Map verification method

Abstract

receiving sensor data of the at least semi-autonomous robot (F) for depicting at least one detection element (S3, S4, S5) representing an environmental element of the at least semi-autonomous robot (F) as detected by an environmental sensor of the at least semi-autonomous robot (F); receiving map data (Dk) depicting a map having at least one map element (A3, A4, A5) representing an environmental element of the at least semi-autonomous robot (F) as plotted on a predetermined map; and a step of receiving location data (D1) indicating a location of a map element (F); a step of determining data uncertainty, including sensor data uncertainty, map data uncertainty, and / or location data uncertainty; a step of initializing a presence probability (P) for at least one map element (A3, A4, A5) with an initial value; and a step of updating the presence probability (P) of the at least one map element (A3, A4, A5) using the map data (Dk), the sensor data (Ds), the location data (D1), and the data uncertainty.

Description

Technical Field

[0001] The present invention relates to a map verification method.

Background Art

[0002] Currently, at least semi-autonomous robots, especially partially automated vehicles, rely heavily on high-resolution maps, so-called HD maps, in order to plan their future behavior. Relying only on sensor data from at least semi-autonomous robots can lead to undesirable errors in the assessment of traffic situations.

Summary of the Invention

Problems to be Solved by the Invention

[0003] However, if the map data is old, the future behavior of at least semi-autonomous robots cannot be planned without errors either.

[0004] Therefore, a method for verifying map data is desired.

Means for Solving the Problems

[0005] Embodiments of the present invention provide a map verification method according to the independent claims. Useful embodiments of the present invention will become apparent from the dependent claims, the specification, and the attached drawings.

[0006] According to one aspect of the present invention, a map verification method, i.e., a card-type authenticity check method, includes the following steps. Receiving sensor data of at least a semi-autonomous robot indicating at least one detected element, where the at least one detected element represents an environmental element of the at least semi-autonomous robot as detected by an environmental sensor of the at least semi-autonomous robot. Receiving map data depicting a map having at least one map element, where the at least one map element represents an environmental element of the at least semi-autonomous robot as plotted on a predetermined map. Determining location-specific data from the received sensor data, where the location-specific data indicates the location of the at least semi-autonomous robot on the map. Determining data uncertainty, where the data uncertainty includes sensor data uncertainty, map data uncertainty, and / or location-specific data uncertainty. Initializing the probability of existence of at least one map element with an initial value. Updating the probability of existence of at least one map element using the map data, sensor data, location-specific data, and data uncertainty.

[0007] Preferably, the sensor data of the at least semi-autonomous robot includes online sensor data detected by a sensor of the at least semi-autonomous robot. Preferably, the sensor data includes camera data, lidar data, radar data, and / or GPS data.

[0008] As used herein, the term "map element" specifically includes traffic signs such as traffic signs or traffic lights, as well as road signs.

[0009] In general, at least a semi-autonomous robot can be an at least partially automated vehicle. Alternatively, the at least semi-autonomous robot may be another mobile robot such as one that moves by flying, swimming, diving, or walking. For example, the mobile robot may be at least a semi-autonomous lawn mower or at least a semi-autonomous cleaning robot, etc.

[0010] Preferably, initializing the occupancy probability includes determining an initial value of the map element occupancy probability, where the initial value indicates a provisional occupancy probability assigned to the map element when there is no sensor data.

[0011] Preferably, the localization data includes GPS data.

[0012] Preferably, the localization data is determined using the received sensor data and the received map data. For example, the localization data is determined by matching the map data with the sensor data, particularly using GPS data. Thus, the localization data indicates at least an estimated position of the at least semi-autonomous robot on the map.

[0013] Thus, the map verification method uses the so-called Markov assumption, according to which the actual state depends only on the previous state and the current sensor data, but not on the complete history of the states. In this way, the simplest computational load of the map verification method can be realized.

[0014] Preferably, the step of determining the data uncertainty includes the step of representing the sensor data, the map data, and the localization data in a probabilistic representation.

[0015] Preferably, the probabilistic representation includes a Bayesian representation.

[0016] The probabilistic representation of the sensor data takes into account the spatial uncertainty of the sensor measurements, the interference rate of the sensor measurements, the detection probability of the map elements by the sensor, and / or the relevance and non-relevance depending on the form used.

[0017] Perturbations, also referred to as clutter, include sensor measurements, i.e., sensor data that are not determined by real objects, i.e., map elements, such as ghost measurements.

[0018] The clutter rate represents the number of clutter measurements per time step, which is typically modeled using a Poisson distribution when the clutter measurements are made independently.

