Methods for determining the position of an automatically and / or autonomously operating vehicle and automatically and / or autonomously operating vehicle

Abstract

The invention relates to a method for determining the position (P) of an automatically and / or autonomously operating vehicle (10), in which point clouds from contours of an environment of the vehicle (10) are acquired by means of at least one sensor (12) and the position (P) of the vehicle (10) is determined by comparing the point clouds, wherein the point clouds are acquired discontinuously. The invention further relates to an automatically and / or autonomously operating vehicle (10).

Description

[0001] The present invention relates to a method for determining the position of an automatically and / or autonomously operating vehicle, in which point clouds are detected from contours of the vehicle's surroundings using at least one sensor. Furthermore, the invention relates to an automatically and / or autonomously operating vehicle that has at least one optical sensor for detecting point clouds from contours of the vehicle's surroundings and at least one evaluation device configured to determine the position of the vehicle based on the detected point clouds.

[0002] Automatic and / or autonomously operating vehicles are used, for example, in logistics to transport objects. Such vehicles use a so-called simultaneous localization and mapping algorithm (SLAM algorithm), in which the environment is continuously scanned using a scanning unit and point clouds are generated. By evaluating the point clouds using the SLAM algorithm, the position required for navigation can be determined. The challenge of the SLAM algorithm is then to calculate the vehicle's trajectory and generate the point cloud at the same time.

[0003] Due to the error influences caused by the movement of the vehicle including the scanning unit, inaccuracies arise, which the SLAM algorithm attempts to compensate for. However, this correction is only possible to a limited extent, which is why this solution cannot be used for higher scanning accuracy.

[0004] To circumvent this disadvantage, EP 3 926 300 A1 discloses a method for determining the position of an automatically and / or autonomously operating vehicle. In the method, a self-position is determined by means of a first incremental estimate from a point cloud acquired with a first optical sensor and a second position value is determined by means of a second incremental estimate from measurement data of a second non-optical sensor, and the self-position is determined on the basis of the first position value and the second position value. At least when the first position value does not lie within a vicinity of the second position value defined by a tolerance deviation, a modified first position value is determined from the point cloud under the boundary condition that the modified first position value lies within the vicinity of the second position value.

[0005] The present invention is based on the object of creating a method and a vehicle which have an improved position determination, in particular an increased accuracy of the position determination.

[0006] To solve the problem, a method having the features of claim 1 and a vehicle having the features of claim 7 are proposed.

[0007] Advantageous embodiments of the method and the vehicle are the subject of the respective dependent claims.

[0008] According to a first aspect of the invention, a method for determining the position of an automatically and / or autonomously operating vehicle is proposed. In the method, point clouds are acquired from contours of the vehicle's surroundings using at least one sensor, and the position of the vehicle is determined by comparing the point clouds, with the point clouds being acquired discontinuously.

[0009] Due to the discontinuous acquisition of point clouds, the method according to the invention is based on a stop-and-go method for capturing point clouds. This means that the automatically and / or autonomously operating vehicle stops and captures a point cloud using a sensor, for example, an optical sensor. The automatically and / or autonomously operating vehicle then continues driving and stops again after a predetermined time or a predetermined distance, and then captures another point cloud. This process is repeated as often as required to determine the position of the automatically and / or autonomously operating vehicle. This increases the accuracy of the individual capture of the point clouds, as better sensors and a longer capture time can be used.

[0010] While traditional mobile mapping systems currently rely on mobile LIDAR systems in conjunction with SLAM algorithms, the method according to the invention uses discontinuous static scans. Furthermore, the method according to the invention can be used to compare or match point clouds in order to calculate trajectories and enable continuous navigation of an automatically and / or autonomously operating vehicle. Discontinuous acquisition significantly increases the accuracy of individual acquisitions or scans, as better sensors and longer acquisition or scan durations can be used. Furthermore, all terrestrial laser scanners can be used to control a mobile, automatic and / or autonomous vehicle or to continuously provide the required navigation information for the vehicle.

[0011] The method according to the invention now proposes converting the known method into a method for live localization through a live evaluation of the co-registration. This enables an automatic and / or autonomously operating vehicle to transmit its position live from position to position, thus enabling continuous navigation. This is a stop-and-go method, since the scans must be performed statically. However, after calculating the current position, the system sends a trajectory to the next defined waypoint to the vehicle, which then moves on to the next position.

[0012] The position can be determined by comparing the point clouds using an evaluation device. The evaluation device can be a data processing device, such as a computer, on which an algorithm, in particular a discrete SLAM algorithm, is stored. This algorithm compares the acquired point clouds and, based on this, determines the position of the vehicle. In an advantageous embodiment, a local coordinate system is constructed based on the determined position of the vehicle.

[0013] In an advantageous embodiment, viewpoints are defined as waypoints, and thus implicitly also trajectories, in advance in a local coordinate system. A first viewpoint can advantageously be defined at the coordinate origin and transferred to the system. By measuring a first contour at a first viewpoint, the local coordinate system can be defined. The system can then determine the tractor's position and send the vehicle along this trajectory to the next waypoint.

