JP2024070849A - Method and device for operating earth-moving machine, and earth-moving machine - Google Patents

Abstract

To provide a method for operating an earth-moving machine (100).SOLUTION: A method for operating an earth-moving machine includes a step of reading a terrain signal (215), a material signal (220) and a machine signal (225), wherein the terrain signal (215) indicates a map where a current contour of a terrain is detected, the material signal (220) indicates at least one material parameter of a material detected in a terrain (105), and the machine signal (225) indicates at least one machine parameter of an earth-moving machine (100). The method includes the steps of: creating a geometric model (227) of a terrain using the terrain signal (215) and creating the geometric model (227) as a voxel network; simulating a material flow of a material using the geometric model (227) and the material signal (220); and outputting a control signal (210) for controlling the earth-moving machine (100) on the basis of the simulated material flow, using the material signal (225) so as to re-distribute the material.SELECTED DRAWING: Figure 2

Description

The invention relates to a method for operating an earth-moving machine, a device and an earth-moving machine of the type according to the preambles of the independent claims. The subject of the invention is also a computer program.

The creation of a flat surface can be achieved by one bulldozer moving the soil or by several bulldozers moving the soil independently of each other. For this purpose, the machine operator receives verbal instructions or CAD data of the surface to be obtained. The CAD data can be communicated to the assistance system, which can then use the geolocation information to create the final terrain contour. The machine operator can then first manually, with the assistance system active, scrape off or distribute such an amount of soil that such an amount of soil remains on the target contour that in the final work step the target contour is obtained.

Disclosure of the invention Given this background, the approach presented herein provides a method for operating an earth-moving machine as set forth in the independent claims, as well as an apparatus using this method, an earth-moving machine and finally a corresponding computer program. Advantageous developments and improvements of the apparatus as set forth in the independent claims are realizable by the measures set forth in the dependent claims.

The present invention advantageously allows for determining an optimal distribution strategy for a given pile, a given topology or surface geometry and a bulldozer. Thus, advantageously, time required can be minimized, fuel consumption can be minimized and machine downtime can be avoided.

A method for the operation of an earthmoving machine is presented, the method comprising the steps of reading a terrain signal, a material signal and a machine signal, where the terrain signal represents a detected map of a current contour of the terrain, the material signal represents at least one material parameter of a material detected in the terrain, and the machine signal represents at least one machine parameter of the earthmoving machine. The method further comprises the steps of creating a geometric model of the terrain using the terrain signal, the geometric model being created as a voxel network, simulating a material flow of the material using the geometric model and the material signal, and outputting a control signal for controlling the earthmoving machine based on the simulated material flow and using the machine signal to redistribute the material. Here, the geometric model may in particular be configured as a spatial configuration model. The geometric model preferably represents a surface, in particular a volume of a near-surface area of the terrain. The geometric model preferably includes a number of elements or components representing a near-surface volume of the terrain. Near-surface may in particular be understood as an area of the terrain that influences the material flow of the material present on the terrain.

The earthworking machine may be, for example, a vehicle for processing the ground, for example for leveling, loosening, loading, transporting or compacting earth lumps or bulk materials. For example, it may be a bulldozer, an excavator or a wheel loader. The earthworking machine may, for example, include a device for performing the method described herein, or the corresponding device may be arranged outside the machine, and the earthworking machine may be configured for receiving and additionally or alternatively providing electronic signals for performing the method. This allows the reading of the terrain signal, the material signal and the machine signal. For example, to provide the terrain signal, firstly the basic spatial configuration of the terrain to be processed by the earthworking machine may be detected. This can be done, for example, via a sensor system on the machine (multiple cameras and/or LiDAR (Lidar)) or by using a drone flyover with a comparable or high resolution 3D sensor system, or also by using other measuring devices. The terrain signal can be used to provide a map of this terrain for the machine. In a similar manner, for example, one or more material parameters of a material detected in the terrain, such as the ground or a bulk pile, can be detected and read in using the material signal. The material parameters can be, for example, the shape, flow behavior or adhesion of the material. In contrast, machine parameters of an earthmoving machine, such as the type, size or drive force of the machine used, can be read in directly, for example, from a corresponding unit of the machine or from an external database.

