for receiving an instruction of a traveling speed, a handle for receiving an instruction of a traveling direction, and the like.
The autonomous driving vehicle 1 has a display unit 10. The display unit 10 displays information on the current traveling situation and various other information, and is composed of a liquid crystal display, an LED lamp, and the like.

The autonomous traveling trolley 1 has a GNSS (Global Navigation Satellite System) receiver 11. From the GNSS receiver 11, the control unit 7 is input with the absolute coordinates (latitude / longitude) of the current location obtained by RTK (Real Time Kinematic) positioning. The absolute coordinates of the current location are stored in the storage unit 27 (described later).

(2) Detailed Configuration of Traveling Unit 3 The traveling unit 3 will be described in detail with reference to FIG. The traveling unit 3 has two traveling wheels 13a and 13b.
The traveling unit 3 further has a pair of motors 15a and 15b, encoders 17a and 17b, and motor driving units 19a and 19b corresponding to the traveling wheels 13a and 13b, respectively. Traveling wheels 13a and 13b are connected to the output rotation shafts of the pair of motors 15a and 15b.

The encoders 17a and 17b are connected to the output rotation shaft of the motor 15a and the output rotation shaft of the motor 15b, respectively, and detect the rotation positions of the motors 15a and 15b, respectively.
The motor drive units 19a and 19b correspond to the motors based on the control amount input from the travel control unit 29 (described later) and the rotation positions of the motors 15a and 15b detected by the corresponding encoders 17a and 17b, respectively. Feedback control is performed on 15a and 15b.

(3) Detailed Configuration of Control Unit 7 The control unit 7 will be described in detail with reference to FIG. The control unit 7 includes an obstacle information conversion unit 21, a teaching data creation unit 23, a SLAM processing unit 25, a storage unit 27, a travel control unit 29, and an I / O port 31 as an input / output device. ..
The obstacle information conversion unit 21 converts the detection signal of the distance sensor 5 into the position information of the obstacle existing around the autonomous driving vehicle 1. For example, the obstacle information conversion unit 21 converts the detection signal into obstacle position information which is a coordinate value on a predetermined coordinate.

The teaching data creation unit 23 creates a running schedule, which is a set of passing point data corresponding to the passing time and the passing time in the teaching running mode.
The SLAM processing unit 25 executes SLAM processing that simultaneously performs position estimation and environment map creation (described later).

The storage unit 27 (an example of storage) is a storage area formed in at least a part of the storage area of the storage device of the control unit 7. The storage unit 27 stores the environmental map restoration data and the environmental map.
The environmental map restoration data is the position information of the obstacle acquired in advance by the distance sensor 5 at a predetermined position in the moving area where the autonomous driving vehicle 1 moves. The environmental map is an environmental map created by the map creation unit 33 at the time of the previous position estimation.

The storage unit 27 further stores the traveling schedule. The traveling schedule indicates a traveling route in which the autonomous traveling vehicle 1 autonomously moves when the reproduction traveling mode is executed. When the reproduction traveling mode is executed, the autonomous traveling vehicle 1 refers to the target position indicated in the traveling schedule and controls the traveling unit 3 so as to reach the target position.
The travel control unit 29 generates a control amount of the motors 15a and 15b based on the input travel command and outputs the control amount to the travel unit 3. The travel control unit 29 is, for example, a motion controller.

The travel command is an instruction input from the operator that is input via the operation unit 9 in the teaching travel mode. On the other hand, in the reproduction running mode, the running command is generated based on the comparison between the position on the environmental map estimated by the SLAM processing unit 25 and the running schedule.
The data of the distance sensor 5 and the GNSS receiver 11 are input to the I / O port 31. Further, the I / O port 31 outputs a controlled amount to the traveling unit 3.

(4) Detailed Configuration of SLAM Processing Unit The SLAM processing unit 25 will be described in detail with reference to FIG.
The SLAM processing unit 25 has a map creation unit 33 and a self-position estimation unit 35.

