OBJECT RECOGNITION BY AN ACTIVE OPTICAL SENSOR SYSTEM

Abstract

According to a method for object recognition by an active optical sensor system (2), a detector unit (2b) detects light (3b) reflected off an object (4) and generates a sensor signal (5a, 5b, 5c, 5d, 5e, 5f) on the basis thereof. A computing unit (2c) ascertains a first pulse width (D1) defined by a predetermined first limit value (G1) for an amplitude of the sensor signal (5a, 5b, 5c, 5d, 5e, 5f) as well as a second pulse width (D2) defined by a predetermined second limit value (G2). The signal pulse is assigned to one of at least two categories according to at least one predefined signal pulse parameter, and a scatter plot for object recognition is generated, said scatter plot containing exactly one entry for the signal pulse, said entry corresponding to the first pulse width (D1) or to the second pulse width (D2) according to the category to which the signal pulse belongs.

Claims

1. A method for object recognition by an active optical sensor system, comprising: registering light reflected by an object in an environment of the sensor system by a detector unit of the sensor system and a sensor signal is generated on the basis of the registered light; determining a first pulse width of a signal pulse of the sensor signal by a computer unit, the first pulse width being established by a predetermined first limit value for an amplitude of the sensor signal; wherein the computer unit is configured to: determine a second pulse width of the signal pulse, the second pulse width being established by a predetermined second limit value for the amplitude of the sensor signal; assign the signal pulse to one of at least two categories as a function of at least one predefined parameter of the signal pulse, and generate a point cloud for object recognition, which contains precisely one entry for the signal pulse, the entry corresponding either to the first pulse width or to the second pulse width as a function of the category of the signal pulse.

2. The method as claimed in claim 1, wherein a radial distance of the object from the sensor system is determined by the computer unit as a function of the signal pulse; and the signal pulse is assigned to the one of at least two categories as a function of the radial distance.

3. The method as claimed in claim 2, wherein the second limit value is greater than the first limit value; the signal pulse is assigned to a first category of the at least two categories if the radial distance is greater than a predefined limit distance; and the point cloud is generated with the entry which corresponds to the first pulse width if the signal pulse has been assigned to the first category.

4. The method as claimed in claim 3, wherein the signal pulse to a second category of the at least two categories if the radial distance is less than the limit distance and the second pulse width is greater than zero; and the point cloud is generated with the entry which corresponds to the second pulse width if the signal pulse has been assigned to the second category.

5. The method as claimed in claim 3, wherein the at least two categories contain at least three categories; the signal pulse to a third category of the at least three categories if the radial distance is less than the limit distance and the second pulse width is equal to zero; and the point cloud is generated with the entry which corresponds to the first pulse width if the signal pulse has been assigned to the third category.

6. The method as claimed in claim 1, wherein the entry contains an identifier that indicates the category to which the signal pulse has been assigned.

7. The method as claimed in claim 1, wherein the object is classified automatically on the basis of the point cloud while taking the entry into account.

8. The method as claimed in claim 1, wherein the second limit value is greater than the first limit value and the first limit value is greater than a predetermined noise level of the detector unit; and/or the second limit value is greater than the first limit value, and the second limit value is greater than a predetermined saturation limit value of the detector unit.

9. A method for the at least partially automatic control of a motor vehicle, comprising: generating a point cloud for object recognition by a method as claimed in claim 1; and controlling the motor vehicle at least partially automatically as a function of the point cloud.

10. The method as claimed in claim 9, wherein the point cloud is generated by object classification automatically on the basis of the point cloud while taking the entry into account; and the motor vehicle is controlled at least partially automatically as a function of a result of the classification.

11. An active optical sensor system, comprising: a detector unit, which is adapted to register light reflected by an object in an environment of the sensor system and to generate a sensor signal on the basis of the registered light; a computer unit, which is adapted to determine a first pulse width of a signal pulse of the sensor signal, the first pulse width being established by a predetermined first limit value for an amplitude of the sensor signal; and wherein the computer unit is adapted to: determine a second pulse width of the signal pulse, the second pulse width being established by a predetermined second limit value for the amplitude of the sensor signal; assign the signal pulse to one of at least two categories as a function of at least one predefined parameter of the signal pulse; and generate a point cloud for object recognition, which contains precisely one entry for the signal pulse, the entry corresponding either to the first pulse width or to the second pulse width as a function of the category of the signal pulse.

