Radar detection filtering approach

1 . A method comprising:

receiving, by a processor from a radar unit, radar data comprising a plurality of values, each value corresponding to one of a plurality of distance bins, wherein each value represents a magnitude of radar returns corresponding to a frequency chirp emitted by the radar unit, wherein each distance bin corresponds to a discrete distance from the radar unit;

determining, by the processor based on the radar data, that a candidate detection event has occurred;

generating, by the processor responsive to determining that the candidate detection event has occurred, a confidence score using signal characteristics of the radar data, the confidence score representing a level of certainty that the candidate detection event comprises actual motion of an object, wherein generating the confidence score using the signal characteristics comprises:

identifying a first value of a first distance bin of the plurality of distance bins, wherein the first value is a first peak of the plurality of values;

identifying a second value of a second distance bin of the plurality of distance bins, wherein the second value is a second peak of the plurality of values lower than the first peak;

determining a first score by generating a ratio between the first value to the second value;

identifying a reference distance bin corresponding to a third distance bin of the plurality of distance bins;

determining a second score based on subtracting a first distance corresponding to the first distance bin from a second distance corresponding to the reference distance bin;

identifying a third value of a fourth distance bin of the plurality of distance bins, wherein the third value is a third peak of the plurality of values, and wherein the third value is lower than the first value and higher than the second value;

determining a third score based on subtracting an average of the first distance and a fourth distance corresponding to the fourth distance bin from the second distance;

identifying a subset of the plurality of distance bins, wherein each distance that corresponds to a distance bin of the subset is within a pre-determined range of distances comprising the second distance;

determining a fourth score based on a count of the values corresponding to one of the subset of the plurality of distance bins which meet a dynamic threshold; and

combining the first, second, third, and fourth scores based on a weighted sum to obtain the confidence score;

determining that the confidence score satisfies a confidence threshold, the confidence threshold representing a minimum level of certainty that the candidate detection event comprises actual motion of an object; and

activating a process to capture video in response to the confidence score satisfying the confidence threshold.

2 . The method of claim 1 , further comprising:

bypassing, responsive to determining that the confidence score satisfies the confidence threshold, a sliding window filter for detecting a predetermined number of candidate detection events within a fixed-size window.

3 . The method of claim 1 , further comprising:

receiving, by the processor from the radar unit, second radar data comprising a second plurality of values, each value corresponding to one of the plurality of distance bins, wherein each value represents a magnitude of radar returns corresponding to a second frequency chirp emitted by the radar unit;

determining, based on the second radar data, that a second candidate detection event has occurred;

generating, responsive to a determination that the second candidate detection event has occurred, a second confidence score using second signal characteristics of the second radar data, the second confidence score representing a level of certainty that the second candidate detection event comprises actual motion of a second object;

determining that the second confidence score does not satisfy the confidence threshold; and

incrementing a counter of a sliding window filter for detecting a predetermined number of candidate detection events within a fixed-size window.

4 . A method comprising:

obtaining radar data from a radar unit;

determining, based on the radar data, that a candidate detection event has occurred;

determining signal characteristics of the candidate detection event, wherein the signal characteristics comprise a plurality of values;

identifying a first value of the plurality of values and a second value of the plurality of values, wherein the second value is greater than a subset of the plurality of values and the first value is greater than the second value;

generating a confidence score using the signal characteristics and based on a ratio between the first value and the second value;

determining that the confidence score satisfies a confidence threshold, the confidence threshold indicating a confidence that the candidate detection event is an actual detection event; and

activating a process to capture one or more of an image or video in response to the confidence score satisfying the confidence threshold.

5 . The method of claim 4 wherein, responsive to determining that the confidence score satisfies the confidence threshold, a sliding window filter signal processing technique for detecting a fixed number of candidate detection events from the radar data within a fixed-size timing window is not utilized.

6 . The method of claim 4 , further comprising:

obtaining second radar data from the radar unit;

determining, based on the second radar data, that a second candidate detection event has occurred;

determining second signal characteristics of the second candidate detection event;

generating a second confidence score using the second signal characteristics;

determining that the second confidence score does not satisfy the confidence threshold; and

incrementing, responsive to a determination that the confidence score does not satisfy the confidence threshold, a counter of a sliding window filter signal processing technique for detecting a fixed number of candidate detection events from the radar data within a fixed-size timing window.

7 . The method of claim 6 , further comprising:

determining a value of the counter satisfies a filter threshold; and

activating a process to capture one or more of an image or video in response to the value of the counter satisfying the filter threshold.

8 . The method of claim 4 , further comprising performing a computer vision analysis on the one or more of the image or the video.

