WO2022091305A1 - 挙動推定装置、挙動推定方法、経路生成装置、経路生成方法、及びコンピュータ読み取り可能な記録媒体 - Google Patents
挙動推定装置、挙動推定方法、経路生成装置、経路生成方法、及びコンピュータ読み取り可能な記録媒体 Download PDFInfo
- Publication number
- WO2022091305A1 WO2022091305A1 PCT/JP2020/040670 JP2020040670W WO2022091305A1 WO 2022091305 A1 WO2022091305 A1 WO 2022091305A1 JP 2020040670 W JP2020040670 W JP 2020040670W WO 2022091305 A1 WO2022091305 A1 WO 2022091305A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- behavior
- environment
- model
- analysis data
- data
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims description 62
- 230000006399 behavior Effects 0.000 claims abstract description 402
- 238000004458 analytical method Methods 0.000 claims abstract description 182
- 238000003891 environmental analysis Methods 0.000 claims abstract description 76
- 230000007613 environmental effect Effects 0.000 claims abstract description 30
- 230000001172 regenerating effect Effects 0.000 claims 1
- 238000010586 diagram Methods 0.000 description 28
- 238000012545 processing Methods 0.000 description 16
- 230000006870 function Effects 0.000 description 9
- 239000000446 fuel Substances 0.000 description 6
- 238000004891 communication Methods 0.000 description 5
- 238000005259 measurement Methods 0.000 description 5
- 238000010276 construction Methods 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 238000012417 linear regression Methods 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 239000006185 dispersion Substances 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 239000011435 rock Substances 0.000 description 2
- 239000004065 semiconductor Substances 0.000 description 2
- 239000002689 soil Substances 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 238000012876 topography Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 241000132092 Aster Species 0.000 description 1
- 241000196324 Embryophyta Species 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 239000004927 clay Substances 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 239000000725 suspension Substances 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/10—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration
- G01C21/12—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning
- G01C21/16—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning by integrating acceleration or speed, i.e. inertial navigation
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W40/00—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models
- B60W40/02—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models related to ambient conditions
- B60W40/06—Road conditions
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W40/00—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models
- B60W40/10—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models related to vehicle motion
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/20—Instruments for performing navigational calculations
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/38—Electronic maps specially adapted for navigation; Updating thereof
- G01C21/3804—Creation or updating of map data
- G01C21/3807—Creation or updating of map data characterised by the type of data
- G01C21/3826—Terrain data
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/20—Control system inputs
- G05D1/24—Arrangements for determining position or orientation
- G05D1/246—Arrangements for determining position or orientation using environment maps, e.g. simultaneous localisation and mapping [SLAM]
- G05D1/2464—Arrangements for determining position or orientation using environment maps, e.g. simultaneous localisation and mapping [SLAM] using an occupancy grid
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/60—Intended control result
- G05D1/644—Optimisation of travel parameters, e.g. of energy consumption, journey time or distance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D2101/00—Details of software or hardware architectures used for the control of position
- G05D2101/10—Details of software or hardware architectures used for the control of position using artificial intelligence [AI] techniques
- G05D2101/15—Details of software or hardware architectures used for the control of position using artificial intelligence [AI] techniques using machine learning, e.g. neural networks
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D2107/00—Specific environments of the controlled vehicles
- G05D2107/30—Off-road
- G05D2107/36—Catastrophic areas
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D2109/00—Types of controlled vehicles
- G05D2109/10—Land vehicles
Definitions
- the present invention relates to a behavior estimation device, a behavior estimation method, a route generation device, and a route generation method used for estimating the behavior of a moving object, and further, a computer-readable program for realizing these is recorded. Regarding recording media.
- Patent Document 1 discloses a behavior prediction device for determining the necessity of updating the behavior prediction model database. According to the behavior prediction device, when it is determined that there is a discrepancy between the actual behavior of the moving object existing around the work vehicle (own vehicle) and the behavior of the moving object predicted by the behavior prediction model. , Estimate the reason why the divergence occurred, and update the behavior prediction model database based on the estimated reason.
- the behavior prediction device disclosed in Patent Document 1 is a device that predicts the behavior of a moving body existing around the work vehicle (own vehicle), the behavior of the work vehicle in an unknown environment cannot be estimated.
- behavior estimators As one aspect, behavior estimators, behavior estimation methods, route generators, route generation methods, and computer readable, which improve the operational efficiency of moving objects by reducing the number of times the model is relearned in an unknown environment.
- the purpose is to provide a flexible recording medium.
- the behavior estimation device in one aspect is The behavior analysis unit that generates the first behavior analysis data that represents the actual behavior of the moving object in the first environment, An environmental analysis unit that analyzes the first environment based on the environmental state data representing the state of the first environment and generates environmental analysis data.
- An estimation unit that inputs the environmental analysis data into a model for estimating the behavior of the moving body in the first environment and estimates the behavior of the moving body in the first environment.
- a confidence interval is set based on the behavior estimation result data estimated by the model, and if the first behavior analysis data exists in the set confidence interval, the model is re-learned in the learning unit for learning the model.
- a learning instruction unit that gives instructions for learning, It is characterized by having.
- the route generation device in one aspect is The behavior analysis unit that generates the first behavior analysis data that represents the actual behavior of the moving object in the first environment, An environmental analysis unit that analyzes the first environment based on the environmental state data representing the state of the first environment and generates environmental analysis data.
- An estimation unit that inputs the environmental analysis data into a model for estimating the behavior of the moving body in the first environment and estimates the behavior of the moving body in the first environment.
- a confidence interval is set based on the behavior estimation result data estimated by the model, and when the first behavior analysis data exists in the set confidence interval, the model is re-learned as a learning means for learning the model.
- a learning instruction unit that gives instructions for learning, When the model is re-learned, the movement route data representing the movement route from the current position to the destination is regenerated based on the behavior estimation result data generated by using the re-learned model.
- the generator and It is characterized by having.
- the behavior estimation method in one aspect is A behavior analysis step that generates the first behavior analysis data that represents the actual behavior of the moving object in the first environment.
- An environmental analysis step that analyzes the first environment based on the environmental state data representing the state of the first environment and generates environmental analysis data.
- a confidence interval is set based on the behavior estimation result data estimated by the model, and if the first behavior analysis data exists in the set confidence interval, the model is re-learned in the learning unit for learning the model.
- a learning instruction step that gives instructions to learn, It is characterized by having.
- the route generation method in one aspect is A behavior analysis step that generates the first behavior analysis data that represents the actual behavior of the moving object in the first environment.
- An environmental analysis step that analyzes the first environment based on the environmental state data representing the state of the first environment and generates environmental analysis data.
- An estimation step of inputting the environmental analysis data into a model for estimating the behavior of the moving body in the first environment and estimating the behavior of the moving body in the first environment.
- a confidence interval is set based on the behavior estimation result data estimated by the model, and when the first behavior analysis data exists in the set confidence interval, the model is re-learned as a learning means for learning the model.
- a learning instruction step that gives instructions to learn, When the model is re-learned, the movement route data representing the movement route from the current position to the destination is regenerated based on the behavior estimation result data generated by using the re-learned model.
- Generation step and It is characterized by having.
- a computer-readable recording medium on which a program according to one aspect of the present invention is recorded may be used.
- a behavior analysis step that generates the first behavior analysis data that represents the actual behavior of the moving object in the first environment.
- An environmental analysis step that analyzes the first environment based on the environmental state data representing the state of the first environment and generates environmental analysis data.
- An estimation step of inputting the environmental analysis data into a model for estimating the behavior of the moving body in the first environment and estimating the behavior of the moving body in the first environment.
- a confidence interval is set based on the behavior estimation result data estimated by the model, and if the first behavior analysis data exists in the set confidence interval, the model is re-learned in the learning unit for learning the model.
- a learning instruction step that gives instructions to learn, It is characterized by recording a program containing an instruction to execute.
- a computer-readable recording medium on which a program in one aspect of the present invention is recorded may be used.
- a behavior analysis step that generates the first behavior analysis data that represents the actual behavior of the moving object in the first environment.
- An environmental analysis step that analyzes the first environment based on the environmental state data representing the state of the first environment and generates environmental analysis data.
- a confidence interval is set based on the behavior estimation result data estimated by the model, and when the first behavior analysis data exists in the set confidence interval, the model is re-learned as a learning means for learning the model.
- a learning instruction step that gives instructions to learn, When the model is re-learned, the movement route data representing the movement route from the current position to the destination is regenerated based on the behavior estimation result data generated by using the re-learned model.
- Generation step and It is characterized by recording a program containing an instruction to execute.
- FIG. 1 is a diagram for explaining the relationship between the tilt angle and the slip in an unknown environment.
- FIG. 2 is a diagram for explaining the estimation of slip on a steep slope in an unknown environment.
- FIG. 3 is a diagram for explaining an example of the behavior estimation device.
- FIG. 4 is a diagram for explaining the re-learning of the model.
- FIG. 5 is a diagram for explaining an example of a system having a behavior estimation device.
- FIG. 6 is a diagram for explaining the generation of movement route data.
- FIG. 7 is a diagram for explaining an example of information regarding the topographical shape.
- FIG. 8 is a diagram for explaining the relationship between the grid and the slip.
- FIG. 9 is a diagram for explaining the relationship between the grid and passable / impossible.
- FIG. 10 is a diagram for explaining an example of a movement route.
- FIG. 11 is a diagram for explaining an example of a movement route.
- FIG. 12 is a diagram for explaining an example of the operation of the behavior estimation device.
- FIG. 13 is a diagram for explaining an example of the operation of the route generation device.
- FIG. 14 is a block diagram showing an example of a computer that realizes a system having a behavior estimation device or a route generation device.
- autonomous work vehicles that work in unknown environments such as disaster areas, construction sites, forests, and planets have acquired image data that captures the unknown environment from the image pickup device mounted on the work vehicle.
- Image processing is performed on the image data, and the state of the unknown environment is estimated based on the result of the image processing.
