KR20150071419A - Method and Apparatus for Path Planning of Unmanned Ground Vehicle in Dynamic Environment - Google Patents

Abstract

The present invention relates to an unmanned autonomous vehicle. The unmanned autonomous vehicle includes: a sensor module (10) which collects the environment information regarding the surrounding terrain of a running vehicle; a location estimation module (20) which estimates the current location around the running vehicle; a driving information module (30) which selects an algorithm appropriate to recognize the environment of the road on which a vehicle currently runs from the various information collected by the vehicle and keep updating and maintaining the latest driving information; a terrain detection processing module (80) which derives the result of the environment recognition process. The present invention selects the quality of service (QoSour) according to the environment recognition result derived from the location information (a) and the quality of service (QoSmax) according to the algorithm appropriateness derived from the terrain detection information (b) as well as the best driving algorithm by sequencially applying the conditions which can be represented as as Δ = QoSmax - QoSour and Δ > a. Accordingly, the present invention can select the optimal algorithm by using the degree of service quality and algorithm bank, thereby updating the environmental awareness performance measure.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an autonomous vehicle,

In particular, the present invention relates to an unmanned autonomous vehicle, and more particularly, it relates to an unmanned autonomous vehicle in which an optimal traveling algorithm is selected from an algorithm bank constructed based on knowledge information of the surrounding environment and a degree of service quality according to a result of processing multiple sensor information, The present invention relates to an autonomous vehicle and an autonomous navigation method based on the dynamic environment.

Generally, in order for an unmanned vehicle to autonomously navigate, it is necessary to determine a travelable area from the recognition and recognition result of the environment to be driven, and to determine a travel route along with a route to be traveled from the result of the determination .

For this purpose, a variety of sensors such as a laser range finder, a camera, and a radar are installed on the unmanned vehicle to collect information on the surrounding environment, and the mounted autonomous drive controller processes and recognizes the collected environment information of the sensor, And the traveling route point is selected.

Korean Patent Publication No. 10-2010-0102815 (September 27, 2010)

The patent document shows an example of an unmanned vehicle control technique in which a traveling route is determined through a landmark recognized from a camera, and self position recognition, steering and speed control are performed.

However, the above-described patent document can not be applied to an environment such as a road where it is difficult to install a separate landmark by recognizing a traveling route through a landmark recognized from a camera.

In particular, due to the environmental characteristics of the field environment, external environmental factors such as the variation of the environment affecting the sensor performance as well as the diversity of the environment are inevitably required.

For this purpose, the unmanned vehicle operated in the field, especially the military-purpose unmanned vehicle, is applied with the multi-sensor so as to perceive the surrounding environment relatively highly, The optimization process is further pursued through an environment recognition procedure that can be supplemented.

However, even if the application of multiple sensors and the convergence of the operating environment and the algorithm are performed, there is a limit to reach the autonomous traveling demand performance.

The present invention has been made in view of the above-mentioned problems, and it is an object of the present invention to provide a method and apparatus for providing a service to an algorithm bank, which is constructed for each sensor based on running information generated using knowledge information about a base surrounding environment, By using the quality measure (Degree of Service Quality), the optimal operating sensor combination and operating sensor-specific processing algorithms are selected, and in particular, the convergence of the selection information updates the environment recognition performance metric of the driving information map, And an object of the present invention is to provide an autonomous vehicle having improved performance and an autonomous driving method based on the dynamic environment.

According to another aspect of the present invention, there is provided an autonomous vehicle including: a sensor module for generating environmental data by collecting environmental information about a surrounding terrain during operation; A position estimation module for generating position data from current position estimation during running; A plurality of different algorithms for driving are provided, and an algorithm selected from the environment data and the position data to suit the environment recognition of the current traveling path is applied to the driving, and the driving information A module; A terrain sensing processing module for deriving an environment recognition result from the resultant performance; Is included.