12. An electronic vehicle guidance system for a motor vehicle comprising: an active optical sensor system as claimed in claim 11; and a control device, which is adapted to generate at least one control signal as a function of the point cloud, in order to control the motor vehicle at least partially automatically.

13.-14. (canceled)

15. A non-transitory computer-readable storage medium which stores a computer program which, when executed by an electronic vehicle guidance system as claimed in claim 12, cause the vehicle guidance system to perform a method comprising: registering light reflected by an object in an environment of the sensor system by a detector unit of the sensor system and generating a sensor signal on the basis of the registered light; determining a first pulse width of a signal pulse of the sensor signal by a computer unit, the first pulse width being established by a predetermined first limit value for an amplitude of the sensor signal; the computer unit being further configured to: determine a second pulse width of the signal pulse, the second pulse width being established by a predetermined second limit value for the amplitude of the sensor signal; assign the signal pulse to one of at least two categories as a function of at least one predefined parameter of the signal pulse; and generate a point cloud for object recognition, which contains precisely one entry for the signal pulse, the entry corresponding either to the first pulse width or to the second pulse width as a function of the category of the signal pulse.

Description

[0066] In the figures:

[0067] FIG. 1 shows a schematic representation of a motor vehicle with an exemplary embodiment of an electronic vehicle guidance system according to the improved concept;

[0068] FIG. 2 shows a schematic representation of sensor signals of a detector unit of an exemplary embodiment of an active optical sensor system according to the improved concept;

[0069] FIG. 3 shows a schematic representation of further sensor signals of a detector unit of a further exemplary embodiment of an active optical sensor system according to the improved concept;

[0070] FIG. 4 shows a schematic representation of further sensor signals of a detector unit of a further exemplary embodiment of an active optical sensor system according to the improved concept;

[0071] FIG. 5 shows a schematic representation of a camera image and a point cloud generated by a further exemplary embodiment of an active optical sensor system according to the improved concept;

[0072] FIG. 6 shows a schematic representation of a camera image and a point cloud; and

[0073] FIG. 7 shows a schematic representation of a camera image and a point cloud.

[0074] FIG. 1 schematically represents a motor vehicle 1 which has an exemplary embodiment of a vehicle guidance system 6 according to the improved concept.

[0075] The electronic vehicle guidance system 6 has, in particular, an active optical sensor system 2 according to the improved concept. Optionally, the vehicle guidance system 6 may also have a control device 7.

[0076] The active optical sensor system 2 has an emitter unit 2a, which contains for example an infrared laser. The sensor system 2 furthermore has a detector unit 2b, which contains for example one or more optical detectors, for example APDs.

[0077] The sensor system 2 furthermore has a computer unit 2c. Functions of the computer unit 2c which are described below may also be undertaken in various configurations by the control device 7, or vice versa.

[0078] The emitter unit 2a emits laser pulses 3a into the environment of the motor vehicle 1, where they are partially reflected by an object 4 and at least partially reflected back as reflected pulses 3b in the direction of the sensor system 2, and in particular of the detector unit 2b. The detector unit 2b, in particular the optical detectors of the detector unit 2b, registers the reflected components 3b and, on the basis thereof, generate a time-dependent sensor signal which has an amplitude that is proportional to the radiation intensity or radiation power of the registered light 3b. Corresponding examples of various signal pulses are represented in FIG. 2 and FIG. 3.

[0079] The computer unit 2c determines a first time interval, during which the sensor signal 5a, 5b, 5c, 5d exceeds a first limit value G1. This first time interval then corresponds to the first pulse width D1 of the corresponding signal pulse. In the same way, the computer unit 2c determines a second pulse width D2 by corresponding comparison of the sensor signal sensor signal 5a, 5b, 5c, 5d with a second limit value G2, which is greater than the first limit value G1.

[0080] The computer unit 2c or the control device 7 may then determine a property of the object 4, for example a reflectivity or an extent of the object 4, on the basis of the first pulse width D1 and the second pulse width D2.

[0081] In particular, the computer unit 2c or the control device 7 may classify the object 4 as a function of the property or of the pulse widths D1, D2.

[0082] On the basis of a result of the classification, or on the basis of the property of the object, the control device 7 then for example generates control signals in order to control the motor vehicle 1 at least partially automatically.