- the state of the unknown environment is, for example, an environment in which the topography, the type of the ground, the state of the ground, etc. are unknown.
- the type of ground is, for example, the type of soil classified according to the content ratio of leki, sand, clay, silt, and the like. Further, the type of ground may include the ground where plants are growing, the ground such as concrete and rock, and the ground where obstacles are present.
- the state of the ground is, for example, the water content of the ground, the looseness (or hardness) of the ground, the stratum, and the like.
- the training data lacks image data of unknown environments and data on terrain that is at high risk for work vehicles such as steep slopes and puddles. Therefore, the learning of the model becomes insufficient. Therefore, it is difficult to accurately estimate the running of the work vehicle even if a model with insufficient learning is used.
- the model is learned using the behavior analysis data generated in an unknown environment and the behavior analysis data generated for each environment in which the vehicle has traveled in the past. Then, by inputting the environmental analysis data obtained by analyzing the state of the unknown environment into the generated model and estimating the behavior of the work vehicle in the unknown environment, the behavior of the work vehicle in the unknown environment is estimated accurately. Proposals have been made.
- the model is relearned every time the work vehicle acquires the behavior analysis data, so that the work vehicle cannot be operated efficiently. Specifically, when the work vehicle acquires the behavior analysis data while the work vehicle is running or the work vehicle is working, the accuracy of the behavior estimation is improved and the safety of the work vehicle is improved. To ensure, the work vehicle must stop running or work and relearn the model.
- the inventor has found a problem that the operational efficiency of the work vehicle is lowered if the behavior of the work vehicle is estimated accurately in an unknown environment by the method as described above. At the same time, we have come up with a means to solve the problem.
- the inventor has come to derive a means for reducing the number of times the model is relearned in an unknown environment. As a result, the behavior of a moving object such as a work vehicle can be estimated accurately, and further, a decrease in operational efficiency of the work vehicle can be suppressed.
- FIG. 1 is a diagram for explaining the relationship between the tilt angle and the slip in an unknown environment.
- FIG. 2 is a diagram for explaining the estimation of slip on a steep slope in an unknown environment.
- the work vehicle 1 which is a moving body shown in FIG. 1, acquires moving body state data representing the state of the moving body from a sensor that measures the state of the working vehicle 1 while traveling in an unknown environment, and the acquired movement.
- the physical condition data is stored in a storage device provided inside or outside the work vehicle 1.
- the work vehicle 1 analyzes the moving body state data acquired from the sensor on a low slope where the risk is low in an unknown environment, and performs a behavior analysis showing the relationship between the inclination angle on the low slope and the slip of the work vehicle 1. Ask for data.
- the image of the behavior analysis data is an image as shown in the graphs of FIGS. 1 and 2.
- the work vehicle 1 learns a model regarding slip on a steep slope in order to estimate the slip of the work vehicle 1 on the steep slope shown in FIG. Specifically, a model for estimating the slip of the work vehicle 1 is learned by using the behavior analysis data on a low slope with a low risk of an unknown environment and a plurality of past behavior analysis data.
- a plurality of past behavior analysis data can be represented by an image as shown in the graph of FIG.
- the known environment is S 1 (cohesive soil), S 2 (sandy ground), and S 3 (rock mass)
- the past multiple behavior analysis data is generated by analyzing the moving body state data in each environment. It is the data showing the relationship between the tilt angle and the slip. It should be noted that a plurality of past behavior analysis data are stored in the storage device.
- the behavior analysis data generated based on the moving object state data measured on the low slope of the unknown environment and the past behavior generated in each of the known environments S1 , S2, and S3. Learn the model using the analysis data.
- the work vehicle 1 analyzes the environmental state data representing the state of the steep slope acquired from the sensor by the work vehicle 1 on a low slope with a low risk of an unknown environment, and obtains the environmental analysis data representing the topographical shape and the like. Generate.
- the work vehicle 1 inputs the environmental analysis data into the model for estimating the behavior of the moving object in the target environment, and estimates the slip of the work vehicle 1 on a steep slope in the target environment.
- the behavior of the moving object can be estimated accurately in an unknown environment. Therefore, the moving body can be controlled accurately even in an unknown environment.
- FIG. 3 is a diagram for explaining an example of the behavior estimation device.
- the behavior estimation device 10 shown in FIG. 3 is a device for learning a model used for accurately estimating the behavior of a moving object in an unknown environment. Further, as shown in FIG. 3, the behavior estimation device 10 includes a behavior analysis unit 11, a learning unit 12, an environment analysis unit 13, an estimation unit 14, and a learning instruction unit 15.
- the behavior estimation device 10 is equipped with, for example, a programmable device such as a CPU (Central Processing Unit) or FPGA (Field-Programmable Gate Array), a GPU (Graphics Processing Unit), or all of them, or two or more thereof. Circuits and information processing devices.
- a programmable device such as a CPU (Central Processing Unit) or FPGA (Field-Programmable Gate Array), a GPU (Graphics Processing Unit), or all of them, or two or more thereof. Circuits and information processing devices.
- the behavior analysis unit 11 generates behavior analysis data (first behavior analysis data) representing the actual behavior of the moving object in the target environment (first environment: unknown environment). Specifically, the behavior analysis unit 11 analyzes the behavior of the moving body based on the moving body state data representing the state of the moving body, and generates behavior analysis data representing the behavior of the moving body.
- the target environment is an unknown environment in which mobile objects move, for example, in disaster areas, construction sites, forests, planets, etc.
- the moving body is, for example, an autonomous vehicle, a ship, an aircraft, a robot, or the like.
- the work vehicle is, for example, a construction vehicle used for work in a disaster area, a construction site, or a forest, or an exploration vehicle used for exploration on a planet.
- the moving body state data is data representing the state of the moving body acquired from a plurality of sensors for measuring the state of the moving body.
- the sensors that measure the state of the moving body are, for example, a position sensor that measures the position of the vehicle, an IMU (Inertial Measurement Unit: 3-axis gyro sensor + 3-axis angular velocity sensor), a wheel encoder, and consumption. Instruments that measure power, instruments that measure fuel consumption, and so on.
- the behavior analysis data is data representing the moving speed, posture angle, etc. of the moving body, which is generated by using the moving body state data.
- the behavior analysis data includes, for example, the traveling speed of the vehicle, the wheel rotation speed of the vehicle, the attitude angle of the vehicle, the slip during traveling, the vibration of the vehicle during traveling, the power consumption, and the fuel consumption. It is data representing such things.
- the environmental analysis unit 13 analyzes the target environment based on the environmental state data representing the state of the target environment, and generates the environmental analysis data.
- the environmental state data is data representing the state of the target environment acquired from a plurality of sensors for measuring the state of the surrounding environment (target environment) of the moving object.
- the sensor for measuring the state of the target environment is, for example, LiDAR (Light Detection and Ringing, Laser Imaging Detection and Ringing), an image pickup device, or the like.
- LiDAR for example, generates 3D point cloud data around the vehicle.
- the image pickup device outputs image data (moving image or still image) by, for example, a camera that captures an image of the target environment.
- a sensor for measuring the state of the target environment a sensor provided in addition to the moving body, for example, a sensor provided in an aircraft, a drone, an artificial satellite, or the like may be used.
- Environmental analysis data is data representing the state of the target environment generated using the environmental state data.
- the environmental state data is data representing a topographical shape such as an inclination angle and unevenness.
- the environmental state data three-dimensional point cloud data, image data, three-dimensional map data, or the like may be used.
- the estimation unit 14 inputs the environmental analysis data into the model for estimating the behavior of the moving body in the target environment, and estimates the behavior of the moving body in the target environment.
- the model is a model for estimating the behavior of a moving object such as a work vehicle 1 in an unknown environment generated by the learning unit 12 described later.
- the learning unit 12 has behavior analysis data (first behavior analysis data) generated in the target environment and behavior analysis data (first behavior analysis data) generated for each known environment in the previously known environment (second environment). The degree of similarity between the target environment and the known environment is calculated using the second behavior analysis data). After that, the learning unit 12 learns a model for estimating the behavior of the moving object in the target environment by using the calculated similarity and the model trained for each known environment.
- the model is a model used for estimating the behavior of a moving object such as a work vehicle 1 in an unknown environment.
- the model can be represented by a function as shown in Equation 1.
- the model to which the number 1 is applied there is a model in which the N Gaussian process regression models fG (Si) shown in the number 2 are combined by a weighted linear sum.
- Each Gaussian process regression model builds a model based on behavior analysis data.
- the weight wi shown in Equation 2 is learned.
- the weight wi is a model parameter representing the degree of similarity between the behavior analysis data corresponding to the target environment and the behavior analysis data corresponding to the known environment.
- Equation 3 N linear regression models fG (Si) shown in Equation 3 are combined by a weighted linear sum.
- the linear regression model builds a model based on a trained model generated for each of several known environments in the past.
- the learning instruction unit 15 sets a confidence interval based on the behavior estimation result data estimated by the model, and when the first behavior analysis data exists in the set confidence interval, the learning unit 12 causes the learning unit 12 to relearn the model. Give instructions.
- FIG. 4 is a diagram for explaining the re-learning of the model.
- the learning instruction unit 15 first acquires the behavior estimation result data regarding the slip of the work vehicle 1 estimated by the estimation unit 14 using the model. Next, the learning instruction unit 15 sets a confidence interval based on the acquired behavior estimation result data.
- the confidence interval is set by the confidence lines 1 and 2 (dotted line) centering on the behavior estimation result data (solid line).
- the width of the confidence interval (distance between the confidence line 1 including the behavior estimation result data and the confidence line 2) is determined by, for example, an experiment, a simulation, or the like, and is stored in the storage unit in advance. The width of the behavior estimation result data and the confidence line 1 and the width of the behavior estimation result data and the confidence line 2 do not have to be the same.
- the variance value can be estimated in addition to the estimated average value corresponding to the solid line in FIG.
- the mean ⁇ a * variance is set as the confidence interval.