[0014] Advantageously, the point clouds are captured from contours of the vehicle's surroundings in 2D or 3D.

[0015] By comparing the point clouds, the algorithm can determine a trajectory of the vehicle.

[0016] In an advantageous embodiment, the vehicle is stopped to capture the point cloud, and a point cloud consisting of contours of the vehicle's surroundings is captured, after which the vehicle continues driving. Due to the stop-and-go method, the distance between two point clouds that need to be co-registered within the algorithm is greater. However, this significantly increases the accuracy of the individual captures.

[0017] In an advantageous embodiment, a laser scanner is used as the sensor. Furthermore, the laser scanner is advantageously a terrestrial scanner. The laser scanner can be mounted on the vehicle.

[0018] Using a terrestrial laser scanner, a contour of an environment can be captured at a first location. A second contour of the environment can then be captured at a further location. Due to different recording situations (due to different recording positions / fields of view / shadowing), the captured contours differ in some parts. However, by choosing suitable locations, both contours also have common contour parts G. Using pattern recognition methods (and optionally with the aid of an approximate position obtained by additional sensors such as an IMU) and a cloud-to-cloud matching method, the two contours can be co-registered by minimizing the distances in points. This means that the further location in relation to the first location is now known. The final co-registration of point clouds can be performed in post-processing, i.e., after all location points have been recorded.

[0019] In an advantageous embodiment, a trajectory is calculated to determine the position by comparing the recorded point clouds. This determines the course of a spatial curve along which the automatically and / or autonomously operating vehicle moves.

[0020] In an advantageous embodiment, at least one additional sensor is used to determine the position, which determines an absolute and / or relative position in space. The additional sensor can support the robustness of the location determination, which in turn can result in greater accuracy of the entire point cloud, as misclassifications can be avoided.

[0021] In an advantageous embodiment, the additional sensor serves to detect and avoid obstacles or to calculate alternative routes based on prior knowledge of the environment. If a predefined waypoint cannot be reached, a position is set as close as possible to the predefined waypoint. Since all waypoints in the local coordinate system are known, a trajectory can also be calculated in this case.

[0022] In an advantageous embodiment, the additional sensor is a camera, a radio positioning system, and / or an inertial measurement unit (IMU). The use of an additional sensor can support the robustness of the position determination. This results in higher point cloud accuracy, as mismatches can be avoided. The camera can be a 3D camera. Using an inertial measurement unit, acceleration, orientation, angular rates, and / or other gravitational forces can be measured. For this purpose, the inertial measurement unit can comprise acceleration sensors and / or gyroscope sensors.

[0023] In an advantageous embodiment, the system comprises additional sensors, such as IMU or (radio) positioning systems, which can provide a rough position of a point before calculating the cloud-to-cloud matching and thus increase the stability of the process.

[0024] According to a further aspect of the invention, an automatically and / or autonomously operating vehicle is proposed. The vehicle has at least one sensor for detecting point clouds from contours of the vehicle's surroundings and at least one evaluation device configured to determine a position of the vehicle based on the detected point clouds. The evaluation device is configured to execute a method according to the invention to determine the position of the vehicle.

[0025] The evaluation device can be a data processing device, such as a computer, on which an algorithm, in particular a discrete SLAM algorithm, is stored. This algorithm compares the acquired point clouds and, based on this, determines the position of the vehicle. In particular, a trajectory of the vehicle is determined by comparing the point clouds. The evaluation device is operatively connected to the sensor. The evaluation device can be connected to the sensor via radio, for example, Wi-Fi, and / or via a data cable.

[0026] In an advantageous embodiment, the sensor has a laser scanner.

[0027] In an advantageous embodiment, the automatically and / or autonomously operating vehicle has an additional sensor by means of which an absolute and / or relative position in space is determined. The additional sensor can be a camera, a radio positioning system, and / or an inertial measurement unit (IMU). The evaluation device can be operatively connected to the additional sensor. The evaluation device can be connected to the additional sensor via radio, for example, Wi-Fi, and / or via a data cable.

[0028] An automatically and / or autonomously operating vehicle, a method for determining a vehicle's position, and other features and advantages are described below using an exemplary embodiment schematically illustrated in the figures. Herein: Fig. 1 a schematic representation of an automatically and / or autonomously operating vehicle; Fig. 2 a schematic representation of a method for detecting the position of the vehicle; and Fig. 3 a schematic representation of a method for calculating the trajectory.

[0029] In Fig. 1 shows an automatically and / or autonomously operating vehicle 10, for example a carrier system for transporting objects in a warehouse.

[0030] The vehicle 10 has a sensor 12 for detecting point clouds from contours of an environment of the vehicle 10 and at least one evaluation device 14 which is designed to determine a position P of the vehicle 10 on the basis of the detected point clouds.

[0031] The sensor 12 is a laser scanner by means of which point clouds from contours of an environment of the vehicle 10 are recorded in 2D or 3D.

[0032] The evaluation device 14 is operatively connected to the sensor 12 and to a vehicle control system (not shown) via a data cable 16. Alternatively, the evaluation device 14 can be connected to the sensor 12 wirelessly, for example, via Wi-Fi.