[0019] Preferably, the map verification method is executed online, i.e., in at least a semi-autonomous robot. The map verification method uses not only map data and sensor data, but also the localization of at least a semi-autonomous robot, i.e., its estimated position relative to the map. The task is to check or remove map elements and thus to detect possible discrepancies between the map data and the sensor data, i.e., the environment of at least a semi-autonomous robot.

[0020] A prerequisite for the proper functioning of the map verification method is not only accurate localization but also accurate sensor calibration. The reason is that the detected divergence between the map data and the sensor data can be caused not only by inaccurate or old maps, but also by insufficient localization and inaccurately calibrated or incomplete sensors.

[0021] Preferably, a common probability of existence of a map element is calculated from different sensors. Alternatively, separate probabilities of existence of a map element are calculated for each sensor or combination of different sensors.

[0022] Preferably, the sensor data includes data from untracked sensor measurements, and the error from the sensor measurements is different from the tracked elements that are filtered over time and has little or no correlation over time. Measurements with temporally uncorrelated measurement errors are much easier to probabilistically model, and this model also makes it easier to perform checks that are part of the internal validity check. The measurements are not lost, and the map verification method can access all available information, including perturbation measurements that would otherwise be filtered by the tracking algorithm.

[0023] Updating the probability of existence corresponds to the concept of element fusion or object fusion. In other words, the probability of existence is calculated for map elements that are directly observable by fusing map data and sensor data. This can be done individually for each sensor, but it is also possible to calculate the simultaneous probability of existence based on all available sensor data or any combination of sensors at once.

[0024] In addition to the probability of existence, it is preferable to calculate the updated position of the map element. This is essentially done by a (multi-)Bernoulli filter in the random finite set approach (abbreviated as the RFS approach), or if necessary, it can be added as an additional Kalman filter via the logit binary Bayesian filter approach (abbreviated as the logit approach).

[0025] The calculation of the updated probability of existence of the map elements within the horizontal line is performed for the visible map elements at each time step using the probabilistic representation of the input data.

[0026] In this way, an improved behavior regarding safety and trajectory planning is provided.

[0027] In this way, an overall, accurate, and realistic representation of the environment of at least a semi-autonomous robot is provided based on the fused sensor data and map data.

[0028] According to a preferred embodiment, the map verification method includes the following steps. Projecting at least one map element into the sensor space of an environmental sensor.

[0029] All map elements that are part of the map horizon are projected into all relevant, for example, forward sensors, i.e., their sensor spaces. This projection is usually easy to achieve, but the back-projection from sensor coordinates to map coordinates is generally more complex or even undefined, for example in the case of a monocular camera. The map horizon is a subset of the overall map and consists of map elements in the vicinity of at least a semi-autonomous robot. This map horizon includes at least the fields of view of all relevant sensors, but is typically larger.

[0030] According to a preferred embodiment, the map verification method comprises the following step, i.e., the step of assigning at least one map element to at least one detected element.

[0031] In the assignment step, a predetermined number of best global assignments of sensor data, i.e., sensor measurements, to represent elements in the sensor space are calculated using the MHT ranked assignment algorithm and, for example, based on the Hungarian method. The ranked assignment can be based, for example, on a multi-object measurement model from RFS theory, which captures not only the spatial uncertainty of sensor measurements but also the perturbation velocity, perturbation intensity, and detection probability. In particular, the (multi-)Bernoulli filter can process multiple associations weighted based on the respective probabilities already calculated in the association step. In particular, these associations also take into account the possibility of missed detections (no map element is assigned to any detection) and spurious detections (no map element is assigned to any detection).

[0032] Preferably, the association of the detected element with the map element is synchronized only once within the system, between all modules that depend on this information. Otherwise, the map verification method is decoupled from other functional modules that can perform completely different associations and uses them without checking the accuracy of these potentially inaccurate associations. The association of the measurements for depicting the map object requires localization as input. The association depends on the predicted localization estimate from the previous time step, and the provided association is used to update the next step of the localization module.

[0033] According to a preferred embodiment, the map verification method comprises the following steps. An evaluation step of evaluating the probability of existence of at least one map element, where evaluating comprises one of confirming the map element, demonstrating that the map element, potentially a new map element, is an error, and making no statement whatsoever, the evaluation step.

[0034] If the sensor data cannot be assigned to an existing map element, a potentially new map element is identified. In this way, a potentially new map element can be identified. In other words, the probability that the measurement, i.e., the corresponding sensor data, cannot be assigned to an existing map element is determined. As a result, a potentially new map element is reported or a new map element is directly initialized.

[0035] According to a preferred embodiment, the step of updating the probability of existence includes a random finite set (RFS) approach or a logit approach.