The map creation unit 33 creates an environmental map including the travel route of the autonomous driving vehicle 1 as a global map based on the position information of the obstacle acquired by the distance sensor 5.
The map creation unit 33 corresponds to the time when the position information of the obstacles located around the autonomous travel vehicle 1 is acquired in the teaching travel mode, and stores it in the storage unit 27 as the environment map restoration data.
In the reproduction travel mode, the map creation unit 33 reads the environment map restoration data corresponding to the time before the current passing point from the storage unit 27, and updates the global map based on the data.

The self-position estimation unit 35 has a function of obtaining the SLAM probability distribution by the following operation. In the following description of (I) to (V), it is assumed that the autonomous driving vehicle 1 has moved from the first position P1 to the second position P2. Also, the probability finally obtained at each position is the "posterior probability", which is used as the "prior probability" at the next position.
(I) When the autonomous traveling vehicle 1 exists at the second position P2, the self-position estimation unit 35 creates a local map (position information of obstacles) using the distance sensor 5 at that position.
(II) The self-position estimation unit 35 is based on dead recognition (movement amount by the traveling unit 3 (calculated from the rotation speeds of the traveling wheels 13a and 13b obtained from the encoders 17a and 17b in this embodiment)). The posterior probability of the first position P1 (= the prior probability of the second position P2) is placed at the estimated position P2'.
(III) The self-position estimation unit 35 uses a plurality of temporary local maps and global as a plurality of temporary local maps at the temporary self-position in which the local map created in (I) is moved around the estimated position P2'. By matching with the map (map matching), the degree of matching of each is calculated. As a result, the likelihood of each temporary local map can be obtained.
(IV) By multiplying the posterior probabilities of the first position P1 in (II) (= prior probabilities of the second position P2) and the likelihoods of (III), the posterior probabilities at each temporary self-position are calculated. NS.
(V) Of the plurality of calculated posterior probabilities, for example, the one having the highest peak is selected as the SLAM probability distribution.

The self-position estimation unit 35 further has a function of obtaining the self-position of the autonomous traveling carriage 1 as a GNSS probability distribution by using the data of the GNSS receiver 11.
Further, the self-position estimation unit 35 creates a SLAM / GNSS probability distribution by synthesizing the SLAM probability distribution and the GNSS probability distribution, and sets the coordinates at which the probability peaks as the self-position of the autonomous traveling vehicle 1. (See below).

(5) Schematic operation of the autonomous traveling vehicle The schematic operation of the autonomous traveling vehicle 1 will be described with reference to FIG. FIG. 2 is a flowchart showing an outline of the operation of the autonomous driving bogie.
In step S1, the control unit 7 determines whether or not the mode has been selected by the operator. Specifically, it is determined that the mode selection has been performed when the instruction input is received by the operation of the operation unit 9 by the operator.
In step S2, the control unit 7 determines whether the selected mode is the teaching driving mode or the reproduction driving mode.

In step S3, the control unit 7 executes the reproduction running mode. The reproduction driving mode is a mode in which the vehicle autonomously travels on a predetermined traveling route, and the control unit 7 estimates the position of the autonomous driving vehicle 1 in the moving region and sets the estimated position of the autonomous driving vehicle 1. , The traveling unit 3 is controlled based on the comparison with the position information shown in the traveling schedule (described later).
In step S4, the control unit 7 executes the teaching running mode. The teaching driving mode is executed to acquire the environment map restoration data and the driving schedule. The control unit 7 controls the traveling unit 3 based on the operation from the operating unit 9 described later while estimating the position of the autonomous traveling vehicle 1 in the moving region.

In step S5, the control unit 7 determines whether or not to end the operation of the autonomous traveling vehicle 1. Specifically, the control unit 7 receives an instruction input indicating that the processing is completed by the operation of the operation unit 9 by the operator, receives an instruction input signal indicating that the processing is completed by the remote controller, or teaches the vehicle. When it is determined that the travel schedule created by the mode has been completed, it is determined that the operation of the autonomous traveling vehicle 1 is terminated.
If it is determined that the operation of the autonomous traveling vehicle 1 is terminated, the process is terminated. On the other hand, if it is determined that the operation of the autonomous traveling vehicle 1 is to be continued, the process returns to step S1.