- the learning instruction unit 15 determines whether or not the behavior analysis data exists in the set confidence interval. When the behavior analysis data does not exist in the set confidence interval, the learning instruction unit 15 instructs the learning unit 12 to relearn the model. Further, when the behavior analysis data exists in the set confidence interval, the learning instruction unit 15 does not instruct the learning unit 12 to relearn the model.
- the learning instruction unit 15 since the actual behavior analysis data 1 (dotted line) of the moving body generated in the target environment exists in the confidence interval, the learning instruction unit 15 causes the learning unit 12 to relearn the model. Do not give instructions for. On the other hand, when the behavior analysis data 2 (dotted line) does not exist in the confidence interval, the learning instruction unit 15 instructs the learning unit 12 to relearn (update) the model.
- the determination of whether or not the behavior analysis data exists in the confidence interval is not determined only by the data at a single time, but the behavior analysis data analyzed in a predetermined period (for example, the latest 10 [m]). It may be determined whether or not 90 [%] or more of them are included in the confidence interval.
- FIG. 5 is a diagram for explaining an example of a system having a behavior estimation device.
- the system 100 shown in FIG. 5 is a system for planning and controlling the movement route of a moving body in an unknown environment. As shown in FIG. 5, it has a route generation device 20, a measurement unit 30, a storage device 40, and a mobile control unit 50.
- the route generation device 20 includes a behavior estimation device 10, a movement route generation unit 16, and a replanning instruction unit 17.
- the measuring unit 30 has a sensor 31 and a sensor 32.
- the sensor 31 is a sensor for measuring the state of the moving body described above.
- the sensor 32 is a sensor for measuring the state of the surrounding environment (target environment) of the moving body described above.
- the sensor 31 measures the state of the moving body and outputs the measured moving body state data to the behavior analysis unit 11.
- the sensor 31 has a plurality of sensors.
- the sensor 31 is, for example, a position sensor for measuring the position of the vehicle, an IMU, a wheel encoder, an instrument for measuring power consumption, an instrument for measuring fuel consumption, and the like.
- the position sensor is, for example, a GPS (Global Positioning System) receiver or the like.
- the IMU measures, for example, the acceleration in the three axes (XYZ axes) of the vehicle and the angular velocity around the three axes of the vehicle.
- the wheel encoder measures the rotational speed of the wheel.
- the sensor 32 measures the state of the surrounding environment (target environment) of the moving object, and outputs the measured environmental state data to the environment analysis unit 13.
- the sensor 32 has a plurality of sensors.
- the sensor 32 is, for example, LiDAR, an image pickup device, or the like.
- the sensor for measuring the state of the target environment may be a sensor provided in a sensor other than the mobile body, for example, a sensor provided in an aircraft, a drone, an artificial satellite, or the like.
- the behavior analysis unit 11 first acquires the moving body state data measured by each of the sensors included in the sensor 31 in the target environment. Next, the behavior analysis unit 11 analyzes the acquired mobile body state data and generates behavior analysis data (first behavior analysis data) representing the behavior of the moving body. Next, the behavior analysis unit 11 outputs the generated first behavior analysis data to the learning unit 12.
- the learning unit 12 acquires the first behavior analysis data output from the behavior analysis unit 11 and the second behavior analysis data stored in the storage device 40 for each known environment. Next, the learning unit 12 learns using the acquired models of the first behavior analysis data and the second behavior analysis data, using the models shown in the numbers 2 and 3. Next, the learning unit 12 stores the model parameters generated by the learning in the storage device 40.
- the learning unit 12 resembles the first behavior analysis data, the second behavior analysis data generated for each second environment, and the geological characteristics of each position in each of the first environment and the second environment. You may train the model using degrees.
- the geological characteristics of the near position are similar, but the geological characteristics of the distant position are likely to be different. Therefore, the accuracy of behavior estimation can be improved by learning the model using the similarity of geological characteristics.
- the model can be represented by a function as shown in Equation 4.
- the behavior estimation model f G based on the topographical information and the relational model f P of the position and the geological characteristics (behavior) are explicitly separated and modeled to behave in the place where the vehicle travels. The accuracy of estimation can be improved.
- the behavior estimation model fG is input with information related to topographical information to estimate the behavior.
- the model f P inputs the position and the geological characteristics as the position information and estimates the behavior.
- the behavior estimation model fG When information on the inclination angle and unevenness is input to the behavior estimation model fG as terrain information, if the terrain in front of the input is the same, the behavior is estimated assuming that the same running is performed. However, for example, even if the input value of the inclination angle x G is the same, there is a possibility that the running behavior will actually differ depending on how far away from the place where the behavior analysis data used for learning was acquired. be.
- the model f G and the model f P are modeled by, for example, Gaussian process regression or linear regression. Further, after learning each of the model f G and the model f P separately, the estimation results of each model may be multiplied. Further, learning may be performed in the form of f G / f P. Further, although the number 4 is modeled in the form of the product of f G and f P as an example, it may be modeled in the form of the sum of these.
- the environmental analysis unit 13 first acquires the environmental state data measured by each of the sensors included in the sensor 32 in the target environment. Next, the environment analysis unit 13 analyzes the acquired environment state data and generates environment analysis data representing the state of the environment. Next, the environmental analysis unit 13 outputs the generated environmental analysis data to the estimation unit 14. Further, the environmental analysis unit 13 may store the environmental analysis data in the storage device 40.
- the estimation unit 14 acquires the environmental analysis data output from the environmental analysis unit 13, the model parameters and hyperparameters stored in the storage device 40. Next, the estimation unit 14 inputs the acquired environment analysis data, model parameters, hyperparameters, etc. into the model for estimating the behavior of the moving body in the target environment, and estimates the behavior of the moving body in the target environment. .. Next, the estimation unit 14 may store the behavior estimation result data in the storage device 40.
- the storage device 40 is a memory for storing various data handled by the system 100.
- the various data include models, model parameters, hyperparameters, first behavior analysis data (eg, new behavior analysis data analyzed in an unknown environment), second behavior analysis data (eg, previously analyzed in a known environment). Multiple behavior analysis data), environment analysis data, behavior estimation result data, etc.
- the storage device 40 is provided in the system 100, but may be provided separately from the system 100. In that case, the storage device 40 may be a storage device such as a database or a server computer.
- the learning instruction unit 15 first acquires behavior estimation result data from the estimation unit 14. Next, the learning instruction unit 15 sets a confidence interval based on the acquired behavior estimation result data. Next, the learning instruction unit 15 determines whether or not the behavior analysis data exists in the set confidence interval. When the behavior analysis data exists in the set confidence interval, the learning instruction unit 15 does not instruct the learning unit 12 to relearn the model. When the behavior analysis data does not exist in the set confidence interval, the learning instruction unit 15 instructs the learning unit 12 to relearn the model.
- the movement route generation unit 16 generates movement route data representing a route from the current position to the destination based on the result of estimating the behavior of the moving object in the target environment (behavior estimation result data). The generation of movement route data will be described later.
- the movement route generation unit 16 acquires an instruction for replanning from the replanning instruction unit 17, the movement route generation unit 16 moves from the current position to the destination based on the behavior estimation result data of the relearned model. Generates travel route data that represents the route.
- the replanning instruction unit 17 acquires behavior estimation result data from the estimation unit 14. Next, the replanning instruction unit 17 determines whether or not to generate movement route data (replanning) based on the acquired behavior estimation result data. When it is determined that the replanning is to be performed, the replanning instruction unit 17 instructs the movement route generation unit 16 to generate the movement route data. Further, when it is determined that the replanning is not performed, the replanning instruction unit 17 does not instruct the movement route generation unit 16 to generate the movement route data.
- the replanning instruction unit 17 instructs the movement route generation unit 16 to generate movement route data when the model is relearned. Further, the replanning instruction unit 17 instructs the movement route generation unit 16 to generate the movement route data when the route correction is necessary even if the model has not been relearned. For example, when an obstacle is detected on the planned route, or when the moving body deviates significantly from the planned route, the replanning instruction unit 17 generates movement route data for the movement route generation unit 16. Give instructions to do.
- the travel route data is sent to the travel route generation unit 16. Do not give instructions to generate.
- FIG. 6 is a diagram for explaining the generation of movement route data. As shown in FIG. 6, at the current position, the slip in front of the route is estimated, and when it is determined that the estimated slip value is higher than the risk threshold value (when the risk is high), the vehicle moves to correct the route. Instruct the route generation unit 16 to generate the movement route data.
- the route is not corrected, so that the travel route data is sent to the travel route generation unit 16. Give instructions to generate.
- the moving body control unit 50 controls and moves the moving body based on the behavior estimation result data and the movement route data.
- the mobile body control unit 50 first acquires the behavior estimation result data and the movement route data. Next, the moving body control unit 50 generates information for controlling each part related to the movement of the moving body based on the behavior estimation result data and the movement route data. Then, the moving body control unit 50 controls the moving body to move it from the current position to the target location.
- Example 1 The behavior estimation device 10 and the route generation device 20 will be specifically described.
- the slip (behavior) of the work vehicle 1 when traveling on a slope in an unknown environment is estimated from the data acquired when traveling on a low slope.
- the slip is modeled as a function of the topographical shape (inclination angle, unevenness) of the target environment.
- the behavior analysis unit 11 causes the work vehicle 1 to travel on a gentle terrain with a low risk of the target environment at a constant speed, and obtains moving object state data from the sensor 31 of the measurement unit 30 at regular intervals. get.
- the behavior analysis unit 11 acquires mobile state data at intervals of, for example, 0.1 [seconds] or 0.1 [m].
- the behavior analysis unit 11 uses the acquired moving body state data to move the moving speeds Vx, Vy, and Vz of the work vehicle 1 in the XYZ directions, the wheel rotation speed ⁇ of the work vehicle 1, and the XYZ of the work vehicle 1.
- the attitude angle around the axis (roll angle ⁇ x, pitch angle ⁇ y, yaw angle ⁇ z) is calculated.
- the movement speed is calculated by, for example, dividing the difference in time between the two points from the difference in GPS latitude, longitude, and altitude between the two points.