Based on the data read in with the terrain signal, a voxel network of the spatial arrangement or surface of the terrain is constructed. A voxel (short for volumetric element) is to be understood as the smallest building block that is characterized via its physical relationship to its neighbors (pressure, adhesion, gravity), its physical properties (color, mass, reflectance, transmittance) and shape. Thus, three-dimensional structures of any shape can be represented, similar to pixels (short for picture elements). This is an advantage over height maps, for example, which cannot represent interrelationships univocally. An important advantage is that voxels are increasingly used in the field of virtual reality computer simulations and computer games, so that dedicated algorithms and hardware accelerators also exist for processing voxel structures at high iteration rates. Furthermore, there are techniques for adaptively designing voxel network structures, i.e. increasing the resolution in particularly relevant areas and decreasing the resolution in other areas. These are controllable via the differences between the interactions in the individual voxels. For example, if all nine voxels of a cube experience equal forces, they can be merged together. If the forces of voxels bordering each other are significantly different, they can be further divided to allow for more accurate tracking of force vectors and material flow in the voxel network, thus saving additional computational costs. In the method presented here, a voxel network can be described by three variables present in the voxel: density, the velocity vector in which the voxel moves, and the force vector that accelerates or brakes the voxel. Collectively, these can be referred to as the density field, the velocity field, and the force field.

The material flow of the material is then simulated using the geometric model and the material signals. For this, for example, a simulation of a voxel network can be initialized with the estimated values. In this case, the basic spatial configuration and the deposit can be simulated separately from each other, or, in simplification, the same material parameters can be used for both. This should reduce the estimation error, since mainly the deposit is distributed and the basic spatial configuration is not mainly scraped off. If this assumption is incorrect, then they should be simulated separately.

Finally, based on the simulated material flow and using the machine signals, in an outputting step, a control signal is output, which allows optimal drive control of the earthmoving machine for the material to be processed, the current terrain, and the task to be accomplished, and allows for redistribution of material in response to predefined work instructions.

According to an embodiment, the method may comprise a step of determining the material parameters using the machine signal and additionally or alternatively the topographical signal. In this case, in the reading step, the material signal may represent the determined material parameters. For example, the material parameters of the pile may vary continuously, for example with the moisture content. Based on very steep collapses or the observed flow behavior, for example the flow angle and adhesion force, can be estimated, for example via LiDAR and also via the external geometry detectable additionally or alternatively via a camera or another sensor system that can provide a 3D point cloud. This can be done, for example, during the unloading process from a truck, if the truck is in the sensor's reach. This provides the advantage that the material flow simulation carried out using the material signal can be optimally carried out. Furthermore, via grain size, reflectivity, texture and color, for example a neural network with fuzzy logic can estimate the material parameters based on a parameter library. Fuzzy neurons can be chosen here, since they advantageously allow good predictions with uncertain parameters even in the case of a small network depth.

Furthermore, in the determining step, at least one material parameter can be estimated using parameters stored in a database. For example, the mass can be determined via the power consumption of the machine during the pile movement, and via measurements the volume and thus the density of the material can be determined. If an estimation is still not possible because an extrusion process has not been carried out, reference values from a geographical database or other measurements can be used. This provides the advantage that the material parameter can already be determined technically easily and energy-savingly before the physical work is carried out by the earthmoving machine.

According to another embodiment, the method may include a step of determining a trajectory for the movement of the earthmoving machine. In this case, in the outputting step, the selected trajectory may be used to output a control signal. For example, to drive and control the distribution of the pile, a trajectory that can result in maximum leveling can first be selected. Here, the criterion for the selection of the trajectory may be that the distribution radius with the distributed material results in a very small height difference. Here, in order to save calculation time, a simplified model can be used in which only a static distribution of the inelastic material without internal flow processes can be considered. This allows advantageously to optimize and output a control signal for the planned work. For example, when distributing the pile via a side change, the distribution can be uniform. However, this may vary based on a material flow simulation.