[0036] The RFS approach refers to a sensor model based on random finite set theory, spatial uncertainty, perturbation velocity, perturbation intensity, and detection probability.

[0037] The logit approach refers to a sensor model based on hit rate and false alarm rate using the logarithmic implementation of a binary Bayesian filter.

[0038] The RFS approach can further provide an updated position of a map element in addition to the existence probability by applying a (multi) Bernoulli filter, where the logit approach has to be combined with a Kalman filter to also provide a position update.

[0039] Using all of this information, it is possible to update the existence probability of a conventional map element using either the RFS or the logit approach. The existence probability identifies whether a map element still exists based on the collected but uncertain sensor data. Map elements that are completely invisible are not updated and thus retain their existence probability from the previous time step. The existence probability of a map element, preferably the update of the position, is performed at each time step at which a set of sensor measurements is available. Both the RFS and the logit approaches follow the Markov assumption, and the actual state of an updated map element depends only on the current measurement and its previous state, rather than on the complete history of the state of the map element.

[0040] The update of the existence probability is performed by either a (multi) Bernoulli filter and an RFS-based measurement model, which further update the map object position using sensor data as a by-product or using a binary Bayesian filter with a log-likelihood (logit) and the corresponding hit / miss rate.

[0041] The multi-Bernoulli filter is based on RFS theory and takes into account the spatial uncertainty of sensor measurements, the perturbation speed / intensity of the sensor, and the detection probability of map elements.

[0042] The binary Bayesian filter is also simply called the logit or log-odds approach, models the hit and miss of elements with specific probabilities, and measurements near a map element are considered hits, and measurements in the vicinity, if visible, are not considered misses of that specific map element.

[0043] In the case of point-like objects or objects sparsely extended with respect to the sensor space, the approach described can be applied directly, for example, to traffic lights, traffic signs, dashed lane markings, poles, tree trunks, points of bots, etc. Solid lane markings, for example, boundaries may need to be divided into smaller segments if they are too long and thus significantly exceed the sensor's field of view. These smaller segments can then be checked for their presence using the proposed approach. Thus, a probability of existence is assigned to each segment of the (continuous) fixed lane boundary.

[0044] As a by-product of using the RFS approach that depends on the (multi) Bernoulli filter, it is possible to calculate the probability that a measurement is not part of an object that already exists. This information is potentially a complete indicator of new map elements. Thus, additional measurements with a high probability of not belonging to any known map element are transferred to enable the inference of new map elements. These measurements can be used to create new map elements and also to update them in the following steps of the map verification method.

[0045] According to a preferred embodiment, the probability of existence is initialized with an initial value of 50%.

[0046] Preferably, the initial value is predetermined. More preferably, the initial value is determined dynamically for each map element. For example, to determine the initial value, a predetermined trend of the probability of existence is determined according to the characteristics of the map element and / or the map data. For example, a map element of the type of road sign is initialized with a lower initial probability than a map element of the type of traffic light, assuming that road signs are usually replaced more frequently than traffic lights.

[0047] According to a preferred embodiment, updating the probability of existence of at least one map element is repeated at time intervals.

[0048] Preferably, the time interval is predetermined. Further preferably, the time interval depends on the detection rate of at least one environmental sensor of the at least semi-autonomous robot. In other words, the prior existence probability of at least one map element is updated as soon as new sensor data becomes available. For example, the environmental sensor of the at least semi-autonomous robot comprises a camera that provides 25 frames per second. As a result, the prior existence probability of at least one map element is updated 25 times per second.

[0049] According to a preferred embodiment, the map verification method comprises the following steps. Determining the visibility of a map element, which is determined using the field of view of the environmental sensor and the occlusion of the map element. Determining a detection probability using the visibility of the map element.

[0050] The visibility of a map element is determined not only by checking the field of view of the sensor, but also by checking the occlusion of each map element.

[0051] Preferably, the detection probability is determined using the true positive rate of the sensor.

[0052] In addition, the visibility takes into account the sensor measurements and the uncertainty of the map element, as well as the uncertainty of the position determination and calibration information via error propagation. The visibility information can then be integrated into the detection probability. The detection probability captures not only the visibility of the map object, but generally also the probability that the sensor or detection algorithm will produce a corresponding measurement. Thus, the detection probability of a non-visible object is zero, but the detection probability of a fully visible object is generally not 1. This is because the sensor typically cannot detect all visible objects and can only detect a certain percentage of them (see the percentage of true positive results).