- the attitude angle is calculated, for example, by integrating the angular velocity of the IMU.
- the moving speed and the posture angle may be calculated based on the Kalman filter using both the moving body state data measured by GPS and the IMU.
- the movement speed and attitude angle may be calculated based on SLAM (Simultaneous Localization and Mapping: a technique for simultaneously estimating the position of a moving object and constructing a peripheral map) based on GPS, IMU, and LiDAR data. good.
- SLAM Simultaneous Localization and Mapping: a technique for simultaneously estimating the position of a moving object and constructing a peripheral map
- the behavior analysis unit 11 calculates the slip based on the speed of the work vehicle 1 and the wheel rotation speed, as shown in Equation 5.
- the slip is a continuous value.
- the behavior analysis unit 11 outputs a plurality of data points (first behavior analysis data) having a roll angle ⁇ x, a pitch angle ⁇ y, and a slip as a set of data points to the learning unit 12.
- the learning unit 12 has a data point (first behavior analysis data) stored in the behavior analysis unit 11 and a data point (second behavior analysis) stored in the storage device 40 and generated in a previously known environment. Based on the degree of similarity with the data), the model related to the roll angle ⁇ x, pitch angle ⁇ y, and slip in the target environment is learned.
- the learning unit 12 has a data point (first behavior analysis data) stored in the behavior analysis unit 11 and a data point (second behavior analysis data) stored in the storage device 40 and generated in a previously known environment. ), The roll angle ⁇ x, the pitch angle ⁇ y, and the model related to slip in the target environment are learned based on the similarity with the model generated based on.
- the likelihood of the behavior analysis data in the target environment when modeled by f (Si) is used.
- Likelihood is the probability of how likely a data point in a target environment is to that model, assuming that each model in a known environment represents a slip phenomenon in the target environment.
- g (wi ) of the number 2 be wi / ⁇ wi .
- a model of f (T) of equation 2 is constructed as the sum of weights of f (Si) with g (wi) as the weight.
- the weight wii is based on the index of how well the data in the target environment can be represented by the model in each known environment. To decide.
- the reciprocal of the mean square error (MSE) when the slip in the target environment is estimated using the model in each known environment is set in the weight wi .
- the coefficient of determination (R 2 ) when the slip in the target environment is estimated using the model in each known environment is set to the weight wi .
- Gaussian process regression can be used to represent not only average estimation but also estimation uncertainty as a probability distribution. can.
- the weight wi the likelihood of the data in the target environment when the slip in the target environment is estimated using each model of the known environment is used.
- a threshold value may be set for the similarity (1 / MSE, R2 , likelihood), and only a model in a known environment in which the similarity is equal to or higher than the threshold value may be used. Further, only the model having the highest similarity may be used, or the specified number of models may be used in descending order of similarity.
- modeling may be performed by a method other than the above-mentioned polynomial regression or Gaussian process regression.
- Other machine learning methods include support vector machines and neural networks.
- modeling the relationship between the input and the output as a black box as in the machine learning method it may be modeled as a white box based on a physical model.
- the model parameters stored in the storage device 40 may be used as they are, or the model parameters may be relearned using the data acquired while traveling in the target environment. You may (update).
- a threshold value may be set for the similarity (1 / MSE, R2 , likelihood), and only a model in a known environment in which the similarity is equal to or higher than the threshold value may be used.
- the model in a plurality of known environments stored in the storage device 40 may be one learned based on the data acquired in the real world, or may be learned based on the data acquired by the physical simulation.
- the environmental analysis unit 13 first acquires environmental state data from the sensor 32 of the measurement unit 30.
- the environment analysis unit 13 acquires, for example, a three-dimensional point cloud (environmental state data) generated by measuring the target environment in front of the work vehicle 1 using LiDAR mounted on the work vehicle 1.
- the environmental analysis unit 13 processes the three-dimensional point cloud to generate topographical shape data (environmental analysis data) related to the topographical shape.
- FIG. 7 is a diagram for explaining an example of information regarding the topographical shape.
- the environmental analysis unit 13 calculates an approximate plane that minimizes the average distance error of the point group from the point group included in the grid itself and the grid in eight directions around the grid for each grid, and the approximate plane thereof. Calculate the maximum tilt angle and tilt direction of.
- the environmental analysis unit 13 generates topographical shape data (environmental analysis data) in association with the coordinates representing the position of the grid, the maximum tilt angle of the approximate plane, and the tilt direction for each grid, and the storage device 40.
- the estimation unit 14 estimates the slip in each grid based on the topographical shape data generated by the environmental analysis unit 13 and the trained slip model.
- the slip estimation method for each grid will be specifically described. (1) Only the maximum tilt angle of the grid is input to the model to estimate the slip. However, in reality, the slip of the work vehicle 1 is determined by which direction the work vehicle 1 faces with respect to the slope. For example, when the work vehicle 1 faces the maximum inclination angle direction (the direction with the steepest inclination), the slip becomes the largest, so it is conservatively predicted to estimate the slip using the maximum inclination angle. Means to do. The slip may be estimated by setting the pitch angle of the work vehicle 1 as the maximum inclination angle and the roll angle as 0.
- the slip is estimated according to the traveling direction of the work vehicle 1 when passing through the grid.
- the roll angle and pitch angle of the work vehicle 1 are calculated based on the maximum inclination angle and the slope direction, and the traveling direction of the work vehicle 1.
- slip is estimated for each grid in the traveling direction of the plurality of work vehicles 1 (for example, at intervals of 15 degrees).
- the mean value and variance value of slip are estimated. Since the behavior of the work vehicle 1 becomes complicated on steep slopes and terrain with severe unevenness, there is a high possibility that the slip variation becomes large. Therefore, by estimating the dispersion as well as the average, the safe work vehicle 1 can be further improved. Can be operated.
- the estimation unit 14 associates the estimated slips (continuous values of slips in the maximum inclination angle direction) with each of the grids, generates behavior estimation result data, and stores the behavior estimation result data in the storage device 40. ..
- FIG. 8 is a diagram for explaining the relationship between the grid and the slip.
- the estimation unit 14 generates behavior estimation result data in association with the estimated slip and the vehicle traveling direction in each of the grids and stores them in the storage device 40.
- the vehicle traveling direction is expressed by using, for example, an angle with respect to a predetermined direction.
- the estimation unit 14 generates behavior estimation result data in association with the estimated slip average, the slip dispersion, and the vehicle traveling direction in each grid, and stores it in the storage device 40.
- the estimation unit 14 determines whether it is passable or impassable based on a preset threshold value for slip, associates information representing the determination result with a grid, generates behavior estimation result data, and stores it in the storage device 40.
- FIG. 9 is a diagram for explaining the relationship between the grid and passable / impossible. “ ⁇ ” shown in FIG. 9 indicates passable, and “ ⁇ ” indicates impassable.
- the slip is modeled using only the terrain shape as a feature amount, but when the work vehicle 1 is equipped with an image pickup device such as a camera, the image data (in addition to the terrain shape) (for example, the brightness value or texture of each pixel) may be added to the input data (feature amount) of the model.
- an image pickup device such as a camera
- the position where the mobile state data was acquired may also be used as the feature quantity.
- the movement speed, the steering operation amount, the change in weight and weight balance due to the increase / decrease in the load of the work vehicle 1, the passive / active change in the shape of the work vehicle 1 due to the suspension or the like may be added to the feature amount.
- Example 1 slip has been described, but as another behavior of the estimation target, for example, there is vibration of the work vehicle 1.
- the basic processing flow is the same as in the case of slip described above.
- the time-series information of the acceleration measured by the IMU is converted into the magnitude and frequency of the vibration by, for example, Fourier transform, and it is modeled as a function of the terrain shape.
- other behaviors of the estimation target include, for example, power consumption, fuel consumption of fuel, and attitude angle of the vehicle.
- the basic learning and estimation flow for each behavior is the same as the slip described above.
- Power consumption and fuel consumption are modeled using the measured values of the corresponding instruments and the terrain shape data.
- the posture angle is almost the same as the inclination angle of the ground in many cases, but depending on the geological characteristics and the severity of the unevenness, the vehicle body tilts more than the inclination angle of the ground and becomes a dangerous state. Therefore, for example, the terrain shape estimated from the point cloud measured in advance by LiDAR and the vehicle attitude angle when actually traveling on the terrain (the attitude angle of the vehicle calculated using the angular velocity measured by the IMU) are paired. As the input / output data of, the attitude angle is modeled as a function representing the topography of the target environment.
- Example 2 In the second embodiment, a method of planning and controlling the movement route of the moving body in an unknown environment will be described. Specifically, in the second embodiment, a movement route is obtained based on the estimation result obtained in the first embodiment, and the moving body is moved according to the obtained movement route.
- the movement path is generated by avoiding the place corresponding to the grid estimated to have a high slip value.
- a case of planning a movement route will be described using an example in which it is determined whether the vehicle can pass or cannot pass from the slip estimated based on the maximum inclination angle shown in FIG.
- any algorithm can be used as the algorithm for planning the movement route.
- a * Aster
- the adjacent node is searched sequentially from the current position, and the route is efficiently searched based on the movement cost between the current search node and the adjacent node and the movement cost from the adjacent node to the target position. Explore.
- each grid is set as one node, and each node can move to the adjacent node in 16 directions.
- the travel cost is the Euclidean distance between the nodes.
- FIG. 10 is a diagram for explaining an example of a movement route.
- the movement route generation unit 16 outputs information representing a series of nodes on the movement route to the movement control unit 50.
- the movement route is generated including the direction of the work vehicle 1.
- the reason is that the direction of movement of the work vehicle 1 is limited, such as the work vehicle 1 cannot move to the side and the steering angle is limited, so that the direction of the vehicle must also be taken into consideration.
- each grid is set as one node, and each node can move to the adjacent node in 16 directions. Since the estimated slip is reflected in the route search, for example, the travel cost between the nodes is not a mere Euclidean distance but a sum of the weights of the distance and the slip shown in Equation 6.
- FIG. 11 is a diagram for explaining an example of a movement route.
