DE102016206631A1 - Device for data fusion of measured data for generating a spatial occupancy grid for a motor vehicle - Google Patents

Abstract

The present invention relates to a device for data fusion of measured data for generating a spatial occupancy grid for a motor vehicle, the device comprising: a sensor device (10) which has at least one sensor and which is designed to provide sensor data; a data fusion device (20) configured to generate an environment update map (MU) based on a fusion of the provided sensor data; and a computing device (30) configured to update a spatial environment map (MG) based on the generated environment update map (MU) and with the aid of an evidence-based sensor data evaluation.

Description

Technical area

The present invention relates to devices and methods for data fusion of measurement data for generating a spatial occupancy grid for a motor vehicle.

Technical background

Environment modeling for driver assistance systems often use an occupancy grid to map the spatial vehicle environment.

A frequently used approach for the acquisition of newly acquired data is the so-called Dempster-Shafer theory, abbreviated DST, which allows existing environment modeling to be adapted by the newly acquired data.

Once a majority of sensors are involved in this update, the problem is usually to provide a powerful overall architecture.

Summary of the invention

It is an object of the present invention to provide an improved sensor data fusion for spatial environment modeling for a motor vehicle. This object is solved by the subject matters of the independent claims.

Embodiments and further developments can be taken from the dependent claims, the description and the figures of the drawings.

A first aspect of the present invention relates to a device for data fusion of measurement data for generating a spatial occupancy grid for a motor vehicle, the device comprising: a sensor device which has at least one sensor and which is designed to provide sensor data; a data fusion device configured to generate an environment update map based on a fusion of the provided sensor data; and a computing device configured to update a spatial environment map based on the generated environment update map and using evidence-based sensor data evaluation.

The present invention advantageously makes it possible to reduce the complexity of the multi-sensor data fusion for generating the spatial occupancy grid.

In particular, the present invention enables the resource consumption to be reduced in the data fusion and the environment modeling, whereby, for example, the computation time and / or the memory requirement can be reduced.

A further, second aspect of the present invention comprises a method for data fusion of measurement data for generating a spatial occupancy grid for a motor vehicle, the method comprising the following method steps: providing sensor data by means of a sensor device; Generating an environment update map based on a fusion of the provided sensor data using a data fusion facility and updating a spatial environment map based on the generated environment update map using evidence-based sensor data evaluation and computing.

Advantageous embodiments of the present invention are characterized in the subclaims.

In an advantageous embodiment of the present invention, it is provided that the data fusion device is designed to generate the environment update map by means of the fusion of the provided sensor data by an accumulation of occupancy probabilities determined by means of the sensor data into a total occupancy probability.

This advantageously makes it possible, by the cumulation or a simple summation, to provide a fusion of a multiplicity of sensor data, which is generated by a plurality of sensors in each case.

In a further advantageous embodiment of the present invention, it is provided that the data fusion device is designed to generate the environment update map by means of the fusion of the provided sensor data by an accumulation of non-occupancy probabilities determined by the sensor data into a total non-occupancy probability.

In a further advantageous embodiment of the present invention, it is provided that the data fusion device is designed to use confidence intervals which indicate a precision of the occupancy estimation in the cumulation of non-occupancy probabilities determined by means of the sensor data into the total non-occupancy probability.

Indications of inaccuracies in the form of confidence intervals advantageously allow subsequent calculations to be simplified as compared to an exact - i. no confidence intervals permitting - Dempster-Shafer theory implementation, short DST implementation.

In a further advantageous embodiment of the present invention, it is provided that the computer device is designed to update the spatial environment map based on the total occupancy probability and the total non-occupancy probability in the form of a thesis and a counter-thesis.

This advantageously makes it possible to minimize the computational effort and the required memory resources when calculating the spatial environment map, since only simplified theses and counter-theses need to be checked by the simplified data interpretation model.

In a further advantageous embodiment of the present invention, it is provided that the data fusion device is designed to carry out, as the accumulation, a successive accumulation of the occupancy probabilities and / or the non-occupancy probabilities.

This advantageously enables a simplified calculation of the total occupancy probability or the total non-occupancy probability, since only simple operations in the form of adding-in need to be performed.

In a further advantageous embodiment of the present invention, it is provided that the data fusion device is designed to determine the occupancy probabilities or the non-occupancy probabilities as a function of a sensor type of the at least one sensor of the sensor device.

This advantageously makes it possible, for example, to carry out a different weighting of different sensor types depending on the place of use and the time of use.

In a further advantageous embodiment of the present invention, it is provided that the data fusion device is designed to scale the occupancy probabilities or the non-occupancy probabilities prior to the cumulation.

This advantageously makes it possible to enable normalization of the values and improved sensor data fusion of the sensor data with the spatial environment map.