The sensor module includes a laser range finder for collecting distance information about the surrounding terrain during operation, a camera for collecting image information about the surrounding terrain during operation, a radar for collecting object information on the dangerous elements from the surrounding terrain, And an environmental sensor for collecting temperature, illuminance and humidity information related to the surrounding terrain while driving.

A QoS (Quality of Service) determination block in which QoS information (Quality of Service) which is the maximum QoS among the plurality of algorithms is selected from the environment data; A travel information selection block for searching travel information included in the travel route from the location data and generating a travel information search result; (Quality of Service) of the algorithm being applied to the running is calculated from the environment recognition using the driving information retrieval result and the driving environment recognition is performed from the QoSmax (Quality of Service) and the QoSour An algorithm selection block for generating an information update result for re-selection of an appropriate algorithm; A travel information update block for keeping the travel route information updated from the information update result; .

The quality of service (QoS) determination block includes a distance-based decision unit for processing distance information among the environment data, an image-based decision unit for processing image information of the environment data, Based decision unit.

Wherein the algorithm selection block includes a distance-based selection unit that selects an algorithm from the distance information processing result among the environment data, an image-based selection unit that selects an algorithm from the image information processing result among the environment data, A radar-based selection unit for selecting an algorithm from the result, and a fusion weight selection unit for weighting the processing result of the environment data.

Wherein the terrain detection processing module comprises: a distance-based recognition unit for deriving an environment recognition result from the processing result of the algorithm according to the distance information processing of the environment data; Based recognition unit for deriving an environment recognition result from a processing result of an algorithm according to the object information processing of the environmental data; And a fusion recognition section for deriving recognition results.

In order to achieve the above object, according to the present invention, there is provided an autonomous navigation system based on dynamic environment of an autonomous vehicle, wherein position information about a periphery of the autonomous vehicle is recognized, An environment recognition mode in which QoSour (Quality of Service) of an algorithm is calculated; The method of claim 1, wherein the terrain sensing information for the periphery of the unmanned autonomous vehicle is recognized, the QoS (Quality of Service) for a plurality of algorithms is compared from the terrain sensing information, Detection mode; An optimal algorithm suitable for recognition of the current driving environment among the plurality of algorithms is selected from the QoS a and a QoS a and the selected algorithm is recorded and stored while being applied to the running of the autonomous vehicle Algorithm update mode; As shown in FIG.

The terrain sensing mode is performed during the environment recognition mode.

In the environment recognition mode, the calculation of the QoS (Quality of Service) may include a DB search step of searching a travel information database based on the location information; An environment recognition step of deriving an environment recognition result from a predetermined algorithm based on a result of the search from the database, and selecting the quality of service (QoS) as a result of the environment recognition; Lt; / RTI >

The environment recognition result is derived using an initial setting value in the case where there is no previous running result.

In the terrain sensing mode, the selection of the quality of service (QoS) may include: a QoS calculation step of calculating a quality of service (QoS) of an appropriate algorithm from the terrain sensing information; A QoS selection step of selecting one QoSmax (Quality of Service) having a maximum value among the QoSs when the QoS is calculated for each of a plurality of algorithms; Lt; / RTI >

In the algorithm update mode, the optimal algorithm selection is performed such that a difference value is determined as a result of comparing the Quality of Service (QoS) and the Quality of Service (QoS), and the difference value is compared with a setting reference value And is then obtained from the above comparison processing result.

The difference value is obtained from Quality of Service (QoS) - QoSour (Quality of Service), and the comparison result is obtained from the difference reference value.

The optimal algorithm is selected when the difference value> the set reference value is satisfied, and the currently applied algorithm is continuously maintained when the difference value> the set reference value is not satisfied.

In the algorithm update mode, the information to be recorded and stored is the optimal algorithm and the QoS information at the selected time, temperature, illumination and humidity.

In the present invention, an optimal driving algorithm is selected from an algorithm bank constructed from knowledge information of the surrounding environment and a service quality scale according to a result of processing a plurality of sensor information, There is an effect that the autonomous driving performance of the unmanned vehicle is improved by updating the environment recognition performance measure.