- the estimation considering uncertainty can be expressed by Gaussian process regression, that is, when the mean value and variance value of slip are estimated for each grid, for example, even if the mean value is small, the variance value ( Generate a movement path so as to avoid a grid with a large prediction uncertainty).
- FIG. 12 is a diagram for explaining an example of the operation of the behavior estimation device.
- FIG. 13 is a diagram for explaining an example of the operation of the route generation device.
- the behavior estimation method and the route generation method are implemented by operating the behavior estimation device 10, the route generation device 20, and the system 100 in the embodiment, the first embodiment, and the second embodiment. Therefore, the description of the behavior estimation method and the route generation method in the embodiment, the first embodiment, and the second embodiment is replaced with the following operation explanations of the behavior estimation device 10, the route generation device 20, and the system 100.
- the behavior analysis unit 11 acquires the moving body state data from the sensor 31 (step A1). Next, the behavior analysis unit 11 analyzes the behavior of the moving body based on the moving body state data representing the state of the moving body, and generates the first behavior analysis data representing the behavior of the moving body (step A2). ..
- the environmental analysis unit 13 acquires the environmental state data from the sensor 32 (step A3).
- the environment analysis unit 13 analyzes the target environment based on the environment state data representing the state of the target environment, and generates the environment analysis data (step A4).
- steps A1, A3 or A3, A1 the processing may be performed in the order of steps A2, A4 or steps A4, A2. Further, after the processing of steps A3 and A4, the processing of steps A1 and A2 may be performed. Further, the processes of steps A1 and A2 and the processes of steps A3 and A4 may be processed in parallel.
- the estimation unit 14 inputs the environmental analysis data into the model for estimating the behavior of the moving object in the target environment, and estimates the behavior of the moving object in the target environment (step A5).
- the learning instruction unit 15 sets a confidence interval based on the behavior estimation result data estimated by the model, and determines whether or not the first behavior analysis data exists in the set confidence interval (step). A6).
- the learning unit 12 is not instructed to re-learn the model (step A7: No).
- the learning unit 12 is instructed to re-learn the model (step A7: Yes).
- the learning unit 12 uses the first behavior analysis data generated in the target environment and the second behavior analysis data generated for each known environment in the previously known environment to be used in the target environment.
- a model for estimating the behavior of the moving body in the above is learned (step A8).
- step A9: Yes when the behavior estimation device 10 receives an instruction to end the behavior estimation process (step A9: Yes), the behavior estimation process is terminated.
- step A9: No the process proceeds to step A1 and the behavior estimation process is continued.
- the behavior estimation processing of steps A1 to A8 is executed.
- the estimation unit 14 inputs the environment analysis data into the relearned model, and newly estimates the behavior of the moving object in the target environment (step B1).
- the replanning instruction unit 17 acquires the behavior estimation result data generated by using the model relearned from the estimation unit 14, and generates the movement route data (replanning) based on the acquired behavior estimation result data. Determine whether or not to do so. (Step B2).
- the replanning instruction unit 17 determines that the replanning is to be performed, the replanning instruction unit 17 instructs the movement route generation unit 16 to generate the movement route data (step B3: Yes). .. Further, when it is determined that the replanning is not performed, the replanning instruction unit 17 does not instruct the movement route generation unit 16 to generate the movement route data (step B3: No).
- the replanning instruction unit 17 instructs the movement route generation unit 16 to generate movement route data when the model is relearned. Further, the replanning instruction unit 17 instructs the movement route generation unit 16 to generate the movement route data when the route correction is necessary even if the model has not been relearned. For example, when an obstacle is detected on the planned route, or when the moving body deviates significantly from the planned route, the replanning instruction unit 17 generates movement route data for the movement route generation unit 16. Give instructions to do.
- the movement route generation unit 16 generates movement route data representing the movement route from the current position to the destination based on the behavior estimation result data (step B4).
- step B4 the movement route generation unit 16 acquires the behavior estimation result data of the moving object in the target environment as shown in FIGS. 8 and 9 from the estimation unit 14.
- step B4 the movement route generation unit 16 applies general route planning processing to the behavior estimation result data of the moving body to generate movement route data.
- the movement route generation unit 16 outputs the movement route data to the moving body control unit 50.
- the moving body control unit 50 controls and moves the moving body based on the behavior estimation result data and the movement route data.
- the mobile body control unit 50 first acquires the behavior estimation result data and the movement route data. Next, the moving body control unit 50 generates information for controlling each part related to the movement of the moving body based on the behavior estimation result data and the movement route data. Then, the moving body control unit 50 controls and moves the moving body from the current position to the target location.
- step B5 when the route generation device 20 receives an instruction to end the route generation process (step B5: Yes), the route generation process is terminated.
- step B5: No When the route generation process is continued (step B5: No), the process proceeds to step A1 and the route generation process is continued.
- the number of times of re-learning of the model in an unknown environment can be reduced.
- the behavior of a moving object such as a work vehicle can be estimated accurately, and further, a decrease in operational efficiency of the work vehicle can be suppressed.
- the program in the embodiment, the first embodiment, and the second embodiment may be a program that causes a computer to execute steps A1 to A9 and steps B1 to B5 shown in FIGS. 12 and 13.
- the computer processor functions as a behavior analysis unit 11, a learning unit 12, an environment analysis unit 13, an estimation unit 14, a learning instruction unit 15, a movement route generation unit 16, a replanning instruction unit 17, and a moving body control unit 50. And process.
- each computer has a behavior analysis unit 11, a learning unit 12, an environment analysis unit 13, an estimation unit 14, a learning instruction unit 15, a movement route generation unit 16, a replanning instruction unit 17, and a moving body control. It may function as any of the parts 50.
- FIG. 14 is a block diagram showing an example of a computer that realizes a system having a behavior estimation device or a route generation device.
- the computer 110 includes a CPU (Central Processing Unit) 111, a main memory 112, a storage device 113, an input interface 114, a display controller 115, a data reader / writer 116, and a communication interface 117. And. Each of these parts is connected to each other via a bus 121 so as to be capable of data communication.
- the computer 110 may include a GPU (Graphics Processing Unit) or an FPGA (Field-Programmable Gate Array) in addition to the CPU 111 or in place of the CPU 111.
- the CPU 111 expands the program (code) in the present embodiment stored in the storage device 113 into the main memory 112, and executes these in a predetermined order to perform various operations.
- the main memory 112 is typically a volatile storage device such as a DRAM (Dynamic Random Access Memory).
- the program in the present embodiment is provided in a state of being stored in a computer-readable recording medium 120.
- the program in the present embodiment may be distributed on the Internet connected via the communication interface 117.
- the recording medium 120 is a non-volatile recording medium.
- the storage device 113 include a semiconductor storage device such as a flash memory in addition to a hard disk drive.
- the input interface 114 mediates data transmission between the CPU 111 and an input device 118 such as a keyboard and mouse.
- the display controller 115 is connected to the display device 119 and controls the display on the display device 119.
- the data reader / writer 116 mediates the data transmission between the CPU 111 and the recording medium 120, reads the program from the recording medium 120, and writes the processing result in the computer 110 to the recording medium 120.
- the communication interface 117 mediates data transmission between the CPU 111 and another computer.
- the recording medium 120 include a general-purpose semiconductor storage device such as CF (Compact Flash (registered trademark)) and SD (Secure Digital), a magnetic recording medium such as a flexible disk, or a CD-.
- CF Compact Flash
- SD Secure Digital
- a magnetic recording medium such as a flexible disk
- CD- Compact Disk Read Only Memory
- optical recording media such as ROM (Compact Disk Read Only Memory).
- the behavior estimation device 10, the route generation device 20, and the system 100 in the first and second embodiments can be realized by using the hardware corresponding to each part instead of the computer in which the program is installed. be. Further, the behavior estimation device 10, the route generation device 20, and the system 100 may be partially realized by a program and the rest by hardware.
- the behavior analysis unit that generates the first behavior analysis data that represents the actual behavior of the moving object in the first environment
- An environmental analysis unit that analyzes the first environment based on the environmental state data representing the state of the first environment and generates environmental analysis data.
- An estimation unit that inputs the environmental analysis data into a model for estimating the behavior of the moving body in the first environment and estimates the behavior of the moving body in the first environment.
- a confidence interval is set based on the behavior estimation result data estimated by the model, and if the first behavior analysis data exists in the set confidence interval, the model is re-learned in the learning unit for learning the model.
- a learning instruction unit that gives instructions for learning, Behavior estimation device with.
- the behavior estimation device (Appendix 2) The behavior estimation device according to Appendix 1.
- the learning unit includes the first behavior analysis data, the second behavior analysis data generated for each second environment, and the geological characteristics of each position in each of the first environment and the second environment.
- a behavior estimation device that learns the model using the degree of similarity.
- the behavior analysis unit that generates the first behavior analysis data that represents the actual behavior of the moving object in the first environment
- An environmental analysis unit that analyzes the first environment based on the environmental state data representing the state of the first environment and generates environmental analysis data.
- An estimation unit that inputs the environmental analysis data into a model for estimating the behavior of the moving body in the first environment and estimates the behavior of the moving body in the first environment.
- a confidence interval is set based on the behavior estimation result data estimated by the model, and if the first behavior analysis data exists in the set confidence interval, the model is re-learned in the learning unit for learning the model.
- a learning instruction unit that gives instructions for learning, When the model is re-learned, the movement route data representing the movement route from the current position to the destination is regenerated based on the behavior estimation result data generated by using the re-learned model.
- the generator and Route generator with.
- the route generator according to Appendix 3,
- the learning unit includes the first behavior analysis data, the second behavior analysis data generated for each second environment, and the geological characteristics of each position in each of the first environment and the second environment.
- a route generator that learns the model using similarity.
- a behavior analysis step that generates the first behavior analysis data that represents the actual behavior of the moving object in the first environment.
- An environmental analysis step that analyzes the first environment based on the environmental state data representing the state of the first environment and generates environmental analysis data.