According to another embodiment, the method may include a step of simulated distribution of material using the simulated material flow and the selected trajectory to obtain a simulated distribution result. Here, in the outputting step, the simulated distribution result can be used to output a control signal. For example, a voxel simulation can be used to simulate the distribution step based on the selected trajectory. In this case, the internal flow processes in the velocity and force fields can also be taken into account. This provides the advantage that the earthmoving machine can be driven and controlled to obtain the best possible distribution result at the lowest possible time and energy costs.

According to another embodiment, the method may further comprise a step of loading optimization parameters, which may represent targets for the distribution of the material. In this case, in the evaluation step, the simulated distribution result can be evaluated using the optimization parameters. For example, a predefined threshold value, for example for the maximum fuel consumption or the working speed, can be loaded by the user interface. After the simulation of the distribution of the material, the distribution step can be evaluated multi-criteria, for example taking into account such and further optimization parameters. On the one hand, the criterion of maximum leveling can be used. On the other hand, other quality criteria, for example the fuel consumption required for the movement of the material or the time required for distribution, can be evaluated, which advantageously allows an optimized approach for a given work order. If no improvement is shown compared to the previous simulation, changes can be made again, returning to the last trajectory. For the initialization of a step, the trajectory from the previous step can be decomposed into two trajectories, which is done by changing it to the left and to the right by a predefined delta. Through such an approach, advantageously, a trajectory representing a local optimum can be found. If the optimization criterion exceeds a predefined limit, the loop can be left and the state obtained in this simulation can be selected as the basis for the subsequent distribution steps.

According to another embodiment, if the evaluating step evaluates that the simulated distribution of the material does not correspond to the optimized parameters, the identifying step, the simulated distribution step, the further reading step and the evaluating step can be repeated. Here, if the evaluating step evaluates that the simulated distribution of the material corresponds to the optimized parameters, a control signal can be output in the outputting step. For example, based on the first step, a subsequent distribution step can be simulated. This approach advantageously makes it possible to avoid that a local optimum leads to a deadlock, from which subsequent steps only worsen the results. Based on the available computing power and computing time, for example, a maximum number of prediction steps can be defined. If the local optimum turns out to be a saddle point, the trajectory of the first distribution step can be reinitialized, for example by starting at the other end of the pile, in order to achieve the largest possible difference.

According to another embodiment, the reading and creating steps can be repeated after the outputting step in order to obtain an updated geometric model. Here, in the comparing step, the updated geometric model can be compared with the simulated distribution result, and the determining step can be repeated if a predefined iteration criterion is met. For example, a comparison of the material redistribution thus obtained with the simulated distribution can be carried out after the work step carried out by the machine, materially, using the control signal. For this, in particular, the terrain signal can be read in again, so that an updated geometric model can be created. If the currently detected spatial configuration corresponds to the simulated spatial configuration, i.e. the target, within a tolerance range, i.e. within a predefined limit value, further work is likewise possible. If no match is confirmed, i.e. for example in the case of an upper or lower exceedance of a limit value, this can be a criterion for repeating the determining step and selecting a new trajectory.

According to another embodiment, the method may include a step of again specifying another trajectory for a further movement of the earthmoving machine and a step of again simulating distribution of the material using the simulated material flow and the selected another trajectory, whereby another simulated distribution result is obtained. The method may further comprise a step of again further reading in the optimization parameters and additionally or alternatively another optimization parameter, which may represent another target for processing the material. In this case, in a step of again evaluation, another simulated distribution result may be evaluated using the other optimization parameters. The repetition of the described steps provides the advantage that the drive control of the earthmoving machine can be continuously adapted and optimized to the changed terrain parameters and material parameters based on the executed work steps.

The method may be implemented, for example, in software or hardware, or in a mixed form of software and hardware, for example in a control device.

