Sky determination in environment detection for mobile platforms, and associated systems and methods
Identifying interfering features in the detection of an environment adjacent to a mobile platform, and associated systems and methods are disclosed herein. A representative method can include identifying candidate regions from a color image obtained by a color vision sensor carried by a mobile platform, filtering out a first region subset of regions based, at least in part, on non-image data obtained by a second sensor carried by the mobile platform, determining a second subset of regions as corresponding to an interfering feature based, at least in part, on color information, and performing environment detection based, at least in part, on the second subset of regions. Furthermore, the method can comprise transforming data corresponding to the subset of regions to integrate with non-color environment data obtained from another sensor carried by the mobile platform.
1. A computer-implemented method for detecting sky using a camera unit and an inertial measurement unit (IMU), both carried by a common autonomous vehicle, the method comprising:
segmenting a two-dimensional (2D) color image obtained by the camera unit into a plurality of regions, wherein the color image depicts at least a portion of an environment surrounding the autonomous vehicle;
determining an indication of the horizon relative to the color image based, at least in part, on sensory data obtained from the IMU;
filtering out a first subset of the regions based, at least in part, on the indication of the horizon to generate a reduced set of regions;
identifying a second subset of regions from the reduced set of regions as corresponding to the sky, based, at least in part, on color information associated with the reduced set of regions, wherein the areas above the horizon line are candidate regions, and statistical information of color-based gradients is used to determine which candidate regions constitute the second subset of regions;
transforming the identified second subset of regions into a reference system of a stereo camera unit, undistorting the data obtained by the stereo camera unit before transforming data corresponding to the identified second subset of regions; and
performing environment detection using at least the stereo camera unit based, at least in part, on exclusion of data corresponding to the second subset of regions, such that the stereo camera unit skips sensing certain parts of the environment.
2. A computer-implemented method for environment sensing using one or more sensors carried by a mobile platform, comprising:
identifying a plurality of regions from an image obtained by a vision sensor carried by the mobile platform;
determining a subset of regions from the plurality of regions as corresponding to an interfering feature, wherein at least one of the identifying the plurality of regions or the determining the subset of regions is based, at least in part, on non-image data obtained by a second sensor carried by the mobile platform; and
performing environment detection based, at least in part, on exclusion of data corresponding to the subset of regions, such that a stereo camera skips sensing certain parts of the environment;
transforming data corresponding to the subset of regions to integrate with data obtained by the stereo camera;
undistorting the data obtained by the stereo camera before transforming data corresponding to the subset of regions,
wherein determining the subset of regions is based, at least in part, on a measurement of gradient, and the measurement of gradient is calculated based, at least in part, on color values.
3. The method of claim 2 , wherein individual regions of the plurality of regions do not overlap with one another.
4. The method of claim 2 , wherein the entire image is divided into the plurality of regions.
5. The method of claim 2 , wherein identifying the plurality of regions comprises:
selecting a target portion from the image based, at least in part, on the non-image data obtained by the second sensor; and
identifying the plurality of regions from the target portion of the image.
6. The method of claim 5 , wherein selecting the target portion from the image comprises determining a line relative to the image based, at least in part, on the non-image data.
7. The method of claim 6 , wherein the target portion of the image resides on one side of the line.
8. The method of claim 2 , wherein identifying the plurality of regions is based, at least in part, on edge detection.
9. The method of claim 2 , wherein the vision sensor includes a monocular color camera.
10. The method of claim 2 , wherein the second sensor includes an inertial measurement unit (IMU).
11. The method of claim 2 , wherein a relative position and/or orientation between the second sensor and the vision sensor is fixed.
12. The method of claim 2 , wherein determining the subset of regions from the plurality of regions as corresponding to an interfering feature further comprises filtering out another subset of regions based, at least in part, on a spatial relationship between the interfering feature and a portion of the image.
13. The method of claim 2 , wherein the interfering feature is at least a threshold distance away from the mobile platform.
14. The method of claim 13 , wherein the interfering feature corresponds to the sky.
15. The method of claim 13 , wherein color associated with the interfering feature conforms to a particular type of color distribution.
16. The method of claim 2 , wherein the stereo camera produces three-dimensional (3D) sensor data.
17. The method of claim 2 , wherein performing environment detection is further based on data obtained by the stereo camera.