- An estimation step of inputting the environmental analysis data into a model for estimating the behavior of the moving body in the first environment and estimating the behavior of the moving body in the first environment.
- a confidence interval is set based on the behavior estimation result data estimated by the model, and when the first behavior analysis data exists in the set confidence interval, an instruction for retraining the model is given. Instruction steps and Behavior estimation method with.
- Appendix 6 The behavior estimation method described in Appendix 5. Using the first behavior analysis data, the second behavior analysis data generated for each second environment, and the similarity of the geological characteristics of each position in each of the first environment and the second environment. A behavior estimation method for learning the model.
- a behavior analysis step that generates the first behavior analysis data that represents the actual behavior of the moving object in the first environment.
- An environmental analysis step that analyzes the first environment based on the environmental state data representing the state of the first environment and generates environmental analysis data.
- An estimation step of inputting the environmental analysis data into a model for estimating the behavior of the moving body in the first environment and estimating the behavior of the moving body in the first environment.
- a confidence interval is set based on the behavior estimation result data estimated by the model, and if the first behavior analysis data exists in the set confidence interval, the model is re-learned in the learning unit for learning the model.
- a learning instruction step that gives instructions to learn, When the model is re-learned, the movement route data representing the movement route from the current position to the destination is regenerated based on the behavior estimation result data generated by using the re-learned model.
- Generation step and Route generation method having.
- Appendix 8 The route generation method described in Appendix 7. Using the first behavior analysis data, the second behavior analysis data generated for each second environment, and the similarity of the geological characteristics of each position in each of the first environment and the second environment. A route generation method for learning the model.
- a behavior analysis step that generates the first behavior analysis data that represents the actual behavior of the moving object in the first environment.
- An environmental analysis step that analyzes the first environment based on the environmental state data representing the state of the first environment and generates environmental analysis data.
- An estimation step of inputting the environmental analysis data into a model for estimating the behavior of the moving body in the first environment and estimating the behavior of the moving body in the first environment.
- a confidence interval is set based on the behavior estimation result data estimated by the model, and when the first behavior analysis data exists in the set confidence interval, an instruction for retraining the model is given.
- Instruction steps and A computer-readable recording medium recording a program that contains instructions to perform processing.
- Appendix 10 The computer-readable recording medium according to Appendix 9, wherein the recording medium is readable.
- the recording medium is readable.
- a behavior analysis step that generates the first behavior analysis data that represents the actual behavior of the moving object in the first environment.
- An environmental analysis step that analyzes the first environment based on the environmental state data representing the state of the first environment and generates environmental analysis data.
- An estimation step of inputting the environmental analysis data into a model for estimating the behavior of the moving body in the first environment and estimating the behavior of the moving body in the first environment.
- a confidence interval is set based on the behavior estimation result data estimated by the model, and if the first behavior analysis data exists in the set confidence interval, the model is re-learned in the learning unit for learning the model.
- a learning instruction step that gives instructions to learn
- the model When the model is re-learned, the movement route data representing the movement route from the current position to the destination is regenerated based on the behavior estimation result data generated by using the re-learned model.
- Generation step and A computer-readable recording medium recording a program that contains instructions to perform processing.
- Appendix 12 The computer-readable recording medium according to Appendix 11, wherein the recording medium is readable. Using the first behavior analysis data, the second behavior analysis data generated for each second environment, and the similarity of the geological characteristics of each position in each of the first environment and the second environment. A computer-readable recording medium for learning the model.
- the present invention As described above, according to the present invention, the number of times of re-learning of the model in an unknown environment can be reduced. As a result, the behavior of a moving object such as a work vehicle can be estimated accurately, and further, a decrease in operational efficiency of the work vehicle can be suppressed.
- the present invention is useful in fields where it is necessary to estimate the behavior of moving objects.
Landscapes
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Transportation (AREA)
- Mechanical Engineering (AREA)
- Mathematical Physics (AREA)
- Aviation & Aerospace Engineering (AREA)
- Navigation (AREA)
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
- Traffic Control Systems (AREA)
Abstract
Description
第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成する、挙動解析部と、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成する、環境解析部と、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定する、推定部と、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを学習する学習部に、前記モデルを再学習させるための指示をする、学習指示部と、
を有することを特徴とする。
第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成する、挙動解析部と、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成する、環境解析部と、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定する、推定部と、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを学習する学習手段に、前記モデルを再学習させるための指示をする、学習指示部と、
前記モデルが再学習された場合、再学習された前記モデルを用いて生成された挙動推定結果データに基づいて、現在位置から目的地までの移動経路を表す移動経路データを再生成する、移動経路生成部と、
を有することを特徴とする。
第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成する、挙動解析ステップと、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成する、環境解析ステップと、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定する、推定ステップと、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを学習する学習部に、前記モデルを再学習させるための指示をする、学習指示ステップと、
を有することを特徴とする。
第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成する、挙動解析ステップと、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成する、環境解析ステップと、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定する、推定ステップと、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを学習する学習手段に、前記モデルを再学習させるための指示をする、学習指示ステップと、
前記モデルが再学習された場合、再学習された前記モデルを用いて生成された挙動推定結果データに基づいて、現在位置から目的地までの移動経路を表す移動経路データを再生成する、移動経路生成ステップと、
を有することを特徴とする。
コンピュータに、
第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成する、挙動解析ステップと、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成する、環境解析ステップと、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定する、推定ステップと、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを学習する学習部に、前記モデルを再学習させるための指示をする、学習指示ステップと、
を実行させる命令を含むプログラムを記録していることを特徴とする。
コンピュータに、
第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成する、挙動解析ステップと、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成する、環境解析ステップと、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定する、推定ステップと、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを学習する学習手段に、前記モデルを再学習させるための指示をする、学習指示ステップと、
前記モデルが再学習された場合、再学習された前記モデルを用いて生成された挙動推定結果データに基づいて、現在位置から目的地までの移動経路を表す移動経路データを再生成する、移動経路生成ステップと、
を実行させる命令を含むプログラムを記録していることを特徴とする。