The approach presented herein further provides an apparatus configured to execute, drive or realize the steps of the method variants presented herein in a corresponding apparatus. This variant of the invention in the form of an apparatus also allows for a rapid and effective solution of the problem on which the invention is based.

For this purpose, the device may have at least one calculation unit for processing signals or data, at least one memory unit for storing signals or data, at least one interface to a sensor or actuator for reading a sensor signal from a sensor or for outputting a data signal or a control signal to an actuator, and/or at least one communication interface for reading or outputting data embedded in the communication protocol. The calculation unit may be, for example, a signal processor, a microcontroller, etc., in which case the memory unit may be a flash memory or a magnetic memory unit. The communication interface may be configured to read or output data wirelessly and/or wired, and a communication interface capable of reading or outputting wired data may read or output these data, for example electrically or optically, from a corresponding data transmission line or output them to a corresponding data transmission line.

Here, a device may be understood as an electrical device that processes sensor signals and outputs control and/or data signals accordingly. The device may have an interface that may be configured on the basis of hardware and/or software. In the case of a hardware-based configuration, the interface may for example be part of a so-called ASIC system, which includes the various functions of the device. However, it is also possible for the interface to be a dedicated integrated circuit or to consist at least partially of separate components. In the case of a software-based configuration, the interface may be a software module, which is for example present on a microcontroller alongside other software modules.

Furthermore, an earthmoving machine is presented, which comprises a variant of the device presented above. For example, the earthmoving machine may be configured for planning the terrain or for compacting the ground structure, and for this purpose may be configured for carrying out a variant of the method presented above. This combination of the earthmoving machine and the device also allows the above-mentioned advantages to be optimally realized.

Also advantageous is a computer program product or a computer program having a program code which can be stored on a machine-readable carrier or on a machine-readable storage medium such as a semiconductor memory, a hard disk memory or an optical memory and which is used to execute, realize and/or drive the steps of the method according to one of the above-mentioned embodiments, in particular when the program product or program is executed on a computer or on an apparatus. In particular, a computer program comprising instructions for causing a computer to carry out a method according to one of the above-mentioned embodiments and/or for causing an apparatus to carry out a method according to one of the above-mentioned embodiments, when the program is executed by a computer, is advantageous.

Examples of the approach presented herein are shown in the drawings and described in more detail in the following specification.

FIG. 1 shows a schematic plan view of an earthmoving machine according to one embodiment. FIG. 1 shows a schematic plan view of an earthmoving machine according to one embodiment. FIG. 1 shows a schematic plan view of an earthmoving machine according to one embodiment. FIG. 1 shows a schematic plan view of an earthmoving machine according to one embodiment. FIG. 1 shows a schematic plan view of an earthmoving machine according to one embodiment. FIG. 2 shows a block circuit diagram of an apparatus for controlling a method for the operation of an earthmoving machine. FIG. 1 illustrates a flow chart of one embodiment of a method for operation of an earthmoving machine. FIG. 1 illustrates a flow chart of one embodiment of a method for operation of an earthmoving machine. FIG. 1 illustrates a schematic diagram of an example of a simulation of distribution of material by a simulated earthmoving machine. FIG. 2 illustrates a schematic diagram of a geometric model according to one embodiment. FIG. 2 illustrates a schematic diagram of a geometric model according to one embodiment. FIG. 2 illustrates a schematic diagram of a geometric model according to one embodiment.

In the following description of the preferred embodiments of the present invention, the same or similar reference numerals are used for elements having similar functions shown in the various figures, and repeated descriptions of these elements are omitted.

1A, 1B, 1C, 1D, and 1E each show a schematic plan view of an earthmoving machine 100 according to one embodiment. In this embodiment, the earthmoving machine 100 is configured as a bulldozer to level a terrain 105 or redistribute material to create a flat target terrain 110.