(6) Reproduction Driving Mode of Autonomous Driving Vehicle The reproduction driving mode of the autonomous driving vehicle will be described with reference to FIG. FIG. 3 is a flowchart showing the operation of the autonomous traveling vehicle during the reproduction traveling mode.
In step S31, the position information of the obstacles existing around the autonomous driving vehicle 1 is acquired. Specifically, the distance sensor 5 irradiates the laser beam and further receives the reflected light reflected from the obstacle. Then, the obstacle information conversion unit 21 converts the detection signal into obstacle position information (for example, coordinate values on predetermined coordinates). The position information of this obstacle is used as a local map acquired at the current position.

In step S32, the self-position estimation unit 35 estimates the current position of the autonomous traveling vehicle 1 (described later).

In step S33, the map creation unit 33 creates an environment map (at the current estimated position) for the next self-position estimation, using the next target arrival point indicated in the travel schedule as the above-mentioned future position. It is stored in the storage unit 27.

In step S34, the traveling unit 3 is controlled based on the determined self-position. Specifically, a travel command created based on a comparison between the current estimated position acquired from the SLAM processing unit 25 and the next target arrival point acquired from the travel schedule indicating the travel route currently performing the reproduction travel. According to this, the travel control unit 29 calculates the control amount of the motors 15a and 15b for moving from the current estimated position to the next target arrival point, and outputs the control amount to the motor drive units 19a and 19b. Therefore, based on the received control amount, the motor drive units 19a and 19b calculate the drive signals for driving the motors 15a and 15b, respectively, and output them to the motors 15a and 15b. As a result, the autonomous traveling vehicle 1 travels from the current estimated position toward the next target arrival point.

In step S35, the control unit 7 determines whether or not the reproduction running mode has ended. For example, when it is determined that the autonomous driving vehicle 1 has passed all the target arrival points indicated in the traveling schedule, it is determined that the reproduction traveling mode has ended.
When it is determined that the reproduction traveling mode is terminated, the control unit 7 stops the control of the traveling unit 3 and ends the reproduction traveling mode. On the other hand, if it is determined that the reproduction running mode is to be continued, the process returns to step S31.
By executing the above-mentioned reproduction traveling mode, the autonomous traveling carriage 1 can move autonomously while reproducing a predetermined traveling schedule stored in the storage unit 27.

The above self-position estimation operation (step S32) will be specifically described with reference to FIG. FIG. 4 is a flowchart of the self-position estimation operation.
In step S41, the self-position estimation unit 35 obtains the SLAM probability distribution using the data of the distance sensor 5 and the map of the moving environment.
In step S42, the self-position estimation unit 35 obtains the GNSS probability distribution using the data of the GNSS receiver 11.
In step S43, the self-position estimation unit 35 determines the ratio of mixing the GNSS probability distribution with the SLAM probability distribution (hereinafter, referred to as “GNSS probability distribution mix coefficient”) (described later).

In step S44, the self-position estimation unit 35 creates a SLAM / GNSS probability distribution by synthesizing the SLAM probability distribution and the GNSS probability distribution using the GNSS probability distribution mix coefficient. The obtained SLAM / GNSS probability distribution is used as a prior probability when the self-position is estimated by SLAM next time.
In step S45, the self-position estimation unit 35 determines the peak position of the SLAM / GNSS probability distribution as the self-position of the autonomous traveling vehicle 1 in the moving environment.

(7) Synthesis of SLAM probability distribution and GNSS probability distribution using GNSS probability distribution coefficient The calculation formula of SLAM / GNSS probability distribution is as follows.
{SLAM probability distribution x (1.0-GNSS probability distribution mix coefficient)} + (GNSS probability distribution x GNSS probability distribution mix coefficient) = SLAM / GNSS probability distribution GNSS probability distribution mix coefficient is 0.0 to 1.0. It is changed in the range.