9 . The method of claim 4 , wherein the plurality of values each correspond to one of a plurality of distances from the radar unit, wherein determining the confidence score comprises:

identifying a reference distance that corresponds to a first distance of the plurality of distances, wherein determining that the candidate detection event has occurred uses the reference distance;

identifying a highest value of the plurality of values, wherein the highest value corresponds to a second distance of the plurality of distances; and

determining a third distance between the reference distance and the second distance.

10 . The method of claim 4 , wherein the plurality of values each correspond to one of a plurality of distances from the radar unit, wherein determining the confidence score comprises:

identifying a reference distance that corresponds to a first distance of the plurality of distances, wherein determining that the candidate detection event has occurred uses the reference distance;

identifying a highest value of the plurality of values, wherein the highest value corresponds to a second distance of the plurality of distances,

wherein the second value corresponds to a third distance of the plurality of distances;

generating a fourth distance by averaging the second distance and the third distance; and

determining a fifth distance between the reference distance and the fourth distance.

11 . The method of claim 4 , wherein the plurality of values each correspond to one of a plurality of distances from the radar unit, wherein determining the confidence score comprises:

identifying a reference distance that corresponds to a first distance of the plurality of distances, wherein determining that the candidate detection event has occurred uses the reference distance;

determining a subset of the plurality of distances, wherein the subset comprises the first distance; and

comparing each value corresponding to one of the subset to a dynamic threshold.

12 . The method of claim 11 , further comprising:

determining that a third value satisfies the dynamic threshold, the third value corresponding to a second distance of the subset, wherein the second distance is less than the first distance;

identifying a first weight for the third value;

determining that a fourth value satisfies the dynamic threshold, the fourth value corresponding to a third distance of the subset, wherein the third distance is greater than the first distance;

identifying a second weight for the fourth value, wherein the first weight is different than the second weight; and

combining the first weight and the second weight.

13 . A device comprising:

a radar unit;

a camera;

a memory device to store radar data, wherein the radar data comprises signal characteristics of radar returns observed by the radar unit, wherein the signal characteristics comprise a plurality of values; and

a processor coupled to the memory device, wherein the memory device stores processor executable instructions which, when executed using the processor, cause the device to perform operations comprising:

determining, based on the radar data, that a candidate detection event has occurred;

identify a first value of the plurality of values and a second value of the plurality of values wherein the second value is greater than a subset of the plurality of values and the first value is greater than the second value;

generate a confidence score using the signal characteristics and based on a ratio between the first value and the second value;

determine that the confidence score satisfies a confidence threshold, the confidence threshold indicating a confidence that the candidate detection event is an actual detection event; and

activate a process to capture, using the camera, one or more of an image or video in response to the confidence score satisfying the confidence threshold.

14 . The device of claim 13 wherein, responsive to determining that the confidence score satisfies the confidence threshold, a sliding window filter signal processing technique for detecting a fixed number of candidate detection events from the radar data within a fixed-size timing window is not utilized.

15 . The device of claim 13 , wherein the processor executable instructions, when executed using the processor, cause the device to perform operations comprising:

obtaining second radar data from the radar unit;

determining, based on the second radar data, that a second candidate detection event has occurred;

determining second signal characteristics of the second candidate detection event;

generating a second confidence score using the second signal characteristics;

determining that the second confidence score does not satisfy the confidence threshold; and

incrementing, responsive to a determination that the confidence score does not satisfy the confidence threshold, a counter of a sliding window filter signal processing technique for detecting a fixed number of candidate detection events from the radar data within a fixed-size timing window.

16 . The device of claim 15 , wherein the processor executable instructions, when executed using the processor, cause the device to perform operations comprising:

determining a value of the counter satisfies a filter threshold; and

activating a process to capture one or more of an image or video in response to the counter satisfying the filter threshold.

17 . The device of claim 13 , wherein the plurality of values each correspond to one of a plurality of distances from the radar unit, wherein the processor executable instructions, when executed using the processor cause the device to perform operations to determine the confidence score comprising:

identifying a reference distance that corresponds to a first distance of the plurality of distances, wherein the processor uses the reference distance to determine that the candidate detection event has occurred;

identifying a highest value of the plurality of values, wherein the highest value corresponds to a second distance of the plurality of distances; and

determining a third distance between the reference distance and the second distance.

18 . The device of claim 13 , wherein the plurality of values each correspond to one of a plurality of distances from the radar unit, wherein the processor executable instructions, when executed using the processor, cause the device to perform operations to determine the confidence score comprising:

identifying a reference distance that corresponds to a first distance of the plurality of distances, wherein the processor uses the reference distance to determine that the candidate detection event has occurred;

determining a subset of the plurality of distances, wherein the subset comprises the first distance; and

comparing each value corresponding to one of the subset to a dynamic threshold.