In an advantageous embodiment of the present invention, it is provided that the computer device is configured to update the spatial environment map based on the generated environment update map and using the evidence-based sensor data evaluation, wherein the evidence-based sensor data evaluation has a linear complexity.

This advantageously allows an accelerated calculation of the sensor data fusion and the combination of sensor data, which leads to an appreciation of the information that would not be possible without the data merge.

In an advantageous embodiment of the present invention it is provided that the device is designed to be coupled with a driver assistance system or a roadside recognition system or a free space recognition system or an automatic parking system.

This advantageously makes it possible to provide an accurate vehicle environment detection centrally for automotive networked safety systems.

The described embodiments and developments can be combined with each other as desired.

Further possible refinements, developments and implementations of the present invention also include combinations of previously or subsequently described embodiments of the present invention which are not explicitly mentioned.

The accompanying drawings are intended to provide further understanding of the embodiments of the present invention. The accompanying drawings illustrate embodiments and, together with the description, serve to explain concepts of the present invention.

Other embodiments and many of the stated advantages will become apparent with reference to the figures of the drawings. The illustrated elements of the figures of the drawings are not necessarily shown to scale to each other.

Brief description of the figures

Show it:

1 a schematic representation of a sensor data fusion system for explaining the present invention;

2 a schematic representation of a system for a sensor data fusion for explaining the present invention;

3 a schematic representation of a sensor data fusion system for explaining the present invention;

4 : a schematic representation of an apparatus for data fusion of measurement data for generating a spatial occupancy grid according to an embodiment of the present invention; and

5 : a schematic representation of a flowchart of a method for data fusion of measurement data for generating a spatial occupancy grid according to another embodiment of the present invention.

Detailed description of the embodiments

In the figures of the drawings, like reference characters designate like or functionally identical elements, components, components or method steps, unless indicated otherwise.

The vehicle is, for example, a motor vehicle or hybrid vehicle, for example a hybrid vehicle with a sailing function, for example a motorcycle, a bus or a truck or a bicycle.

Driver assistance systems are electronic auxiliary devices in motor vehicles to assist the driver in certain driving situations. For example, the device for data fusion of measurement data for generating a spatial occupancy grid is coupled to a driver assistance system, for example with a parking aid, a lane departure warning system, a distance control, or a brake assist.

The 1 shows a schematic representation of a sensor data fusion system for explaining the present invention.

In the 1 sensors are shown. The 1 shows for example a number of n sensors, where the n sensors are to be combined by a multi-sensor data fusion.

An update map, also known as "Update Map" in English, is used to first summarize the sensor data of each individual sensor SE1, SE2, ..., SEn before a Dempster-Shafer fusion, also called DST-Fusion, with the already existing environment map MG is performed.

In this case, a newly calculated update card MU is used for each cycle, so that the environment map is also recalculated when going through each cycle. Thus, an updating card MU and a spatial environment card MG can be assigned to a single cycle.

In this case, each individual sensor SE1, SE2,..., SEn can change the updating card, and this ensures its contribution MS1, MS2,..., MSn of the respectively assigned sensor data to the sensor data fusion.

The final for each calculation cycle, short cycle, newly created update card is then updated by a data fusion with the existing spatial environment map, so that at the end of the cycle a newly calculated spatial environment map is provided and at the beginning of the cycle an initial existing spatial environment map available is.

For example, the newly calculated spatial environment map of the previous cycle can be used as the initially used spatial environment map of each cycle.

The fusion of the spatial environment map with the generated environment update map can be done through a DST merger. Inconsistencies regarding the occupancy grid, which may already exist from previous cycles, are processed accordingly.

In the 2 is a modified calculation concept with a changed data and computer architecture in comparison to the 1 shown.

Instead of a pure accumulation of the sensor data of each sensor SE1, SE2, ..., SEn, a DST fusion function block is introduced, this function block has the task of merging sensor data of a respective sensor with a respective region of the update card. The advantage of this concept lies in the ability to handle conflicting data from different sensors more efficiently.

Each DST step involves a fusion of at least two data values, such as an update card with sensor or an update card with a spatial environment map.

The number of required DST function blocks corresponds to the number of sensors n + 1, that is to say n + 1.

For each nth sensor, the following DST equations can be set up and solved, for example, by using a Dempster-Shafer occupancy grid function or by using certain data values for empty and occupied: Equation 1

An equivalent formula may also be used for the merge between the update map and the spatial environment map.

This recursive merger, as in 2 has the advantage that less memory must be used and the same sensor memory can be used multiple times for the different sensors, however, the successive calculation of the update of the spatial environment map by a plurality of steps is computationally intensive. Thus, a low demand for storage space resurfaces a higher demand for computer power.