The GNSS probability distribution mix coefficient is set large when it is desired to strongly reflect the GNSS probability distribution, and is set small when it is desired to strongly reflect the SLAM probability distribution.
In the former case, the accuracy of GNSS positioning is high. For example, the autonomous driving vehicle 1 is an open sky (an area that is outdoors and away from a building and has no obstruction such as a roof or trees). ).
The latter case is a case where the accuracy of GNSS positioning is low, for example, a case where the autonomous traveling trolley 1 travels in the building (including underground) and around the building. In particular, GNSS positioning is not possible inside buildings, under roofs and trees, or between buildings.

As the data for determining the GNSS probability distribution mix coefficient, for example, the detection amount (that is, the number of detected data) which is the sensing status of the distance sensor 5 is used. The reason is that the number of detected data decreases in the order of, for example, indoors, around the outdoor building, and open skies.

Hereinafter, a method for determining the GNSS probability distribution mix coefficient and the peak coordinates of the SLAM / GNSS probability distribution in three types of situations will be described.
First, indoors, the sensing condition of the distance sensor 5 is good, that is, the number of detected data of the distance sensor 5 is large, so that the GNSS probability distribution mix coefficient is set to zero or extremely small. This is because, in an indoor environment, since an object exists within the distance measuring range of the distance sensor 5, the reliability of position determination by SLAM is high, and conversely, the reliability of measurement by GNSS is low or positioning is impossible. In this case, the peak coordinates of the SLAM probability distribution become the peak coordinates of the SLAM / GNSS probability distribution as they are.
Secondly, in the case of an open sky environment, the number of detected data of the distance sensor 5 is eliminated, so that the GNSS probability distribution mix coefficient is set large (for example, 1). This is because the measurement by GNSS is highly reliable in an open sky environment. As a result, in this case, the peak coordinates of the GNSS probability distribution become the peak coordinates of the SLAM / GNSS probability distribution as they are.

Thirdly, since the number of detection data of the distance sensor 5 is larger in the vicinity of obstacles such as buildings even outdoors, the GNSS probability distribution mix coefficient is set smaller than in the open sky environment. This is because the reliability of position determination by SLAM is high and the reliability of measurement by GNSS is low compared to the open sky environment (for example, the signal from the artificial satellite cannot be detected and the positioning becomes unstable, or the positioning becomes unstable. It becomes impossible).
As a result of the above, when GNSS positioning does not work well in an environment close to an obstacle such as a building, not only indoors but also outdoors, the self-position estimation by SLAM can be supplemented. As a result, the autonomous traveling trolley 1 can recognize its own position in various environments regardless of whether it is indoors or outdoors, and thereby the traveling range can be expanded.

(8) Example A situation in which the autonomous driving vehicle 1 actually travels outdoors will be described with reference to FIG. FIG. 5 is a schematic plan view showing a traveling state of the autonomous traveling vehicle.
As shown in FIG. 5, the autonomous driving trolley 1 travels from the first region A, which is an open sky environment with no surrounding buildings, to the second region B, which is a building peripheral region near the buildings C and D. do. The first area A includes a flower bed E and a tennis court F.

The SLAM / GNSS probability distribution and the self-position estimation operation in the first region A will be described with reference to FIGS. 6A to 6C. FIG. 6A is a graph showing the SLAM probability distribution. FIG. 6B is a graph showing the GNSS probability distribution. FIG. 6C is a graph showing the SLAM / GNSS probability distribution.
As shown in FIG. 6A, in the SLAM probability distribution, the peak is near the coordinate "8".
As shown in FIG. 6B, in the GNSS probability distribution, the peak is near the coordinate "5".

As shown in FIG. 6C, in the SLAM / GNSS probability distribution, the peak is in the vicinity of the coordinate "5" as in the GNSS probability distribution. The reason is that the GNSS probability distribution mix coefficient is set large in the first region A, so that the GNSS probability distribution is strongly reflected in the SLAM / GNSS probability distribution.