Here, in the individual figures, an exemplary flow for distributing the deposit in five steps is shown. Through side changes, the distribution can be uniform. However, this may vary based on material flow simulations. That is, first of all, the trajectory 115 that results in the greatest leveling is selected. That is, the criterion for the selection of the trajectory is that the distribution radius with the distributed material results in a very small height difference. Here, in one embodiment, in order to save calculation time, a simplified model is used in which only the static distribution of inelastic material without internal flow processes is considered.

Figure 2 shows a block diagram of an apparatus 200 for controlling a method for the operation of an earth moving machine 100. The drive-controllable earth moving machine 100 shown in Figure 2 corresponds to or is similar to the earth moving machines described in the preceding figures and is configured to read control signals 210 available from the apparatus 200 and to act autonomously using the control signals 210.

To drive and control the earthmoving machine, the device 200 is configured to read a terrain signal 215, a material signal 220 and a machine signal 225, where the terrain signal 215 represents a detected map of the current contour of the terrain and in this embodiment may be provided by a sensor device 230 for detecting the terrain by sensor technology. Using the terrain signal 215, the device 200 is configured to create a geometric model 227 of the terrain, where the geometric model may be created as a voxel network.

The material signal 220 represents at least one material parameter of the material detected in the terrain, which in this example may be provided by an external database 235. The parameters stored in the database 235 may be used to exemplarily estimate the at least one material parameter.

The machine signal 225 represents at least one machine parameter of the earthwork machine 100 and may be provided, by way of example only, by a unit 240 of the earthwork machine 100.

Furthermore, the apparatus 200 is configured to simulate a material flow of the material using the geometric model 227 and the material signal 220. A control signal 210 can be output based on this simulated material flow and using the machine signal to redistribute the material.

That is, the device 200 allows for determining the optimal distribution strategy for a given pile, a given spatial configuration, and a bulldozer, thus minimizing turnaround time, minimizing fuel consumption, and avoiding machine downtime.

Figure 3 shows a flow chart of one embodiment of a method 300 for operating an earthmoving machine. The method 300 shown in Figure 3 can be implemented, for example, using an apparatus such as that described in the preceding Figure 2 and for driving and controlling an earthmoving machine such as that described in the preceding Figures.

The method 300 includes reading 305 a terrain signal, a material signal, and a machine signal, where the terrain signal represents a detected map of a current contour of the terrain, the material signal represents at least one material parameter of a material detected in the terrain, and the machine signal represents at least one machine parameter of the earthmoving machine.

Further, the method 300 includes a step 310 of creating a geometric model of the terrain using the terrain signal, where the geometric model is created as a voxel network.

Further, the method 300 includes simulating 315 a material flow of the material using the geometric model and the material signal, and outputting 320 a control signal for controlling an earthmoving machine based on the simulated material flow and using the machine signal to redistribute the material.

Figure 4 illustrates a flow chart of one embodiment of a method 300 for operating an earthmoving machine. The method 300 illustrated in Figure 4 corresponds to or is similar to the method described in the preceding Figure 3, but differs in that it includes additional steps.

After the merely exemplary start step 400 of the method, in this embodiment, in a checking step 402, it is checked using the terrain signal whether the material pile has already been unloaded by the vehicle. Subsequently, in this embodiment, a step 405 of determining material parameters is performed, where step 405 is shown merely exemplary in three partial steps 405a, 405b, 405c. If the material has already been unloaded, in partial step 405a, the material parameters are determined using the machine signal and the terrain signal, where in one embodiment, the geometry of the pile is detected, for example, by a camera and LiDAR. If the material has not yet been unloaded, in an alternative partial step 405b, the material parameters are determined during unloading of the material, using the machine signal and the terrain signal, where in one embodiment, the flow behavior is detected, for example, by a camera and LiDAR. Sub-step 405a or 405b is followed in this embodiment by sub-step 405c, in which a material hypothesis is made from the color and texture or the maximum flow angle of the deposition volume is estimated. Additionally or alternatively, in the determining step 405, at least one material parameter may be estimated using parameters stored in a database.

Then, in a subsequent reading step 305 in this embodiment, the material signal represents the material parameters determined in the preceding step 405.