In addition, the present invention has the effect that the driving information database is updated and updated information by updating the environment recognition performance measure of the driving information map based on the surrounding environment.

In addition, the present invention maintains the expansion and update of the travel information database so as to select the optimal algorithm based on the service quality of the sensor based on the environmental factors affecting the sensors collecting various environmental and environmental information, It is possible to greatly improve the autonomous driving performance of the military unmanned vehicle repeatedly operated in the environment of the vehicle.

FIG. 1 is a block diagram of an autonomous vehicle according to the present invention, FIG. 2 is an operation flow for autonomous driving based on a dynamic environment of an autonomous vehicle according to the present invention, FIG. This is a conceptual diagram of time-sharing QoS calculation for maintaining real-time performance in autonomous vehicle based on dynamic environment.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings, which illustrate exemplary embodiments of the present invention. The present invention is not limited to these embodiments.

FIG. 1 shows a block configuration of an autonomous vehicle according to the present invention, according to the present invention.

As shown in the figure, the autonomous vehicle includes a sensor module 10 for collecting environmental information about the surrounding terrain during running, a position estimation module 20 for estimating the current position during running, A travel information module 30 that selects an algorithm suitable for the environment recognition of the vehicle and maintains and maintains the travel information, and a terrain sensing processing module 80 that derives the processed environment recognition result.

The sensor module 10 is configured to collect environmental information about surrounding terrain in at least one or more ways.

For example, the sensor module 10 may include a laser range finder 11 installed in the autonomous vehicle and capable of collecting distance information about the surrounding terrain by emitting laser beams in the front and rear directions or in the left and right directions.

The sensor module 10 may include a camera 12 installed in the autonomous vehicle and capable of collecting image information about the surrounding terrain during the running by photographing the images in the forward and backward directions and the left and right direction.

The sensor module 10 may include a radar 13 installed in the autonomous vehicle and capable of collecting object information on a dangerous element from the surrounding terrain by emitting radio waves in the front and rear directions or in the left and right directions.

The sensor module 10 may include an environmental sensor 14 installed in the autonomous vehicle and capable of collecting temperature, illuminance and humidity information related to the surrounding terrain during traveling.

In addition, the position estimation module 20 may include a GPS so that the current position of the unmanned autonomous vehicle installed in the autonomous vehicle may be grasped.

The travel information module 30 includes a quality of service (QoS) determination block 40 for determining QoS (Quality of Service) for each algorithm that can be utilized in the collected data processing, A traveling information selecting block 50 for searching and providing traveling information included in a traveling route to which the autonomous vehicle belongs; an algorithm selecting block 60 for selecting an algorithm suitable for the traveling environment recognition; And a travel information updating block 70 for keeping the travel route information updated based on the result.

The quality of service (QoS) decision block 40 is configured to be associated with the components of the sensor module 10 by providing all the information collected by the sensor module 10.

For example, the quality-of-service (QoS) decision block 40 is provided with a distance-based decision unit 41. The distance-based decision unit 41 is associated with the laser distance measurer 11, ) Are processed.

The quality-of-service (QoS) determination block 40 includes an image-based determination unit 42. The image-based determination unit 42 associates the camera 12 with the camera 12, .

The quality of service (QoS) decision block 40 is provided with a radar-based decision unit 43. The radar-based decision unit 43 is associated with the radar 13, .

The traveling information selection block 50 is configured to be linked to the position estimation module 20 by providing all the data collected by the position estimation module 20. In the traveling information selection block 50, the traveling information database is searched based on the data collected by the position estimation module 20, and the traveling information data included in the traveling route to which the unmanned autonomous vehicle to provide.

In the algorithm selection block 60, all the data processed in the QoS (Quality of Service) decision block 40 is provided, thereby linking with the QoS (quality of service) decision block 40, 50 and the navigation data selection block 50 are provided.