This illustrates that the DST calculation does not have a linear complexity. Any probability of free or occupied space must also be calculated using the data sets of all other sensors and adds to the complexity of the calculation.

In the 3 The architecture of the computer and the concept of the data function used a comprehensive DST fusion step, which simultaneously handles the fusion of the sensor data from all recorded sensor data. The update card will then be replaced by the already known DST operation 2 merged with the spatial environment map.

However, this approach only allows a less effective treatment of inconsistencies between the individual sensor data from different sensors.

The 4 shows a schematic representation of an apparatus for data fusion of measurement data for generating a spatial occupancy grid.

The device 100 includes a sensor device 10 , a data fusion facility 20 and a computing device 30 ,

The sensor device 10 has at least one sensor SE1, SE2, ..., SEn and is configured to provide sensor data. For example, as in the 1 shown, the sensor device 10 a plurality n of sensors.

The data fusion device 20 For example, it is configured to generate an environment update map MU based on a fusion of the provided sensor data.

The computer device 30 For example, it is designed to update a spatial environment map MG based on the generated environment update map MU and using an evidence-based sensor data evaluation.

Evidence-based sensor data assessment may involve, as the fusion of the data, splitting the sensor data fusion into thesis and counter-thesis.

This advantageously enables a simplified calculation and interpretation of the sensor data. Furthermore, this allows the DST equations to be advantageously converted for calculation or fusion.

wobei „H“ für eine Hypothese steht, e⁺(H) steht für eine Akkumulation von die Beweise unterstützenden Hypothesen, wobei e^–(H) für eine Akkumulation von die Beweise nicht unterstützenden Hypothesen, sprich Gegenthesen, darstellt, wobei pro(H) die Wahrscheinlichkeit für die Hypothese darstellt, wobei con(H) die Wahrscheinlichkeit gegen die Hypothese darstellt. The evidence-based sensor data evaluation can be done as follows: Equation 2:

where "H" stands for a hypothesis, e ⁺ (H) stands for an accumulation of evidence supporting hypotheses, where e ^- (H) represents an accumulation of hypotheses, that is counter-theses, which do not support the evidence, where pro (H) represents the probability of the hypothesis, where con (H) represents the probability against the hypothesis.

This can also be represented in individual steps as follows. The implementation of a multi-sensor data function or a fusion can be carried out as follows:

In a first step, an accumulation of the data values for free and occupied spatial subsections in the update map, which is calculated for all sensors.

Equation 3:

• m _updfree (k) = m _sensorFree1 (k) + m _sensorFree2 (k) ... ... m _sensorFreeN (k) m _updocc (k) = m _sensorcc1 (k) + m _sensorcc2 (k) ... .. m _sensorccN (k)

This pure adding operation in the form of adding to an already existing value in the occupancy grid of the environment map or the update map enables a simple calculation.

It is not necessary to read in the respective occupancy grid value - the probability value for "occupied" - first, where k is used to address the grid and m_sensorfree1 (k) is the probability value or data value for the kth grid element that it is free and this was detected with the first sensor.

Correspondingly, m_sensorocc1 is the probability value for the kth grid element that it is not free and this was detected with the first sensor.

As a second calculation step, for example, a scaling of the occupancy probabilities takes place during the fusion.

Gleichung 5:

This scaling is needed to allow fusion with the spatial map MG: Equation 4:

Equation 5:

In this case, the occupancy probability or non-occupancy probability can represent a simple value, wherein this value can be changed for each type of sensor used.

For example, as a third step, the values for the non-occupancy probabilities are scaled for the total number of n sensors. Equation 6:

In this case, an accuracy error in the environment update map MU, that is to say in the "update map", can advantageously be permitted, for example in the form of permitted confidence intervals or confidence intervals to be achieved, which specify the precision of the location estimate of the occupancy probabilities or the non-occupancy probabilities.

In Equation 6, one introduces the desired inaccuracy in the form of confidence intervals or confidence intervals, thereby simplifying the following equations accordingly, compared to an exact DST implementation.

This loss of accuracy of the non-occupancy probabilities has no further relevance to intended applications, for example in a driver assistance system, since the loss of accuracy by the final or subsequent DST step, i. the combination of the update map with the grid map, is compensated.

In other words, there is no influence of the introduced loss of accuracy for the step S3 of updating the surrounding area map MG based on the generated environment update map MU by means of an evidence-based sensor data evaluation and by means of a computing device 30 ,

As a fourth method step, for example, combining the probability values for the support and for not supporting the hypothesis.

Equation 7:

• accufree (k) = 1 - (1 - m _updfree (k)) ^N accuocc (k) = 1 - (1 - m _{occ_contri1} (k)) (1 - m _{occ_contri2} (k)) ... (1 - m _{occ_contriN} (k))

As a fifth step, for example, the occupancy probabilities and the non-occupancy probabilities for the environment update map MU are calculated.