[0053] The visibility of map elements is estimated separately for each sensor, in particular by checking the line of sight from the sensor origin to each map element, and thus checking the sensor's perceptible field of view and any occlusion by other objects in front of the map element. This can be done, for example, by using a Stixel representation for a stereo camera system that provides a 3D representation of the environment. This representation can then be used to check whether such Stixels are located in front of the map element and thus obscure it. An alternative is to use a labeled pixel image from a monocular camera, which can be used specifically for all map elements on the ground. The basic idea is to check whether the pixels in front of the expected map element are classified as objects, such as cars, pedestrians, etc., and thus obscure the map element on the ground behind it, such as a lane marker.

[0054] The detection probability of a map element is then derived not only from its estimated visibility but also from the sensor's true positive rate for that particular map element type. The detection probability specifies the likelihood that a map element generates a corresponding sensor reading. This is closely related to the hit rate in the logit representation and is directly integrated into the measurement model of the RFS approach. The detection probability is zero for map elements that are not fully visible and less than or equal to 1 for objects that are fully visible. There is only one detection probability for a fully visible map element, and a sensor with a true positive rate of 1, i.e., a sensor that does not provide any false negative measurements in this case.

[0055] The true positive rate specifies the ratio of the detected map elements to the total number of map elements present in the environment, or in other words, the ratio of the true positive measurements to the sum of the true positive and false negative measurements.

[0056] The detection probability includes the probability that an existing map element generates a corresponding measurement, i.e., sensor data. This includes the visibility of the map element and the sensor's true positive rate in the environment.

[0057] According to a preferred embodiment, data uncertainty is used to determine the visibility of map elements.

[0058] According to a preferred embodiment, the probability of existence of at least one map element having a detection probability below a predetermined threshold is not updated.

[0059] According to a preferred embodiment, the map verification method comprises the following steps. That is, a step of verifying the validity of the probability of existence.

[0060] To verify the validity of the map validity verification method itself, the probabilistic assumptions used to calculate the probability of existence are checked in the context of internal credibility or validity checks in at least a semi-autonomous robot. This validity check reports statistically significant deviations of the hit and miss rates in the logit approach from the assumed parameters, for example, the spatial uncertainty, perturbation rate, and detection probability in an RFS-based approach, or the online estimated parameters at a predefined significance level. Furthermore, the error propagation from the map to the sensor space potentially requires the linearization of a non-linear function, which is also checked for strong non-linearity in the relevant area as a validity check.

[0061] The probabilistic and algorithmic assumptions used to update the occupancy probability are verified online as internal validity or consistency checks. Preferably, the measurement model of the environmental sensor is checked for consistency. The relevant parameters of the measurement model, e.g., the spatial uncertainty of the sensor measurements, the sensor perturbation rate, and the detection probability when using the RFS-based formulation, are estimated online. Subsequently, the validity check verifies that the assumptions made to initially calculate the occupancy probability are within the confidence intervals of these online-estimated parameters. For this purpose, a statistical hypothesis test is preferably performed. Preferably, the parameters are estimated online. However, they are not adjusted accordingly to avoid self-fulfilling prophecies. Thus, the estimated parameters are only checked against the assumed parameters to detect the probabilistic deviation between the two and to report to a further module that the credibility check itself is no longer credible. This is a form of self-evaluation of the credibility check itself.

[0062] According to a preferred embodiment, sensor data from different sensors of at least a semi-autonomous robot are compared to verify the validity of the occupancy probability.

[0063] According to a further aspect of the invention, a map verification system is configured to execute the map verification method specified herein.

[0064] According to a further aspect of the invention, a method for controlling at least a semi-autonomous robot comprises the following steps. Executing the map verification method specified herein to determine the occupancy probability of at least one map element. Determining a robot trajectory using sensor data, map data, localization data, and the occupancy probability of at least one map element. Controlling at least the semi-autonomous robot based on the determined robot trajectory.

[0065] In this way, the map verification method enables action and trajectory planning to safely use map data by checking relevant parts of the map using sensor data, especially the online sensor data of at least semi-autonomous robots, before relying on their presence in the environment of at least semi-autonomous robots. Based on the output of the map verification method, the action planning module knows which map elements are confirmed by the sensor data, which map elements are unknown, e.g., due to occlusion, and which map elements no longer exist. Thus, the map verification method provides an overall, accurate, and most important actual representation of the environment of at least semi-autonomous robots based on fused sensor and map data.

[0066] In the following examples, action plans are shown.

[0067] Example 1 The vehicle is approaching an intersection where no traffic signal can be seen. If the traffic signal still exists, it will not be hidden. The probability of existence slowly decreases from 50% to 0% as the vehicle approaches the unseen traffic signal. Action generation decelerates the vehicle quite early in response to the decreasing probability of existence and finally reaches a safe stop when the traffic signal is safety-critical on the map, avoiding entering the intersection while relying on old map data.