For example, the algorithm selection block 60 is provided with a distance-based preference unit 61, and the distance-based preference unit 61 is associated with the distance-based decision unit 41, All the data is used.

The algorithm selection block 60 is provided with an image-based selection unit 62. The image-based selection unit 62 is associated with the image-based determination unit 42 so that all data processed by the image- Is used.

Based selection section 63 is provided in the algorithm selection block 60 and the radar based selection section 63 is associated with the radar based decision section 43 so that all the data processed in the radar based decision section 43 Is used.

In particular, the algorithm selection block 60 is provided with a convergence weight selection section 64, and the convergence weight selection section 64 assigns necessary weights to the processing results, so that an optimal algorithm suitable for recognition of the travel environment can be selected Let's do it.

The traveling information updating block 70 is configured to be linked to the algorithm selecting block 60 so that all data processed in the algorithm selecting block 60 is received and the traveling route information on which the unmanned vehicle travels is updated to the latest Information can be updated and maintained.

The terrain sensing processing module 80 is configured in association with the algorithm selection block 60 by providing all the information processed in the algorithm selection block 60.

The terrain sensing processing module 80 includes a distance based recognition unit 81, an image based recognition unit 82, a radar based recognition unit 83, and a fusion recognition unit 84.

The distance-based recognition unit 81 processes the environment recognition based on the distance from the information provided after the processing in the algorithm selection block 60, and the image-based recognition unit 82 processes the distance- Based on the information provided, the environment recognition based on the image is processed. The radar-based recognition unit 83 processes the environment recognition based on the radar from the information provided after being processed in the algorithm selection block 60, Is processed in the algorithm selection block 60 and the environment recognition based on fusion is processed from the provided information.

FIG. 2 shows an operation flow for autonomous driving based on a dynamic environment of the autonomous vehicle according to the present embodiment.

S10 is a process in which information is recognized for autonomous driving control based on a real-time dynamic environment of an unmanned autonomous vehicle under running, and the information is divided into position information (a) and terrain sensing information (b). The terrain sensing information b is provided in the sensor module 10 of Fig. 1, and the position information a is provided in the position estimation module 20 of Fig.

 In the process of the location information a, one QoS (Quality of Service) is selected from the algorithm as a result of the environment recognition by passing through S20 to S22, and in the processing of the terrain sensing information b, S30 to S33 And another QoS (Quality of Service) is selected as a result of the algorithm processing.

In this embodiment, the processing of the position information (a) is performed first and the processing of the terrain sensing information (b) is performed during the processing of the position information (a).

This method is illustrated in FIG. 3 as a concept that real-time property maintenance is performed when a time-divisional QoS (Quality of Service) is calculated. In particular, when a self-driving vehicle runs an autonomous vehicle, Assignment is illustrated.

As shown in the figure, based on a predetermined algorithm, the environment recognition is performed at a predetermined time in real time, and the real-time performance of the environment recognition is not hindered by sequential QoS calculation for each algorithm constituting the algorithm bank with the spare time .

This is because the moving distance of the unmanned vehicle that can be moved during the unit time is not large, and the change of the environment does not occur rapidly.

Therefore, the QoS is calculated up to the nth algorithm through sequential execution, and then an algorithm suitable for recognition of the traveling environment based on the QoS for each algorithm constituting the algorithm bank in the algorithm selection section can be selected according to the predetermined rule.

The process of the position information (a) is performed as follows.

When the driving information database is searched in S20 based on the position information (a), the environment recognition is performed using the algorithm selected based on the result of the driving information database search result in S21, and the environment recognition is performed in S22 (Quality of Service) is selected.

In the present embodiment, the environment recognition is performed using the initial set values in S21 if there is no previous running result, and the selected one QoS (Quality of Service) is defined as QoSour in S22.

The processing of the terrain sensing information (b) is performed as follows.