従来、被災地、建設現場、山林、惑星などの未知の環境において作業をする自律型の作業車両は、作業車両に搭載された撮像装置から未知の環境を撮像した画像データを取得し、取得した画像データに対して画像処理をし、画像処理の結果に基づいて未知の環境の状態を推定している。
以下、図面を参照して実施形態について説明する。図3を用いて、本実施形態における挙動推定装置10の構成について説明する。図3は、挙動推定装置の一例を説明するための図である。
図3に示す挙動推定装置10は、未知の環境において、移動体の挙動を精度よく推定するために用いるモデルを学習する装置である。また、図3に示すように、挙動推定装置10は、挙動解析部11と、学習部12と、環境解析部13と、推定部14と、学習指示部15とを有する。
モデルは、未知の環境において作業車両1などの移動体の挙動を推定するために用いるモデルである。モデルは、数1に示すような関数で表すことができる。
続いて、図5を用いて、本実施形態におけるシステム100の構成について説明する。図5は、挙動推定装置を有するシステムの一例を説明するための図である。
挙動推定装置10と経路生成装置20について具体的に説明する。実施例1では、未知の環境における作業車両1の斜面走行時のスリップ(挙動)を、低斜面を走行時に取得したデータから推定する場合について説明する。実施例1では、スリップを推定するので、スリップを、対象環境の地形形状(傾斜角、凹凸)の関数としてモデル化する。
実施例1の学習において、挙動解析部11は、作業車両1を、対象環境のリスクの低いなだらかな地形を一定速度で走行させ、一定間隔で、計測部30のセンサ31から移動体状態データを取得する。挙動解析部11は、例えば、0.1[秒]間隔、又は0.1[m]間隔などで移動体状態データを取得する。
推定において、作業車両1がこれから走行する地形形状を計測し、学習したモデルに基づいて対象環境におけるスリップを推定する。
環境解析部13は、まず、図7に示すように、対象環境(空間)を格子に区切り、格子それぞれに点群を割り振る。図7は、地形形状に関する情報の一例を説明するための図である。
(1)格子の最大傾斜角のみをモデルに入力してスリップを推定する。ただし、実際には、作業車両1のスリップは、斜面に対して作業車両1がどの向きを向いているかどうかによって決まる。例えば、最大傾斜角方向(一番傾斜が急な向き)を作業車両1が向いている場合、最もスリップが大きくなるので、最大傾斜角を使用してスリップを推定することは、保守的に予測を行うことを意味する。なお、作業車両1のピッチ角=最大傾斜角、ロール角=0として、スリップを推定してもよい。
実施例2では、未知の環境における移動体の移動経路の計画及び移動制御の方法について説明する。具体的には、実施例2では、実施例1で求めた推定結果に基づいて移動経路を求め、求めた移動経路にしたがって移動体を移動させる。
Cost = a * L + b * Slip
Cost :ノード間の移動コスト
L :ユークリッド距離
Slip :スリップ
a,b :移動経路を生成に用いる重み(0以上の値)
次に、本発明の実施形態、実施例1、実施例2における挙動推定装置10、経路生成装置20の動作について図を用いて説明する。
図12に示すように、まず、挙動解析部11は、センサ31から移動体状態データを取得する(ステップA1)。次に、挙動解析部11は、移動体の状態を表す移動体状態データに基づいて、移動体の挙動を解析し、移動体の挙動を表す第一の挙動解析データを生成する(ステップA2)。
図13に示すように、まず、ステップA1からA8の挙動推定処理を実行する。次に、推定部14は、環境解析データを、再学習したモデルに入力して、新たに対象環境における移動体の挙動を推定する(ステップB1)。
以上のように実施形態、実施例1、実施例2によれば、未知の環境におけるモデルの再学習の回数を低減できる。その結果、作業車両などの移動体の挙動を精度よく推定でき、更に、作業車両の運用効率の低下を抑制できる。
実施形態、実施例1、実施例2におけるプログラムは、コンピュータに、図12、図13に示すステップA1からA9、ステップB1からB5を実行させるプログラムであればよい。このプログラムをコンピュータにインストールし、実行することによって、実施形態、実施例1、実施例2における挙動推定装置10、経路生成装置20、システム100とそれらの方法を実現することができる。この場合、コンピュータのプロセッサは、挙動解析部11、学習部12、環境解析部13、推定部14、学習指示部15、移動経路生成部16、再計画指示部17、移動体制御部50として機能し、処理を行なう。
ここで、実施形態、実施例1、実施例2におけるプログラムを実行することによって、挙動推定装置10、経路生成装置20、システム100を実現するコンピュータについて図14を用いて説明する。図14は、挙動推定装置又は経路生成装置を有するシステムを実現するコンピュータの一例を示すブロック図である。
以上の実施形態に関し、更に以下の付記を開示する。上述した実施形態の一部又は全部は、以下に記載する(付記1)から(付記12)により表現することができるが、以下の記載に限定されるものではない。
第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成する、挙動解析部と、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成する、環境解析部と、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定する、推定部と、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを学習する学習部に、前記モデルを再学習させるための指示をする、学習指示部と、
を有する挙動推定装置。
付記1に記載の挙動推定装置であって、
前記学習部は、前記第一の挙動解析データと、第二の環境ごとに生成された第二の挙動解析データと、前記第一の環境及び前記第二の環境それぞれにおける位置ごとの地質特性の類似度とを用いて、前記モデルを学習する
挙動推定装置。
第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成する、挙動解析部と、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成する、環境解析部と、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定する、推定部と、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを学習する学習部に、前記モデルを再学習させるための指示をする、学習指示部と、
前記モデルが再学習された場合、再学習された前記モデルを用いて生成された挙動推定結果データに基づいて、現在位置から目的地までの移動経路を表す移動経路データを再生成する、移動経路生成部と、
を有する経路生成装置。
付記3に記載の経路生成装置であって、
前記学習部は、前記第一の挙動解析データと、第二の環境ごとに生成された第二の挙動解析データと、前記第一の環境及び前記第二の環境それぞれにおける位置ごとの地質特性の類似度とを用いて、前記モデルを学習する
経路生成装置。
第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成する、挙動解析ステップと、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成する、環境解析ステップと、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定する、推定ステップと、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを再学習させるための指示をする、学習指示ステップと、
を有する挙動推定方法。
付記5に記載の挙動推定方法であって、
前記第一の挙動解析データと、第二の環境ごとに生成された第二の挙動解析データと、前記第一の環境及び前記第二の環境それぞれにおける位置ごとの地質特性の類似度とを用いて、前記モデルを学習する
挙動推定方法。
第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成する、挙動解析ステップと、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成する、環境解析ステップと、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定する、推定ステップと、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを学習する学習部に、前記モデルを再学習させるための指示をする、学習指示ステップと、
前記モデルが再学習された場合、再学習された前記モデルを用いて生成された挙動推定結果データに基づいて、現在位置から目的地までの移動経路を表す移動経路データを再生成する、移動経路生成ステップと、
を有する経路生成方法。
付記7に記載の経路生成方法であって、
前記第一の挙動解析データと、第二の環境ごとに生成された第二の挙動解析データと、前記第一の環境及び前記第二の環境それぞれにおける位置ごとの地質特性の類似度とを用いて、前記モデルを学習する
経路生成方法。
コンピュータに、
第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成する、挙動解析ステップと、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成する、環境解析ステップと、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定する、推定ステップと、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを再学習させるための指示をする、学習指示ステップと、
処理を実行させる命令を含む、プログラムを記録しているコンピュータ読み取り可能な記録媒体。
付記9に記載のコンピュータ読み取り可能な記録媒体であって、
前記第一の挙動解析データと、第二の環境ごとに生成された第二の挙動解析データと、前記第一の環境及び前記第二の環境それぞれにおける位置ごとの地質特性の類似度とを用いて、前記モデルを学習する
コンピュータ読み取り可能な記録媒体。
コンピュータに、
第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成する、挙動解析ステップと、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成する、環境解析ステップと、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定する、推定ステップと、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを学習する学習部に、前記モデルを再学習させるための指示をする、学習指示ステップと、
前記モデルが再学習された場合、再学習された前記モデルを用いて生成された挙動推定結果データに基づいて、現在位置から目的地までの移動経路を表す移動経路データを再生成する、移動経路生成ステップと、
処理を実行させる命令を含む、プログラムを記録しているコンピュータ読み取り可能な記録媒体。
付記11に記載のコンピュータ読み取り可能な記録媒体であって、
前記第一の挙動解析データと、第二の環境ごとに生成された第二の挙動解析データと、前記第一の環境及び前記第二の環境それぞれにおける位置ごとの地質特性の類似度とを用いて、前記モデルを学習する
コンピュータ読み取り可能な記録媒体。
10 挙動推定装置
11 挙動解析部
12 学習部
13 環境解析部
14 推定部
15 学習指示部
16 移動経路生成部
17 再計画指示部
20 経路生成装置
30 計測部
31、32 センサ
40 記憶装置
50 移動体制御部
100 システム
110 コンピュータ
111 CPU
112 メインメモリ
113 記憶装置
114 入力インターフェイス
115 表示コントローラ
116 データリーダ/ライタ
117 通信インターフェイス
118 入力機器
119 ディスプレイ装置
120 記録媒体
121 バス
Claims (12)
- 第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成する、挙動解析手段と、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成する、環境解析手段と、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定する、推定手段と、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを学習する学習手段に、前記モデルを再学習させるための指示をする、学習指示手段と、
を有する挙動推定装置。 - 請求項1に記載の挙動推定装置であって、
前記学習手段は、前記第一の挙動解析データと、第二の環境ごとに生成された第二の挙動解析データと、前記第一の環境及び前記第二の環境それぞれにおける位置ごとの地質特性の類似度とを用いて、前記モデルを学習する
挙動推定装置。 - 第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成する、挙動解析手段と、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成する、環境解析手段と、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定する、推定手段と、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを学習する学習手段に、前記モデルを再学習させるための指示をする、学習指示手段と、
前記モデルが再学習された場合、再学習された前記モデルを用いて生成された挙動推定結果データに基づいて、現在位置から目的地までの移動経路を表す移動経路データを再生成する、移動経路生成手段と、
を有する経路生成装置。 - 請求項3に記載の経路生成装置であって、
前記学習手段は、前記第一の挙動解析データと、第二の環境ごとに生成された第二の挙動解析データと、前記第一の環境及び前記第二の環境それぞれにおける位置ごとの地質特性の類似度とを用いて、前記モデルを学習する
経路生成装置。 - 第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成し、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成し、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定し、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを再学習させるための指示をする
挙動推定方法。 - 請求項5に記載の挙動推定方法であって、
前記第一の挙動解析データと、第二の環境ごとに生成された第二の挙動解析データと、前記第一の環境及び前記第二の環境それぞれにおける位置ごとの地質特性の類似度とを用いて、前記モデルを学習する
挙動推定方法。 - 第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成し、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成し、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定し、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを再学習させるための指示をし、
前記モデルが再学習された場合、再学習された前記モデルを用いて生成された挙動推定結果データに基づいて、現在位置から目的地までの移動経路を表す移動経路データを再生成する
経路生成方法。 - 請求項7に記載の経路生成方法であって、
前記第一の挙動解析データと、第二の環境ごとに生成された第二の挙動解析データと、前記第一の環境及び前記第二の環境それぞれにおける位置ごとの地質特性の類似度とを用いて、前記モデルを学習する
経路生成方法。 - コンピュータに、
第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成し、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成し、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定し、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを再学習させるための指示をする
処理を実行させる命令を含む、プログラムを記録しているコンピュータ読み取り可能な記録媒体。 - 請求項9に記載のコンピュータ読み取り可能な記録媒体であって、
前記第一の挙動解析データと、第二の環境ごとに生成された第二の挙動解析データと、前記第一の環境及び前記第二の環境それぞれにおける位置ごとの地質特性の類似度とを用いて、前記モデルを学習する
コンピュータ読み取り可能な記録媒体。 - コンピュータに、
第一の環境において移動体の実際の挙動を表す第一の挙動解析データを生成し、
前記第一の環境の状態を表す環境状態データに基づいて前記第一の環境について解析をし、環境解析データを生成し、
前記環境解析データを、前記第一の環境における移動体の挙動を推定するためのモデルに入力して、前記第一の環境における前記移動体の挙動を推定し、
前記モデルにより推定された挙動推定結果データに基づいて信頼区間を設定し、設定した前記信頼区間に前記第一の挙動解析データが存在する場合、前記モデルを再学習させるための指示をし、
前記モデルが再学習された場合、再学習された前記モデルを用いて生成された挙動推定結果データに基づいて、現在位置から目的地までの移動経路を表す移動経路データを再生成する
処理を実行させる命令を含む、プログラムを記録しているコンピュータ読み取り可能な記録媒体。 - 請求項11に記載のコンピュータ読み取り可能な記録媒体であって、