The SLAM / GNSS probability distribution and the self-position estimation operation in the second region B will be described with reference to FIGS. 7A to 7C. FIG. 7A is a graph of the SLAM probability distribution. FIG. 7B is a graph of the GNSS probability distribution. FIG. 7C is a graph of SLAM / GNSS probability distribution.
As shown in FIG. 7A, in the SLAM probability distribution, the peak is near the coordinate "8".
As shown in FIG. 7B, in the GNSS probability distribution, the peak is near the coordinate "5".

As shown in FIG. 7C, in the SLAM / GNSS probability distribution, the peak is near the coordinate "8" as in the SLAM probability distribution. The reason is that in the second region B, the GNSS probability distribution mix coefficient is set small, so that the GNSS probability distribution is weakly or not reflected in the SLAM / GNSS probability distribution.
As a result, it is estimated that the peak of the SLAM / GNSS probability distribution is the current position of the autonomous driving vehicle 1.

2. Description of the Embodiment The above embodiment can also be described as follows.
The control unit 7 (an example of a travel control device) is a distance sensor 5 (first) that acquires the distance from the autonomous driving vehicle 1 (an example of the autonomous driving vehicle) to the surrounding obstacles two-dimensionally or three-dimensionally. (Example of a sensor), a GNSS receiver 11 (an example of a second sensor) that acquires the absolute position of the autonomous driving vehicle 1 on the ground, and a traveling unit 3 (an example of a traveling unit) to plan a moving environment. It is used for an autonomous driving vehicle 1 that moves autonomously or unplannedly.
The control unit 7 has a storage unit 27, an I / O port 31, and a self-position estimation unit 35.
The storage unit 27 (an example of storage) stores a map of the mobile environment.
The I / O port 31 (an example of an input / output device) receives data from the distance sensor 5 and the GNSS receiver 11, and outputs a travel command to the travel unit 3.

The self-position estimation unit 35 (an example of an arithmetic unit) executes the following operations.
◎ Using the data of the distance sensor 5 and the map of the moving environment, the self-position of the autonomous traveling trolley 1 on the map of the moving environment is obtained as the SLAM probability distribution (an example of the first probability distribution).
(1) Using the data of the GNSS receiver 11, the self-position of the autonomous traveling carriage 1 is obtained as a GNSS probability distribution (an example of the second probability distribution).
◎ The SLAM / GNSS probability distribution (an example of the third probability distribution) is created by synthesizing the SLAM probability distribution and the GNSS probability distribution.
◎ The SLAM / GNSS probability part distribution peak position is determined as the self-position of the autonomous driving vehicle 1 in the moving environment.
◎ Create a running command based on the determined self-position.

In this case, the autonomous traveling trolley 1 travels seamlessly indoors and outdoors. In particular, since the SLAM / GNSS probability distribution is created and the peak position is determined as the self-position, the most appropriate self-position estimation is possible according to the driving environment. Therefore, for example, indoors, by emphasizing the SLAM probability distribution, the self-position can be accurately grasped. Further, in the first region A (an example of an outdoor region where there are no obstacles such as buildings nearby), the self-position can be accurately grasped by emphasizing the GNSS probability distribution. Further, in the second region B (an example of an region where there is an obstacle such as a building outdoors nearby), the self-position can be accurately grasped by emphasizing the SLAM probability distribution.

3. 3. Other Embodiments Although one embodiment of the present invention has been described above, the present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the gist of the invention. In particular, the plurality of embodiments and modifications described herein can be arbitrarily combined as needed.
In the above-described embodiment, the reproduction driving mode has been mainly described, but the present invention can also be applied to the teaching driving mode.
In the above embodiment, LRF is mentioned as an example of a distance sensor, but any sensor that can obtain distance data may be used, for example, a TOF camera or an ultrasonic sensor.
In the above embodiment, the absolute position of the autonomous traveling vehicle on the ground is acquired by the GNSS method, but the method of acquiring the absolute position of the autonomous traveling vehicle may be another method.