[0068] Example 2 The vehicle is approaching an intersection where the traffic signal is hidden. Action generation decelerates the vehicle and / or drives around the object hiding the traffic signal if possible. The idea in this regard is to collect more information about the presence of the traffic signal. If the traffic signal cannot be seen and is safety-critical, the vehicle stops in front of the intersection.

[0069] Example 3 The vehicle is approaching an intersection where it can clearly see the traffic signal. The probability of the existence of the traffic signal slowly increases from 50% to 100% as the vehicle approaches the intersection. Since the vehicle can perceive the traffic signal and its state, there is no need to decelerate to safely cross the intersection.

[0070] According to a preferred embodiment, a method for controlling at least a semi-autonomous robot comprises the following steps. A step of determining a control mode using sensor data, map data, location identification data, and the probability of the existence of at least one map element, and a step of controlling at least the semi-autonomous robot based on the determined control mode.

[0071] Preferably, the control mode includes predetermined behavior characteristics of at least the semi-autonomous robot. For example, the control mode includes a "normal drive" mode, a "pre-crash safety" mode, and / or a "safe stop" mode. Depending on the control mode, at least the semi-autonomous robot is given the same input data, namely sensor data, map data, and / or location identification data, and is controlled differently.

[0072] Preferably, the computer program includes instructions that cause the computer program to execute the map verification method specified herein when the computer program is executed by a computer.

[0073] Preferably, a machine-readable storage medium stores the computer program described herein.

[0074] Advantageously, map data including at least one map element is received from a remote server, and the result of the map verification method is advantageously sent to the remote server and used there as a basis for determining whether an update of at least one map element or a map resurvey at the location of at least one map element should be triggered. The result of the map verification method is, in particular, the updated probability of existence of at least one map element or information derived therefrom, indicating whether the existence of at least one map element is confirmed, or proven to be an error, or whether no statement can be made about its existence at all.

[0075] Further means for improving the present invention will be described in more detail below with reference to the drawings and the description of the preferred embodiments of the present invention.

Brief Description of the Drawings

[0076]

Figure 1a

Figure 1b

Figure 1c

Figure 2

Figure 3a

Figure 3b

Figure 3c

Figure 4

Mode for Carrying Out the Invention

[0077] Figure 1a shows the first traffic situation V1 at the first time step. At least a semi-autonomous robot, in this case an at least partially automated driving vehicle, is moving towards the intersection. Two map elements of the first traffic signal A1 and the second traffic signal A2 can be obtained from the map data of the high-precision (HD) map. However, in this case, the first traffic signal A1 has been removed due to construction work. In this regard, the map data regarding the first traffic signal A1 is outdated. According to the map verification method, for the first traffic signal A1 and the second traffic signal A2, the first existence probability P1 and the second existence probability P2 are initialized with a value of 50%. At this first time step, the vehicle F is still relatively far from the intersection. The sensor data provided by the environmental sensors of the vehicle F does not have detection elements for either the first traffic signal A1 or the second traffic signal A2. At the first time step, a credibility check is performed in which the first existence probability P1 and the second existence probability P2 are updated. In the credibility check, based on the internal statistics and positioning data of the vehicle F, it is determined that the vehicle F is still quite far from the intersection such that the environmental sensors of the vehicle F cannot detect either the first traffic signal A1 or the second traffic signal A2. In other words, the first traffic signal A1 and the second traffic signal A2 are not within the field of view of the environmental sensors of the vehicle F. Therefore, based on the sensor data, it is determined that no further statements can be made regarding the first existence probability P1 and the second existence probability P2. In this regard, the first existence probability P1 remains at 50% and the second existence probability P2 remains at 50%. Based on the sufficient distance of the vehicle F to the intersection for the control of the at least partially automated driving vehicle, it is determined to continue driving at normal speed. The control mode of the vehicle in this case is "normal drive".