[0033] The evaluation device 14 is a data processing device, for example a computer, on which an algorithm, in particular a discrete SLAM algorithm, is stored, by means of which the position P of the vehicle 10 is determined, as will be explained in more detail below.

[0034] One possible method for determining the position P of the vehicle 10, on which the algorithm is based, involves converting the known method described above into a method for live localization through a live evaluation of the co-registration. This enables the automatically and / or autonomously operating vehicle 10 to transmit its position P live from position to position, thus enabling continuous navigation. This is a stop-and-go method, since the scans must be performed statically. However, after calculating the current position P, the evaluation device 14 provides the vehicle 10 with a trajectory to the next defined waypoint, and the vehicle will then move on to the next position.

[0035] As in Fig. 2, the vehicle 10 is initially located at a starting point S. The starting point S forms the zero point for determining the position P of the vehicle 10. At the starting point S, a scan of the contours of the surroundings is carried out using the sensor 12, and a first point cloud is generated from this. This point cloud is transmitted to the evaluation device 14. The vehicle 10 then travels along a roadway 18 and stops after a predetermined distance. At this first position S1, the contours of the surroundings are again scanned using the sensor 12, and a second point cloud is generated from this. The second point cloud is also sent to the evaluation device 14. The algorithm stored on the evaluation device 14 compares the first point cloud with the second point cloud and, based on this, determines the position P of the vehicle 10. The vehicle 10 then travels on and stops at a second position P2.At the second location S2, the sensor 14 scans the contours of the surroundings and generates a third point cloud from them. The third point cloud is sent to the evaluation device 14, and the SLAM algorithm compares the third point cloud with the second point cloud and determines the position P of the vehicle 10 based on this.

[0036] As in Fig. As shown in Figure 3, the viewpoints S1, S2, and S3 are defined as waypoints W1, W2, and W3, and thus implicitly also the trajectories T12 and T23, in advance in a local coordinate system, where S1 can be defined at the coordinate origin, and are transferred to the evaluation device 14. The local coordinate system can be defined by measuring the contour K1 at the viewpoint S1. The evaluation device 14 can now determine the tractor track T12 and send the vehicle 10 along this trajectory to the next waypoint W2.

[0037] In an alternative embodiment not shown, the vehicle 10 has additional sensors, such as IMUs or (radio) positioning systems, which provide a rough position of point S2 before calculating the cloud-to-cloud matching at point S2. The additional sensors merely support the robustness of the location determination, which in turn can result in a higher accuracy of the entire point cloud, as mismatches can be avoided.

[0038] In another alternative embodiment (not shown), the vehicle 10 has sensors to detect and avoid obstacles or to calculate alternative routes based on prior knowledge of the environment. If the defined waypoint W2 cannot be reached, the position S2 is set as close as possible to the predefined waypoint W2. Since all waypoints in the local coordinate system are known, a trajectory T23 can also be calculated in this case.

[0039] The vehicle 10 and the method are characterized by the discontinuous acquisition of point clouds. This significantly increases the accuracy of individual acquisitions or individual scans, as better sensors and longer acquisition or scanning times can be used. List of reference symbols 10 automatic and / or autonomously operating vehicle 12 optical sensor 14 Evaluation device 16 data cables 18 Roadway P Position S starting point S1 first viewpoint S2 second viewpoint QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] EP 3 926 300 A1

[0004]

Claims

[1] Method for determining a position (P) of an automatically and / or autonomously operating vehicle (10), in which point clouds from contours of an environment of the vehicle (10) are detected by means of at least one sensor (12) and the position (P) of the vehicle (10) is determined by comparing the point clouds, wherein the point clouds are detected discontinuously. [2] Method according to claim 1, characterized by that in order to capture the point cloud, the vehicle (10) is stopped and a point cloud of contours of the surroundings of the vehicle (10) is captured and then the vehicle (10) continues driving. [3] Method according to one of the preceding claims, characterized by that a laser scanner is used as the sensor (12). [4] Method according to one of the preceding claims, characterized by that to determine the position (P) a trajectory is calculated by comparing the recorded point clouds. [5] Method according to one of the preceding claims, characterized by that at least one further sensor (12) is used to determine the position (P), by means of which an absolute and / or relative position (P) in space is determined. [6] Method according to one of the preceding claims, characterized by that the further sensor (12) is a camera, a radio location and / or an inertial measuring unit (IMU). [7] Automatically and / or autonomously operating vehicle (10) comprising at least one sensor (12) for detecting point clouds from contours of an environment of the vehicle (10) and at least one evaluation device (14) which is designed to determine a position (P) of the vehicle (10) on the basis of the detected point clouds, wherein the evaluation device (14) is set up to carry out a method according to one of claims 1 to 6 for determining the position (P). [8] Vehicle according to claim 7, characterized bythat the optical sensor (12) has a laser scanner. [9] Vehicle (10) according to claim 7 or 8, characterized by another sensor by means of which an absolute and / or relative position (P) in space is determined.

Images