In the above embodiment, RTK-GPS (Real Time Kinetics Global Positioning System) was used as the positioning method by GNSS, but GPS (Global Positioning System), QZSS (Quasi-Zenith System) and QZSS (Quasi-Zenith System) You may.
In the above embodiment, the detection amount of the distance sensor 5 is used as the reference data for changing the GNSS probability distribution mix coefficient, but the estimated self-position of the autonomous traveling vehicle 1 on the map of the moving environment or the GNSS positioning result is used. May be. In this case, the SLAM probability distribution and the GNSS probability distribution can be reflected in the SLAM / GNSS probability distribution according to the accurate position and surrounding conditions.

The present invention can be widely applied to a traveling control device for an autonomous traveling vehicle and an autonomous traveling vehicle.

1: Autonomous traveling trolley 3: Traveling unit 5: Distance sensor 7: Control unit 8: Coordinates 9: Operation unit 10: Display unit 11: GNSS receiver 13a: Traveling wheel 13b: Traveling wheel 15a: Motor 15b: Motor 17a: Encoder 17b: Encoder 19a: Motor drive unit 19b: Motor drive unit 21: Obstacle information conversion unit 23: Teaching data creation unit 25: SLAM processing unit 27: Storage unit 29: Travel control unit 31: I / O port 33: Map creation Part 35: Self-position estimation part

Claims (6)

A first sensor that acquires the distance from the autonomous driving vehicle to the surrounding obstacles two-dimensionally or three-dimensionally, a second sensor that acquires the absolute position of the autonomous driving vehicle on the ground, and a traveling unit. In the driving control device of the autonomous driving vehicle that autonomously moves the moving environment systematically or unplannedly, including
Storage that stores the map of the mobile environment and
An input / output device for inputting data from the first sensor and the second sensor and outputting a travel command to the travel unit.
Using the data of the first sensor and the map of the moving environment, the self-position of the autonomous traveling vehicle on the map of the moving environment is obtained as a first probability distribution, and the data of the second sensor is used to obtain the self-position. It determined its own position of the autonomous traveling vehicle as a second probability distribution, wherein the first probability distribution second combines the probability distribution to create a third probability distribution, the third peak of the probability distribution An arithmetic device that determines the position as the self-position of the autonomous traveling vehicle in the moving environment and creates the traveling command based on the determined self-position.
Equipped with a,
The arithmetic unit changes the coefficient for synthesizing the first probability distribution and the second probability distribution according to the estimated self-position of the autonomous traveling vehicle on the map of the moving environment. Control device. The arithmetic unit changes the coefficient for synthesizing the first probability distribution and the second probability distribution according to the detection amount of the first sensor or the detection amount of the second sensor. The driving control device for the autonomous traveling vehicle described in 1. The autonomous driving trolley seamlessly travels between an indoor environment and an outdoor environment.
The travel control device for an autonomous driving vehicle according to claim 1 or 2 , wherein a coefficient is set in which the second probability distribution is more strongly reflected when the autonomous traveling vehicle is in an outdoor environment. The first sensor is a laser range sensor that two-dimensionally acquires the distance from the autonomous driving vehicle to the obstacle, and the map of the moving environment and the first probability distribution are SLAM (Simultaneus Localization and Mapping). The traveling control device for an autonomous driving vehicle according to any one of claims 1 to 3 , which is created by The second sensor is a receiver of a satellite positioning system (Navigation Satellite System), and the absolute position obtained by RTK (Real Time Kinematic) positioning is input to the I / O port, claims 1 to 3. The traveling control device for the autonomous traveling vehicle according to any one of the above. An autonomous trolley that autonomously moves in a moving environment in a planned or unplanned manner.
A first sensor that two-dimensionally or three-dimensionally acquires the distance from the autonomous driving vehicle to surrounding obstacles, and
A second sensor that acquires the absolute position of the autonomous driving vehicle on the ground, and
With the running part
The traveling control device according to any one of claims 1 to 5.
Autonomous traveling trolley equipped with.

Images