[0078] Figure 1b shows the first traffic situation V1 at the second time step. The vehicle F, which is at least partially automated, is closer to the intersection than at the first time step. At the second time step, a plausibility check is performed in which the first presence probability P1 and the second presence probability P2 are updated. In the plausibility check, based on the internal statistics and positioning data of the vehicle F, it is determined that the vehicle F is already close enough to the intersection for the environmental sensors of the vehicle F to be able to detect both the first traffic light A1 and the second traffic light A2. In other words, the first traffic light A1 and the second traffic light A2 are within the field of view of the environmental sensors of the vehicle F. Thereby, it is determined that statements about the first presence probability P1 and the second presence probability P2 can be made based on the sensor data. Since the first traffic light A1 no longer exists as defined, the environmental sensors of the vehicle F also do not detect any elements at the position expected from the map data. In contrast, the second traffic light A2 is still at the position expected from the map data. In this regard, the first presence probability P1 is reduced from 50% to 35%, and the second presence probability P2 is increased from 50% to 70%. For the control of the at least partially automated vehicle, it is decided to slightly decelerate the vehicle F based on the first presence probability P1 and the second presence probability P2, so that there is more time to measure the environment in order to be able to make further statements, especially about the first traffic light A1. The control mode of the vehicle F is changed from "normal drive" to "pre-crash safety".

[0079] In this case, in the plausibility check, a difference in the first traffic light A1 between the sensor data and the map data was detected. If the first traffic light A1 was also detected by the environmental sensors of the vehicle F, the first presence probability P1 was increased from 50% to 70%, and the control mode of the vehicle F remained unchanged in "normal drive".

[0080] FIG. 1c shows a first traffic situation V1 in a third time step. The at least partially automated vehicle F is closer to the intersection than in the second time step. In the third time step, a credibility check is performed, in which the first presence probability P1 and the second presence probability P2 are updated. In the credibility check, based on the internal statistics and localization data of the vehicle F, it is determined that the vehicle F is still close enough to the intersection that the environmental sensor of the vehicle F should be able to detect both the first traffic light A1 and the second traffic light A2. In other words, the first traffic light A1 and the second traffic light A2 are within the visibility range of the environmental sensor of the vehicle F. It is therefore determined that, based on the sensor data, a statement can be made about the first presence probability P1 and the second presence probability P2. As specified, the first traffic light A1 is no longer present, so the environmental sensor of the vehicle F does not continue to detect the element in the position expected from the map data. In contrast, the second traffic light A2 is still in the position expected from the map data. In this respect, the first presence probability P1 is reduced from 35% to 5%, and the second presence probability P2 is increased from 70% to 99%. Based on the first presence probability P1 and the second presence probability P2, it is determined that a possible critical situation caused by the absence of the first traffic light A1 is highly likely, so that the control of the at least partially automated vehicle is to bring the vehicle F to a safe stop before the intersection. Here, the control mode of the vehicle F is changed from "preventive safety" to "safety stop". In this situation, a remote operation of the vehicle F is requested from the monitoring center, whereby the traffic situation of the vehicle F is manually resolved. Furthermore, an update of the map data or updated measurements of the intersection are initiated.

[0081] FIG. 2 shows a map verification method including the following steps. In a first step S10, vehicle sensor data indicating at least one detected element is received, and the at least one detected element represents an environmental element of the vehicle detected by an environmental sensor of the vehicle. In a second step S20, map data indicating a map having at least one map element is received, and the at least one map element represents an environmental element of the vehicle plotted on a predetermined map. In a third step S30, positioning data is determined from the received sensor data, and the positioning data indicates the position of the vehicle on the map. In a fourth step S40, data uncertainty is determined, and the data uncertainty includes sensor data uncertainty, map data uncertainty, and / or positioning data uncertainty. In a fifth step S50, the existence probability of at least one map element is initialized. In a sixth step S60, the existence probability of at least one map element is updated using the map data, the sensor data, the positioning data, and the data uncertainty.

[0082] FIG. 3a shows a second traffic situation V2 at a first time step.

[0083] The at least partially automated driving vehicle F is arranged near an intersection. From the map data of a high-precision (HD) map, three map elements, namely, a third traffic signal A3, a fourth traffic signal A4, and a fifth traffic signal A5, are seen. In this case, the third traffic signal A3, the fourth traffic signal A4, and the fifth traffic signal A5 actually still exist. In this regard, the map data is real.

[0084] However, a third detection element S3 that can be assigned to the third traffic signal A3 and a fourth detection element S4 that can be assigned to the fourth traffic signal A4 result from the sensor data.