After the QoS (Quality of Service) of the algorithm based on the terrain sensing information b is calculated in S30 and the QoS (Quality of Service) of each algorithm is calculated in S31, the respective Quality of Service One QoS (Quality of Service) having a maximum value is selected.

In this embodiment, a selected QoS (Quality of Service) is defined as QoSmax in S32.

Meanwhile, in S40, the QoS difference between the currently executed algorithm and the selected QoS information is calculated based on the QoS information calculated based on the location information (a) and the selected QoS information based on the terrain information (b).

This calculation is judged as a difference (Δ) between QoSmax and QoSour, and is defined as Δ = QoSmax - QoSour.

S50 is a process in which the calculated difference? Is compared with the setting criterion (a) so as to determine whether to change the algorithm to be applied to the autonomous vehicle. For this,?>? Is applied, and? Is applied according to the specification of the autonomous vehicle, so that a specific limitation is not exemplified.

S60 is the time when Δ> α is satisfied as a result of the check in S50. In this case, the current running algorithm is replaced with the algorithm selected by QoSmax, so that the unmanned autonomous vehicle travels in a state optimally adapted to the surrounding environment .

S70 is a process of updating QoS (time, DOP) and environment information (temperature, illuminance, humidity) at that time along with the new traveling algorithm selected in S60. The QoS (time, DOP) 20), and the environmental information (temperature, illuminance, humidity) is obtained from the environmental sensor 14 of the sensor module 10 of Fig.

On the other hand, S100 does not satisfy Δ> α as a result of the check in S50. In this case, the current running algorithm is maintained.

As described above, the autonomous vehicle according to the present embodiment includes the sensor module 10 for collecting environmental information about the surrounding terrain during running, the position estimation module 20 for estimating the current position during running, A traveling information module 30 for selecting an algorithm suitable for the environment recognition of the current traveling route from the information and keeping the traveling information up to date, and a terrain sensing processing module 80 for deriving the processed environment recognition result, From the location information (a), we select QoSour (Quality of Service) according to the environment recognition result and QoSmax (Quality of Service) according to the algorithm fitness result from the terrain sensing information (b) The optimum algorithm is selected from the algorithm bank and the degree of service quality by sequentially applying the condition and the Δ> a condition, Hi there from which it can be renewed environmental awareness performance measure.

10: Sensor module 11: Laser range finder
12: camera 13: radar
14: Environmental sensor
20: position estimation module 30: travel information module
40: Quality of Service (QoS) decision block
41: distance-based determination unit 42:
43: Radar-based determination unit 50: Driving information selection block
60: algorithm selection block 61: distance-based selection
62: image-based preference 63: radar-based preference
64: Convergence weight selection section 70: Travel information update block
80: terrain detection processing module 81: distance-based recognition unit
82: image-based recognition unit 83: radar-based recognition unit
84: fusion recognition section

Claims (16)

A sensor module for generating environmental data by collecting environment information about the surrounding terrain while driving;
A position estimation module for generating position data from current position estimation during running;
A plurality of different algorithms for driving are provided, and an algorithm selected from the environment data and the position data to suit the environment recognition of the current traveling path is applied to the driving, and the driving information A module;
A terrain sensing processing module for deriving an environment recognition result from the resultant performance;
And an unmanned vehicle.
[Claim 2] The method according to claim 1, wherein the sensor module comprises: a laser distance measurer for collecting distance information about the surrounding terrain during running; a camera for collecting image information about the surrounding terrain in operation; And an environment sensor for collecting temperature, illuminance and humidity information related to the surrounding terrain while the vehicle is running.