As described above in FIG. 3, in the illustratively following creating step 310 and simulating step 315, a geometric model is created as a voxel network and material flow is simulated.

In this embodiment, the method 300 further includes step 410 of identifying a trajectory for movement of the earthmoving machine, illustratively in which a trajectory along which the machine should be moved to dispense the material is selected.

This is followed, in one embodiment, by a step 415 of simulated distribution of material using the simulated material flow and the selected trajectory to obtain a simulated distribution result.

In this embodiment, the method 300 further comprises a step 420 of further reading in optimization parameters, which represent targets for the distribution of the material. Here, merely by way of example, the optimization criteria are read from a user input. Using the optimization parameters, merely by way of example, the simulated distribution result is evaluated in an evaluation step 425. Here, merely by way of example, it is checked whether the read in optimization criteria has been improved or whether a predefined limit value is exceeded.

If the evaluating step 425 determines that the simulated distribution of the material does not correspond to the optimized parameters, for example because no refinement of the optimization criterion has been performed or because the optimization criterion has not reached a limiting value, then in one embodiment the identifying step 410, the simulated distribution step 415, the further reading step 420, and the evaluating step 425 are repeated. If the evaluating step 425 determines that the simulated distribution of the material does correspond to the optimized parameters, then the method 300 continues in this embodiment as described below.

Merely by way of example, the method 300 shown in FIG. 4 further comprises a step 430 of re-identification of another trajectory for a further movement of the earthmoving machine. This is followed in one embodiment by a step 435 of re-simulated distribution of the material using the simulated material flow and the selected other trajectory to obtain another simulated distribution result. This is followed by a step 440 of further reading in, and for example, another optimization parameter, which represents another target for processing the material, again, by way of example only, whereby, using the other optimization parameter, in this embodiment, in a step 445 of re-evaluation, another simulated distribution result is evaluated, whereby, if in the step 445 of re-evaluation, it is evaluated that the re-simulated distribution of the material does not correspond to this or the other optimization parameter, the steps 430 of re-identification, 435 of re-simulated distribution, 440 of further reading in, and 445 of re-evaluation are repeated.

In one embodiment, the step of re-evaluation 445 is followed by a step of checking 450, in which, merely by way of example, it is checked whether a maximum number of prediction steps has been reached. If the exemplarily predetermined maximum number has not yet been reached, then in this embodiment, the steps of re-simulated distribution 435, re-further reading 440 and re-evaluation 445 are repeatedly performed. If this maximum number has been reached, then in this embodiment, a step of outputting a control signal 320 is performed in order for the earthmoving machine to perform a distribution of material according to the previously simulated distribution result.

...

In other words, the method 300 shown in Figures 3 and 4 can be described as follows:

To achieve the Pareto optimum, a cost function is minimized by estimating the material parameters of the deposits, measuring the spatial configuration, considering the machine parameters, and performing iterative optimization based on simulation.

For this, first the basic spatial configuration of the terrain is detected and provided to the machine. This can be done via a sensor system on the machine (multiple cameras and/or LiDAR) or using a drone overflying with a comparable or higher resolution 3D sensor system.

Based on this data, a spatially arranged voxel network is constructed.

A voxel network can be described by three variables present within the voxel: density, the velocity vector in which the voxel moves, and the force vector that accelerates/brakes the voxel. Collectively, these can be referred to as the density field, velocity field, and force field. The assumption here is that the material is homogeneous, so density is sufficient for this description.

The material parameters of the sediment are constantly estimated. They change continuously, for example due to the moisture content. Through the external geometry detected via LiDAR and/or a camera or another sensor system providing a 3D point cloud, the flow angle and adhesion force can be estimated based on very steep collapses or the observed flow behavior. This may already be done during the unloading process from the truck, if the truck is in the sensor's range. Furthermore, through the grain size, reflectivity, texture and color, a neural network with fuzzy logic can estimate the material parameters based on a parameter library. Fuzzy neurons are chosen here because they allow good predictions with uncertain parameters even at low network depths.