[0085] According to the map verification method, for the third traffic signal A3, the fourth traffic signal A4, and the fifth traffic signal A5, the third probability of existence P3, the fourth probability of existence P4, and the fifth probability of existence P5 are initialized with a value of 50%. In the first time step, a credibility check is performed in which the third probability of existence P3, the fourth probability of existence P4, and the fifth probability of existence P5 are updated. In the credibility check, based on the internal statistics and positioning data of the vehicle F, it is determined that the vehicle F is close enough to the intersection where the environmental sensor of the vehicle F should be able to detect the third traffic signal A3, the fourth traffic signal A4, and the fifth traffic signal A5. However, the sensor data from the vehicle F also reveals that there are trucks L, LKW in the vicinity of the vehicle F within the line of sight between the environmental sensor of the vehicle F and the fifth traffic signal A5. In other words, the third traffic signal A3 and the fourth traffic signal A4 are within the field of view of the environmental sensor of the vehicle F, and the fifth traffic signal A5 is not within the field of view of the environmental sensor of the vehicle F. Therefore, it is determined that no further statement can be made about the fifth probability of existence P5, but statements can be made about the third probability of existence P3 and the fourth probability of existence P4. In this regard, in the first time step, the fifth probability of existence P5 remains at 50%, and the third probability of existence P3 and the fourth probability of existence P4 are increased from 50% to 77.1%.

[0086] Figure 3b shows the second traffic situation V2 in the second time step.

[0087] Furthermore, from the sensor data, only the third detection element S3 that can be assigned to the third traffic signal A3 and the fourth detection element S4 that can be assigned to the fourth traffic signal A4 exist.

[0088] At the second time step, a credibility check is performed in which the third probability of presence P3, the fourth probability of presence P4, and the fifth probability of presence P5 are updated. In the credibility check, based on the internal statistics and location data of vehicle F, it is determined that vehicle F is still not close enough to the intersection where the environmental sensors of vehicle F should be able to detect the third traffic signal A3, the fourth traffic signal A4, and the fifth traffic signal A5. However, the sensor data of vehicle F continues to reveal that there are trucks L, LKW in the vicinity of vehicle F within the field of view between the environmental sensors of vehicle F and the fifth traffic signal A5. In other words, the third traffic signal A3 and the fourth traffic signal A4 are within the field of view of the environmental sensors of vehicle F, and the fifth traffic signal A5 is not within the field of view of the environmental sensors of vehicle F. Therefore, it is determined that no further statement can be made about the fifth probability of presence P5, but statements can be made about the third probability of presence P3 and the fourth probability of presence P4. In this regard, at the second time step, the fifth probability of presence P5 remains at 50%, and the third probability of presence P3 and the fourth probability of presence P4 are increased from 77.1% to 99.5% respectively.

[0089] Figure 3c shows the second traffic situation V2 at the third time step.

[0090] Here, the sensor data yields a third detection element S3 that can be assigned to the third traffic signal A3, a fourth detection element S4 that can be assigned to the fourth traffic signal A4, and a fifth detection element S5 that can be assigned to the fifth traffic signal A5.

[0091] At the third time step, a credibility check is performed in which the third probability of existence P3, the fourth probability of existence P4, and the fifth probability of existence P5 are updated. In the credibility check, based on the internal statistics and positioning data of vehicle F, it is determined that vehicle F is still close enough to the intersection for the environmental sensors of vehicle F to be able to detect the third traffic signal A3, the fourth traffic signal A4, and the fifth traffic signal A5. Also, the sensor data of vehicle F reveals that the truck L in the vicinity of vehicle F is no longer within the field of view between the environmental sensor of vehicle F and the fifth traffic signal A5. That is, the third traffic signal A3, the fourth traffic signal A4, and the fifth traffic signal are within the field of view of the environmental sensors of vehicle F. Thereby, it is determined that statements can be made about the fifth probability of existence P5, the third probability of existence P3, and the fourth probability of existence P4. In this regard, at the third time step, the fifth probability of existence P5 increases from 50% to 69.5%, and the third probability of existence P3 and the fourth probability of existence P4 are increased from 99.5% to 100%.

[0092] Figure 4 shows a map verification system 10 including an object mapping unit 11 and an object fusion unit 12. The map verification system 10 is coupled as an input to a sensor system 20, a positioning system 30, and a map system 40. Further, the map verification system 10 is coupled as an output to an action planning system 50.

[0093] The sensor system 20 provides sensor data Ds. The map system 40 provides map data Dk. The positioning system 30 determines positioning data Dl from the sensor data Ds and the map data Dk. The sensor data Ds, the map data Dk, and the positioning data Dl are provided to the map verification system 10, particularly to the object mapping unit 11.

[0094] The object mapping unit 11 is specific to the sensor type and is initialized once for each sensor. The object mapping unit 11 is configured to project map elements onto each sensor space, for example, onto a camera coordinate system, and to estimate the visibility of map elements in the sensor space, for example, based on stixels that potentially block the line of sight to these map elements. The object mapping unit 11 is further configured to calculate the k best associations, for example, using Marty's algorithm for the probabilistic formulation of sensor data and a ranked assignment algorithm. The assignment of objects is outsourced from the credibility check of the actual map. This is not specific to the map, and the result can be used in other modules such as position determination. Thus, the object mapping unit 11 projects map elements into the sensor space, or in other words, into the sensor coordinate system, specifically estimates the visibility of map elements, and assigns sensor data, i.e., sensor measurements, to depict map elements in the sensor space.