[0083] FIG. 2 shows two exemplary sensor signals 5a, 5b. While the signal pulse of the sensor signal 5a is reaching a saturation limit value GS, its amplitude thus in particular exceeds the second limit value G2, but this does not apply for the signal pulse of the sensor signal 5b. However, both sensor signals exceed the first limit value G1. Consequently, the first pulse width D1 is greater than zero for both sensor signals 5a, 5b. The second pulse width D2, on the other hand, is greater than zero only for the sensor signal 5a and is equal to zero for the sensor signal 5b.

[0084] If APDs are used as optical detectors, for example, the saturation limit value GS may for example be of the order of a few hundreds of mV, for example lying between 100 mV and 1000 mV.

[0085] FIG. 3 shows a further example of two further sensor signals 5c, 5d, Here, both the first pulse width D1 and the second pulse width D2 are greater than zero for both sensor signals 5c, 5d.

[0086] The computer unit 2c generates a point cloud on the basis of the signal pulses, in particular respectively on the basis either of the first pulse width D1 or of the second pulse width D2. For this purpose, the signal pulses are assigned to one of three categories I, II, III, as represented schematically in FIG. 4.

[0087] If the radial distance r which has been determined for the signal pulse on the basis of the time-of-flight measurement is greater than a predefined limit distance R, it is unlikely that an object 4 in the vicinity of the motor vehicle 1 will reflect light with a sufficiently high intensity for the second limit value G2 to be exceeded, and the second pulse width D2 is accordingly greater than zero. Accordingly, the signal pulse is in this case assigned to a first category I. For signal pulses of category I, the corresponding entry for the signal pulse is generated in such a way that it reproduces the first pulse width D1.

[0088] If the radial distance r is less than the limit distance R, however, it may be assumed that pulses which do not reach the second limit value G2 are due to noise effects or are unreliable for other reasons.

[0089] Accordingly, signal pulses for which the particular radial distance is less than the limit distance R, and whose second pulse width D2 is greater than zero, are assigned to a second category II and those signal pulses for which the radial distance r is less than the limit distance R, and whose second pulse width D2 is equal to zero, are assigned to a third category III.

[0090] For signal pulses of category II, the entry of the point cloud relates to the second pulse width D2, and for signal pulses of category III the entry relates to the first pulse width D1.

[0091] FIG. 5 schematically represents a point cloud 9 which has been generated on the basis of the improved concept as described. Only those points which correspond to category I or category II are represented in the point cloud 9. Signal pulses of category III are not represented. This is possible by each entry being assigned an identifier or flag, which indicates the corresponding category I, II, III, or such an identifier being stored for each entry.

[0092] In this way, the points of category III may be filtered out since with a high probability they are due to noise effects.

[0093] By signal pulses having a radial distance r greater than the limit distance R being taken into account with the second pulse width D2, however, points are represented in the point cloud 9 even for relatively large distances from the sensor system.

[0094] FIG. 5 also schematically represents a corresponding camera image 8.

[0095] FIG. 6 schematically represents a further point cloud. In the point cloud of FIG. 6, the same signal pulses were used as a basis as were used as a basis for generating the point cloud 9 of FIG. 5. For FIG. 6, however, the second pulse width D2 was used for each entry. As may be seen from a comparison of FIG. 5 and FIG. 6, the effective range in the point cloud of FIG. 6 is less than for the point cloud 9 of FIG. 5.

[0096] FIG. 7 shows a further point cloud, which was again based on the same signal pulses. For FIG. 7, however, the first pulse width D1 was used for all signal pulses. Correspondingly, it 5 may be seen that the range is comparable with the one that is provided by the point cloud 9 of FIG. 5. In the near field, however, the point cloud of FIG. 7 contains additional points which with a high probability are due to noise.

[0097] The improved concept, as described, makes it possible to provide a point cloud for object recognition or object classification, which on the one hand has a lesser influence of noise and, on the other hand, emulates a high sensitivity of the sensor system. Furthermore, the improved concept saves on memory space, particularly in comparison with the storage of two complete point clouds.

[0098] In various configurations of the improved concept, a hybrid point cloud is thus generated in order to convert the analog signal pulses into the discrete domain. For example, each point may be assigned a flag according to how great the radial distance from the sensor system is and how high the maximum amplitude of the signal pulse is. In this way, points whose radial distance is small and whose maximum amplitude is likewise small may be filtered out since they are due to interference or noise.