The mass can be determined via the power consumption of the machine during the movement of the pile, and via measurements the volume of the material and therefore its density. If an estimation is still not possible because the extrusion process was not carried out, reference values from a geographical database or other measurements can be used.

The simulation of the voxel network is now initialized with the estimated values. In this case, the basic spatial configuration and the deposit can be simulated separately from each other, or, to simplify, the same material parameters can be used for both. This should reduce the estimation error, since mainly the deposit is distributed and the basic spatial configuration is not mainly scraped off. If this assumption is wrong, they should be simulated separately.

Through the side changes, it is ensured that the distribution is uniform. However, this may be changed based on the material flow simulation. That is, first of all, the trajectory that results in the maximum leveling is selected. The criterion for the selection of the trajectory is that it results in a very small height difference in the distribution radius with the distributed material. Here, in order to save calculation time, a simplified model is used, whereby only the static distribution of the inelastic material without internal flow processes is considered.

Then, using voxel simulation, the distribution step is simulated based on the selected trajectory, taking into account the internal flow processes in the velocity and force fields.

After the simulation, the distribution step is evaluated on multiple criteria. On the one hand, the criterion of maximum evenness is used. On the other hand, other quality criteria can be evaluated, for example the fuel consumption required for the movement of the material or the time required for distribution.

If no improvement is shown compared to previous simulations, we go back to the last trajectory and make changes again.

For the initialization of a step, the trajectory from the previous step is decomposed into two trajectories by changing it to the left and to the right by a predefined delta. Through this method, a trajectory that represents a local optimum should be found.

If the optimization criterion exceeds a predefined limit, the loop is left and the state obtained in this simulation is selected as the basis for the subsequent distribution steps.

Now, based on the first step, the following distribution steps are simulated. This approach should eliminate local optima that lead to deadlocks, from which the following steps will only worsen the results. Based on the available computing power and computing time, the maximum number of prediction steps is determined.

If the local optimum is found to be a saddle point, the trajectory for the first dispensing step is reinitialized by starting at the other end of the pile to achieve the largest possible difference.

Based on the determined trajectory, the distribution step is then physically carried out. Here, the actual result is compared with the simulation result. Via a predefined limit of agreement, it is determined whether the material parameters should be re-estimated or whether a direct transition to the next trajectory can be made. If the simulation is highly accurate, this trajectory can be taken over directly from the previous step, since in this case the next distribution step has already been evaluated. Thus, the computational costs after the initial cost can be reduced. If the parameter estimation is poor, the entire simulation, including the subsequent steps, must be re-estimated.

Now, if the pile is distributed so that no further improvement can be achieved by further distribution steps, an interruption is made. Rather, a simulation of a certain number of predictive steps may indicate in advance how many steps this state will be reached after. Thus, new truck loads can be requested in good time, thereby avoiding downtime.

Figure 5 shows a schematic diagram of an example of a simulation 500 of material distribution by a simulated earthmoving machine 505. By way of example only, this is a voxel-based simulation of a crawler dozer. In the simulation 500 shown in Figure 5, by way of example, the distribution of material is depicted using simulated material flows and selected trajectories.

6A, 6B, and 6C each show a schematic diagram of a geometric model 227 according to an embodiment. Here, FIG. 6A shows an exemplary spatial arrangement of the deposition, where the flow behavior of the deposition can be determined based on a voxel model. FIG. 6B shows a surface diagram of the geometric model 227, and FIG. 6C shows a mesh diagram.

The geometric model 227 allows for example for simulating sediments, including internal friction, adhesion and runoff. Here, a voxel (short for volumetric element) is the smallest building block that is characterized via its physical relationship to its neighbours (pressure, adhesion, gravity), its physical properties (colour, mass, reflectance, transmittance) and shape. Thus, three-dimensional structures of any shape can be represented, as can pixels (short for picture elements).