[0095] The object fusion unit 12 is configured to update the probability of existence of map elements and, optionally, also the position of map elements. Further, the object fusion unit 12 is configured to identify sensor data, i.e., sensor measurements not assigned to any existing map element, and thus potentially identify new map elements. The object fusion unit 12 is configured to check the accuracy of algorithmic / probabilistic assumptions, i.e., to check the validity of the probability of existence.

[0096] As a result, the map verification system 10 provides the map elements with the probability of existence P, the unassigned measurement Mu, and the validity check result V to the action planning system 50. As a result, the action planning system 50 controls the behavior of the automated driving vehicle, at least partially, based on the provided data.

Claims

1. A map verification method, comprising: receiving (S10) sensor data of at least a semi-autonomous robot (F) for depicting at least one detection element (S3, S4, S5) representing an environmental element of the at least semi-autonomous robot (F) as detected by an environmental sensor of the at least semi-autonomous robot (F); receiving (S20) map data (Dk) depicting a map having at least one map element (A3, A4, A5) representing an environmental element of the at least semi-autonomous robot (F) as plotted on a predetermined map; receiving (S30) position identification data (Dl) indicating the position of the at least semi-autonomous robot (F) on the map; determining (S40) data uncertainty, the data uncertainty including sensor data uncertainty, map data uncertainty, and / or position identification data uncertainty; projecting at least one map element (A3, A4, A5) onto the sensor space of the environmental sensor; determining the visibility of the map element (A3, A4, A5), which is determined using the field of view of the environmental sensor and the occlusion of the map element (A3, A4, A5); determining a detection probability using the visibility of the map element (A3, A4, A5); initializing (S50) the existence probability (P) for at least one map element (A3, A4, A5) with a predetermined initial value; updating (S60) the existence probability (P) of at least one map element (A3, A4, A5) using the map data (Dk), the sensor data (Ds), the position identification data (Dl), and the data uncertainty; wherein the existence probabilities (P3, P4, P5) of the at least one map element (A3, A4, A5) having the detection probability below a predetermined threshold are not updated. A map verification method.

2. The map verification method according to claim 1, further comprising assigning at least one map element (A3, A4, A5) to at least one detection element (S3, S4, S5).

3. The method according to claim 1 or 2, further comprising evaluating the existence probabilities (P3, P4, P5) of the at least one map element (A3, A4, A5). Evaluating includes one of checking map elements, proving that a map element or potentially new map element is incorrect, and making no statement whatsoever, a map verification method.

4. In the map verification method according to any one of claims 1 to 3, the step of updating the existence probabilities (P3, P4, P5) includes a random finite set (RFS) approach or a logit approach, a map verification method.

5. In the map verification method according to any one of claims 1 to 4, the existence probabilities (P3, P4, P5) are initialized with an initial value of 50%, a map verification method.

6. In the map verification method according to any one of claims 1 to 5, the step of updating the existence probabilities (P3, P4, P5) of the at least one map element (A3, A4, A5) is repeated at time intervals, a map verification method.

7. In the map verification method according to any one of claims 1 to 6, the data uncertainty is used to determine the visibility of the map element (A3, A4, A5), a map verification method.

8. In the map verification method according to any one of claims 1 to 7, comprising the step of verifying the validity of the existence probabilities (P3, P4, P5), a map verification method.

9. In the map verification method according to claim 8, sensor data (Ds) from different sensors of the at least semi-autonomous robot (F) are compared with each other to verify the validity of the existence probabilities (P3, P4, P5), a map verification method.

10. A map verification system (10) configured to execute the method according to any one of claims 1 to 9.

11. A method for controlling at least a semi-autonomous robot, executing the map verification method according to any one of claims 1 to 9 for determining the existence probability of at least one map element; determining a robot trajectory by using sensor data (Ds), map data (Dk), location identification data (Dl), and existence probabilities (P3, P4, P5) of at least one map element (A3, A4, A5); controlling at least the semi-autonomous robot based on the determined robot trajectory; comprising a method.

12. In the method according to claim 11, A step of determining a control mode by using sensor data (Ds), map data (Dk), position identification data (Dl), and existence probabilities (P3, P4, P5) of at least one map element (A3, A4, A5); A step of controlling at least a semi-autonomous robot based on the determined control mode. A method comprising these steps.

Images