The method of claim 1, wherein the travel information module includes a Quality of Service (QoS) determination block in which QoSmax (Quality of Service) that is the maximum of QoS (Quality of Service) of the plurality of algorithms is selected from the environment data;
A travel information selection block for searching travel information included in the travel route from the location data and generating a travel information search result;
(Quality of Service) of the algorithm being applied to the running is calculated from the environment recognition using the driving information retrieval result and the driving environment recognition is performed from the QoSmax (Quality of Service) and the QoSour An algorithm selection block for generating an information update result for re-selection of an appropriate algorithm;
A travel information update block for keeping the travel route information updated from the information update result;
Wherein the vehicle is an unmanned autonomous vehicle.
4. The apparatus of claim 3, wherein the Quality of Service (QoS) determination block comprises: a distance based decision unit for processing distance information of the environment data; an image based decision unit for processing image information of the environment data; And a radar-based determination unit for processing information.
The image processing apparatus according to claim 3, wherein the algorithm selection block includes: a distance-based selection unit that selects an algorithm from the distance information processing result among the environment data; an image-based selection unit that selects an algorithm from the image information processing result among the environment data; A radar-based selection unit for selecting an algorithm from the object information processing result of the data, and a fusion weight selection unit for assigning a weight to the processing result of the environment data.
The method according to claim 1, wherein the terrain detection processing module comprises: a distance-based recognition unit for deriving an environment recognition result from the processing result of the algorithm according to the distance information processing of the environment data; An image recognition unit for deriving an environment recognition result from the processing result; a radar-based recognition unit for deriving an environment recognition result from the processing result of the algorithm according to the object information processing of the environmental data; And a fusion recognition unit for deriving an environment recognition result from the weighting result.
An environment recognition mode in which position information about the periphery of the autonomous vehicle is being recognized and QoSour (Quality of Service) of the algorithm being applied to the travel is calculated from the environment recognition using the position information;
The method of claim 1, wherein the terrain sensing information for the periphery of the unmanned autonomous vehicle is recognized, the QoS (Quality of Service) for a plurality of algorithms is compared from the terrain sensing information, Detection mode;
An optimal algorithm suitable for recognition of the current driving environment among the plurality of algorithms is selected from the QoS a and a QoS a and the selected algorithm is recorded and stored while being applied to the running of the autonomous vehicle Algorithm update mode;
Wherein the step of performing the autonomous navigation is based on the dynamic environment of the autonomous vehicle.
8. The method of claim 7, wherein the terrain sensing mode is performed in the course of performing the environment recognition mode.
8. The method of claim 7, wherein, in the environment recognition mode, the calculation of the QoS (Quality of Service) includes a DB search step of searching a travel information database based on the location information;
An environment recognition step of deriving an environment recognition result from a predetermined algorithm based on a result of the search from the database, and selecting the quality of service (QoS) as a result of the environment recognition; Wherein the step of performing the autonomous navigation is based on the dynamic environment of the autonomous vehicle.
[12] The method of claim 9, wherein the environment recognition result is derived using an initial set value in the absence of the preliminary running result.
8. The method according to claim 7, wherein in the terrain sensing mode, the selection of the quality of service (QoS) is performed by calculating a quality of service (QoS) of an appropriate algorithm from the terrain sensing information;
A QoS selection step of selecting one QoSmax (Quality of Service) having a maximum value among the QoSs when the QoS is calculated for each of a plurality of algorithms;
Wherein the step of performing the autonomous navigation is based on the dynamic environment of the autonomous vehicle.

8. The method of claim 7, wherein in the algorithm update mode, the optimal algorithm selection is such that a difference value is determined as a result of comparing the Quality of Service (QoS) and Quality of Service (QoS) And the comparison result is obtained from the result of the comparison process.
13. The method of claim 12, wherein the difference value is obtained from a Quality of Service (QoS) - QoSour (Quality of Service).
13. The method of claim 12, wherein the comparison result is obtained from the set reference value.
15. The method of claim 14, wherein the optimal algorithm is selected when the difference value > the set reference value is satisfied, and the currently applied algorithm is maintained when the difference value > Autonomous driving method based on dynamic environment of vehicle.
The method of claim 7, wherein in the algorithm update mode, the information to be recorded and stored is the optimal algorithm, QoS information at a selected time point, temperature, illuminance and humidity. .

Images