Equation 8:

• m _updfree (k) = accufree (k) (1 - accuocc (k)) / 1 - accufree (k) accuocc (k) m _updocc (k) = accuocc (k) (1 - accufree (k)) / 1 - accufree (k) accuocc (k)

This calculation makes it possible to increase the calculation with a linear complexity in relation to the number of sensors.

Also, the memory requirement is reduced for all process steps.

For example, the calculation for a spatial environment map with a 1024 having 1024 occupancy grid with a memory requirement of less than 1 megabyte of volatile memory perform.

Furthermore, the method advantageously allows inconsistencies between the individual sensors within a respective cycle to be calculated and eliminated.

The 5 shows a schematic representation of a flowchart of a method for data fusion of measurement data for generating a spatial occupancy grid, the method comprising the following method steps.

As a first step of the method, sensor data S1 is provided by means of a sensor device 10 ,

As a second method step, generation S2 of an environment update map MU is based on a fusion of the provided sensor data by means of a data fusion device 20 ,

As a third step of the method, updating S3 of a spatial environment map MG takes place based on the generated environment update map MU, by means of an evidence-based sensor data evaluation and by means of a computer device 30 ,

Although the present invention has been described above with reference to preferred embodiments, it is not limited thereto but is modifiable in a variety of ways. In particular, the present invention can be varied or modified in many ways without departing from the gist of the invention.

In addition, it should be noted that "comprising" and "having" does not exclude other elements or steps, and "a" or "an" does not exclude a plurality.

It should also be appreciated that features or steps described with reference to any of the above embodiments may also be used in combination with other features or steps of other embodiments described above. Reference signs in the claims are not to be considered as limiting.

Claims (10)

Contraption ( 100 ) for data fusion of measurement data for generating a spatial occupancy grid for a motor vehicle, the apparatus comprising: 10 ), which has at least one sensor (SE1, SE2, ..., SEn) and which is designed to provide sensor data; A data fusion device ( 20 ) configured to generate an environment update map (MU) based on a fusion of the provided sensor data; and - a computer device ( 30 ), which is adapted to update a spatial environment map (MG) based on the generated environment update map (MU) and with the aid of an evidence-based sensor data evaluation. Contraption ( 100 ) according to claim 1, wherein the data fusion device ( 20 ) is designed to generate the environment update map (MU) by means of the fusion of the provided sensor data by an accumulation of occupancy probabilities determined by means of the sensor data into a total occupancy probability. Contraption ( 100 ) according to claim 1 or 2, wherein the data fusion device ( 20 ) is adapted to generate the environment update map (MU) by means of the fusion of the provided sensor data by an accumulation of non-occupancy probabilities determined by the sensor data into a total non-occupancy probability. Contraption ( 100 ) according to claim 3, wherein the data fusion device ( 20 ) is adapted to use in the accumulation of non-occupancy probabilities determined by the sensor data in the total non-occupancy probability confidence intervals, by which a precision of the non-occupancy estimation is specified. Contraption ( 100 ) according to one of the preceding claims 2 to 4, wherein the computer device ( 30 ) is adapted to update the spatial environment map (MG) based on the total occupancy probability and the total non-occupancy probability in the form of a thesis and a Gegenenthese; and / or wherein the data fusion device ( 20 ) is designed to perform as the accumulation a successive accumulation of occupancy probabilities or a successive accumulation of the non-occupancy probabilities. Contraption ( 100 ) according to one of the preceding claims, wherein the data fusion device ( 20 ) is adapted to determine the occupancy probabilities or the non-occupancy probabilities as a function of a sensor type of the at least one sensor of the sensor device ( 10 ) to investigate. Contraption ( 100 ) according to one of the preceding claims, wherein the data fusion device ( 20 ) is adapted to scale the occupancy probabilities or non-occupancy probabilities prior to cumulation. Contraption ( 100 ) according to one of the preceding claims, wherein the computer device ( 30 ) is adapted to update the spatial environment map (MG) based on the generated environment update map (MU) and using the evidence-based sensor data score, wherein the evidence-based sensor data score has a linear complexity. Contraption ( 100 ) according to any one of the preceding claims, wherein the device ( 100 ) is adapted to be coupled with a driver assistance system or a roadside recognition system or a free space recognition system or an automatic parking system. Method for data fusion of measured data for generating a spatial occupancy grid for a motor vehicle, the method comprising the following method steps: - Providing (S1) sensor data by means of a sensor device ( 10 ); Generating (S2) an environment update map (MU) based on a fusion of the provided sensor data by means of a data fusion device ( 20 ); and - updating (S3) a spatial environment map (MG) based on the generated environment update map (MU), with the aid of an evidence-based sensor data evaluation and by means of a computer device ( 30 ).

Images