前記第一の挙動解析データと、第二の環境ごとに生成された第二の挙動解析データと、前記第一の環境及び前記第二の環境それぞれにおける位置ごとの地質特性の類似度とを用いて、前記モデルを学習する
コンピュータ読み取り可能な記録媒体。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US18/032,864 US20240019250A1 (en) | 2020-10-29 | 2020-10-29 | Motion estimation apparatus, motion estimation method, path generation apparatus, path generation method, and computer-readable recording medium |
PCT/JP2020/040670 WO2022091305A1 (ja) | 2020-10-29 | 2020-10-29 | 挙動推定装置、挙動推定方法、経路生成装置、経路生成方法、及びコンピュータ読み取り可能な記録媒体 |
JP2022558717A JP7444277B2 (ja) | 2020-10-29 | 2020-10-29 | 挙動推定装置、挙動推定方法、経路生成装置、経路生成方法、及びプログラム |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2020/040670 WO2022091305A1 (ja) | 2020-10-29 | 2020-10-29 | 挙動推定装置、挙動推定方法、経路生成装置、経路生成方法、及びコンピュータ読み取り可能な記録媒体 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022091305A1 true WO2022091305A1 (ja) | 2022-05-05 |
Family
ID=81382085
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2020/040670 WO2022091305A1 (ja) | 2020-10-29 | 2020-10-29 | 挙動推定装置、挙動推定方法、経路生成装置、経路生成方法、及びコンピュータ読み取り可能な記録媒体 |
Country Status (3)
Country | Link |
---|---|
US (1) | US20240019250A1 (ja) |
JP (1) | JP7444277B2 (ja) |
WO (1) | WO2022091305A1 (ja) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115220449A (zh) * | 2022-07-14 | 2022-10-21 | 小米汽车科技有限公司 | 路径规划的方法、装置、存储介质、芯片及车辆 |
WO2024069781A1 (ja) * | 2022-09-28 | 2024-04-04 | 日産自動車株式会社 | 車両の走行支援方法及び車両の走行支援装置 |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH1074188A (ja) * | 1996-05-23 | 1998-03-17 | Hitachi Ltd | データ学習装置およびプラント制御装置 |
JP2011150627A (ja) * | 2010-01-25 | 2011-08-04 | Ihi Corp | 通路検出方法、装置、及びプログラム |
JP2018529175A (ja) * | 2015-08-28 | 2018-10-04 | インペリアル・カレッジ・オブ・サイエンス・テクノロジー・アンド・メディスン | 多方向カメラを使用した空間のマッピング |
WO2019176354A1 (ja) * | 2018-03-13 | 2019-09-19 | 住友電気工業株式会社 | 学習用データ収集方法、学習用データ収集装置、異変検知システム及びコンピュータプログラム |
JP2019529898A (ja) * | 2016-09-16 | 2019-10-17 | ポラリス インダストリーズ インコーポレーテッド | ルート計画計算デバイスを改良するためのデバイス及び方法 |
JP2020067980A (ja) * | 2018-10-26 | 2020-04-30 | 富士通株式会社 | 予測プログラム、予測方法及び予測装置 |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP6553148B2 (ja) | 2017-10-05 | 2019-07-31 | ヤフー株式会社 | 判定装置、判定方法及び判定プログラム |
-
2020
- 2020-10-29 US US18/032,864 patent/US20240019250A1/en active Pending
- 2020-10-29 JP JP2022558717A patent/JP7444277B2/ja active Active
- 2020-10-29 WO PCT/JP2020/040670 patent/WO2022091305A1/ja active Application Filing
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH1074188A (ja) * | 1996-05-23 | 1998-03-17 | Hitachi Ltd | データ学習装置およびプラント制御装置 |
JP2011150627A (ja) * | 2010-01-25 | 2011-08-04 | Ihi Corp | 通路検出方法、装置、及びプログラム |
JP2018529175A (ja) * | 2015-08-28 | 2018-10-04 | インペリアル・カレッジ・オブ・サイエンス・テクノロジー・アンド・メディスン | 多方向カメラを使用した空間のマッピング |
JP2019529898A (ja) * | 2016-09-16 | 2019-10-17 | ポラリス インダストリーズ インコーポレーテッド | ルート計画計算デバイスを改良するためのデバイス及び方法 |
WO2019176354A1 (ja) * | 2018-03-13 | 2019-09-19 | 住友電気工業株式会社 | 学習用データ収集方法、学習用データ収集装置、異変検知システム及びコンピュータプログラム |
JP2020067980A (ja) * | 2018-10-26 | 2020-04-30 | 富士通株式会社 | 予測プログラム、予測方法及び予測装置 |
Non-Patent Citations (6)
Title |
---|
BO ZHOU ; KUN QIAN ; XUDONG MA ; XIANZHONG DAI: "Probabilistic terrain modeling of mobile robots for outdoor applications using a scanning laser", MECHATRONICS AND AUTOMATION (ICMA), 5 August 2012 (2012-08-05), pages 1716 - 1721, XP032224580, ISBN: 978-1-4673-1275-2, DOI: 10.1109/ICMA.2012.6284395 * |
FANKHAUSER PETER, BLOESCH MICHAEL, HUTTER MARCO: "Probabilistic Terrain Mapping for Mobile Robots With Uncertain Localization", IEEE ROBOTICS AND AUTOMATION LETTERS, vol. 3, no. 4, 1 October 2018 (2018-10-01), pages 3019 - 3026, XP055872229, DOI: 10.1109/LRA.2018.2849506 * |
FUJII, AKINOBU : "AC1H1-03 Automatic map update system based on the probability of deviation occurrence between the map and the measured value", PROCEEDINGS OF THE 35TH ANNUAL CONFERENCE OF THE RSJ (ROBOTICS SOCIETY OF JAPAN); TOKYO, JAPAN; SEPTEMBER 9-14, 2017, vol. 35, September 2017 (2017-09-01), pages 1 - 4, XP009538128 * |
MASHA DITEBOGO; BURKE MICHAEL; TWALA BHEKISIPHO: "Slip estimation methods for proprioceptive terrain classification using tracked mobile robots", 2017 PATTERN RECOGNITION ASSOCIATION OF SOUTH AFRICA AND ROBOTICS AND MECHATRONICS (PRASA-ROBMECH), 30 November 2017 (2017-11-30), pages 150 - 155, XP033300808, DOI: 10.1109/RoboMech.2017.8261139 * |
MOROMOTO, JUNYA; HAYASHI, TAKUYA; FUJIMOTO, HIROYUKI; ISHI, HIROYUKI; OHYA, JUN; YAMATO, JUNJI; TAKANISHI ATSUO: "S1-3 Study of the Path Planning Based on SLAM and Semantic Segmentation by Deep Learning for Forest Environmental Monitoring Robot", VISUAL/MEDIA COMPUTING CONFERENCE 2018 PREPRINTS OF THE 46TH ANNUAL CONFERENCE OF THE INSTITUTE OF IMAGE ELECTRONICS ENGINEERS OF JAPAN; 2018/06/21 - 2018/06/23, vol. 46, 30 November 2017 (2017-11-30) - 23 June 2018 (2018-06-23), pages 1 - 5, XP009538127, ISSN: 2436-4371, DOI: 10.11371/aiieej.46.0_3 * |
ZHOU RUYI; DING LIANG; GAO HAIBO; FENG WENHAO; DENG ZONGQUAN; LI NAN: "Mapping for Planetary Rovers from Terramechanics Perspective*", 2019 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS), 3 November 2019 (2019-11-03), pages 1869 - 1874, XP033695554, DOI: 10.1109/IROS40897.2019.8967984 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115220449A (zh) * | 2022-07-14 | 2022-10-21 | 小米汽车科技有限公司 | 路径规划的方法、装置、存储介质、芯片及车辆 |
CN115220449B (zh) * | 2022-07-14 | 2023-11-21 | 小米汽车科技有限公司 | 路径规划的方法、装置、存储介质、芯片及车辆 |
WO2024069781A1 (ja) * | 2022-09-28 | 2024-04-04 | 日産自動車株式会社 | 車両の走行支援方法及び車両の走行支援装置 |
Also Published As
Publication number | Publication date |
---|---|
JPWO2022091305A1 (ja) | 2022-05-05 |
JP7444277B2 (ja) | 2024-03-06 |
US20240019250A1 (en) | 2024-01-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Scherer et al. | River mapping from a flying robot: state estimation, river detection, and obstacle mapping | |
US20200042656A1 (en) | Systems and methods for persistent simulation | |
WO2022091305A1 (ja) | 挙動推定装置、挙動推定方法、経路生成装置、経路生成方法、及びコンピュータ読み取り可能な記録媒体 | |
JP2017004373A (ja) | 情報処理装置、情報処理プログラム、および情報処理システム | |
JP6856575B2 (ja) | 能動的探索なしの強化学習に基づく制御ポリシー学習及び車両制御方法 | |
CN111145251A (zh) | 一种机器人及其同步定位与建图方法和计算机存储设备 | |
CN110488818A (zh) | 一种基于激光雷达的机器人定位方法、装置和机器人 | |
CN114442621A (zh) | 一种基于四足机器人的自主探索和建图系统 | |
CN110929402A (zh) | 一种基于不确定分析的概率地形估计方法 | |
Schwendner et al. | Using embodied data for localization and mapping | |
CN115639823A (zh) | 崎岖起伏地形下机器人地形感知与移动控制方法及系统 | |
EP3732549A1 (en) | Method for predicting a motion of an object, method for calibrating a motion model, method for deriving a predefined quantity and method for generating a virtual reality view | |
McAllister et al. | Motion planning and stochastic control with experimental validation on a planetary rover | |
Elisha et al. | Active online visual-inertial navigation and sensor calibration via belief space planning and factor graph based incremental smoothing | |
WO2020050084A1 (ja) | 誘導制御プログラムを格納した記憶媒体 | |
Ugur et al. | Fast and efficient terrain-aware motion planning for exploration rovers | |
Inotsume et al. | Adaptive terrain traversability prediction based on multi-source transfer gaussian processes | |
WO2022034679A1 (ja) | 挙動学習装置、挙動学習方法、挙動推定装置、挙動推定方法、及びコンピュータ読み取り可能な記録媒体 | |
Toscano-Moreno et al. | DEM-AIA: Asymmetric inclination-aware trajectory planner for off-road vehicles with digital elevation models | |
CN115828463A (zh) | 基于数字孪生六维模型的掘进机位姿估计方法、系统 | |
WO2023283561A1 (en) | An imu based system for vertical axis joint angle estimation for swing boom excavators | |
JP2022174734A (ja) | 建設現場用のオフロード車両のための方策を学習するための装置および方法 | |
Jaroszek et al. | Localization of the wheeled mobile robot based on multi-sensor data fusion | |
Agishev et al. | MonoForce: Self-supervised learning of physics-aware grey-box model for predicting the robot-terrain interaction | |
Odry et al. | Adaptive Gradient-Descent Extended Kalman Filter for Pose Estimation of Mobile Robots with Sparse Reference Signals |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 20959823 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2022558717 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 18032864 Country of ref document: US |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 20959823 Country of ref document: EP Kind code of ref document: A1 |