Exemplary techniques exist for adaptively designing the voxel network structure, i.e. increasing the resolution in particularly relevant regions and decreasing the resolution in other regions. These can be controlled via the differences between the interactions in the individual voxels. For example, if all nine voxels of a cube experience equal forces, they can be merged together. If the forces of voxels bordering each other are very different, they can be further subdivided to allow a more accurate tracking of the force vectors and material flow in the voxel network. Thus, additional computational costs are saved.

If an example includes an "and/or" combination between a first feature and a second feature, this should be interpreted as meaning that the example, according to one embodiment, has both the first feature and the second feature, and according to another embodiment, has only the first feature or only the second feature.

Claims (13)

A method (300) for operation of an earthmoving machine (100), comprising:
The method (300) comprises the following steps (305, 310, 315, 320):
reading (305) a terrain signal (215), a material signal (220) and a machine signal (225), wherein the terrain signal (215) represents a detected map of a current contour of the terrain (105), the material signal (220) represents at least one material parameter of a material detected in the terrain (105), and the machine signal (225) represents at least one machine parameter of the earthmoving machine (100);
creating (310) a geometric model (227) of the terrain (105) using the terrain signal (215), the geometric model (227) being created as a voxel network;
simulating (315) a material flow of the material using the geometric model (227) and the material signal (220);
outputting (320) a control signal (210) for controlling the earthmoving machine (100) based on the simulated material flow and using the machine signal (225) to redistribute the material;
The method (300). The method (300) comprises:
determining (405) the material parameters using the machine signal (225) and/or the topographical signal (215);
In the reading step (305), the material signal (220) represents the determined material parameter.
2. The method of claim 1 (300). In the determining step (405), the at least one material parameter is estimated using parameters stored in a database (235).
3. The method of claim 2 (300). The method (300) comprises:
identifying (410) a selected trajectory (115) for movement of the earthmoving machine (100);
In the outputting step (320), the selected trajectory (115) is used to output the control signal (210).
The method (300) of any one of claims 1 to 3. The method (300) comprises:
a step (415) of simulating dispensing said material using said simulated material flow and said selected trajectory (115) to obtain a simulated dispensing result,
In the outputting step (320), the simulated distribution result is used to output the control signal (210).
5. The method of claim 4 (300). The method (300) comprises:
further comprising the step of reading (420) optimization parameters;
the optimization parameters represent targets for the distribution of the material;
In an evaluating step (425), the simulated dispensing results are evaluated using the optimized parameters.
6. The method (300) of claim 5. if in the evaluating step (425) it is evaluated that the simulated dispensing of the material does not correspond to the optimized parameters, iteratively performing the identifying step (410), the simulated dispensing step (415), the further reading step (420) and the evaluating step (425);
If, in the evaluating step (425), it is evaluated that the simulated distribution of the material corresponds to the optimized parameters, in the outputting step (320), outputting the control signal (210).
7. The method (300) of claim 6. repeating the reading step (305) and the creating step (310) after the outputting step (320) to obtain an updated geometric model (227);
In a comparing step (455), the updated geometric model (227) is compared with the simulated distribution result;
repeating said determining step (405) if a predetermined iteration criterion is met;
The method (300) of any one of claims 1 to 7. The method (300) comprises:
- again determining (430) another trajectory for a further movement of said earth moving machine (100);
a step (435) of dispensing the material again using the simulated material flow and the selected alternative trajectory, thereby obtaining an alternative simulated dispensing result;
re-reading (440) the optimization parameters and/or further optimization parameters;
Including,
the different optimization parameters represent different targets for processing the material;
and evaluating the other simulated distribution results using the other optimized parameters in a re-evaluation step (445).
9. The method (300) of claim 7 or 8. An apparatus (200) configured to perform and/or drive and control steps (305, 310, 315, 320) of the method (300) according to any one of claims 1 to 9 in a corresponding unit. An earthmoving machine (100) equipped with the device (200) according to claim 10. A computer program configured to implement and/or drive and control steps (305, 310, 315, 320) of the method (300) according to any one of claims 1 to 9. A machine-readable storage medium on which the computer program of claim 12 is stored.

Images