Autonomous agent operation using histogram images
An apparatus, including an interface configured to receive images with measured distance information of an environment in which an autonomous agent is designed to operate; and processing circuitry that is configured to: generate distance histogram images over time, wherein the distance histogram images include the measured distance information in corresponding picture elements; perform a distribution-based outlier analysis on the distance histogram images to classify each picture element of each of the received images as either an outlier picture element representing a dynamic object or a non-outlier picture element representing a static portion of the environment; and track a dynamic object over time and cause an action by the autonomous agent if it is determined, based on a result of the distribution-based outlier analysis, a distance between the dynamic object and the autonomous agent is less than a predefined distance.
1 . An apparatus, comprising:
an interface configured to receive images with measured distance information of an environment of an autonomous agent; and
processing circuitry that is configured to:
generate distance histogram images over time, wherein the distance histogram images include the measured distance information in corresponding picture elements;
perform a distribution-based outlier analysis on the distance histogram images to classify each picture element of each of the received images; and
track a dynamic object over time and cause an action by the autonomous agent if it is determined, based on a result of distribution-based outlier analysis, that a distance between the dynamic object and the autonomous agent is less than a predefined distance.
2 . The apparatus of claim 1 , wherein the processing circuitry is configured to:
perform the distribution-based outlier analysis on the distance histogram images to classify each picture element of each of the received images as either an outlier picture element representing the dynamic object or a non-outlier picture element representing a static portion of the environment.
3 . The apparatus of claim 2 , wherein the processing circuitry is configured to:
identify, based on location information of the autonomous agent, any outlier picture elements representing a portion of the autonomous agent, and ignore any identified outlier picture elements.
4 . The apparatus of claim 2 , wherein the processing circuitry is configured to:
fuse the outlier picture elements representing the dynamic object based on the received images from a plurality of sensors with different fields-of-view; and
identify the dynamic object based on the fused outlier picture elements.
5 . The apparatus of claim 2 , wherein the processing circuitry is configured to:
generate dynamic images comprising the outlier picture elements representing the dynamic object alone.
6 . The apparatus of claim 2 , wherein the processing circuitry is configured to:
generate a static image comprising the non-outlier picture elements representing the static portion of the environment alone, wherein the static image is a complement of a dynamic image comprising the outlier picture elements representing the dynamic object.
7 . The apparatus of claim 1 , wherein the distance histogram images are based on Gaussian distributions.
8 . The apparatus of claim 1 , wherein the processing circuitry is configured to:
convert the measured distance information into a cartesian space of the autonomous agent.
9 . The apparatus of claim 1 , wherein the processing circuitry is located in a cloud or at an edge.
10 . The apparatus of claim 2 , wherein the processing circuitry is configured to classify an individual picture element as an outlier picture element or a non-outlier picture element by comparing its Gaussian distribution against Gaussian distributions of other picture elements.
11 . A non-transitory computer-readable medium having stored thereon instructions that, when executed by processing circuitry, cause the processing circuitry to:
receive images with measured distance information of an environment of an autonomous agent;
generate distance histogram images over time, wherein the distance histogram images include the measured distance information in corresponding picture elements;
perform a distribution-based outlier analysis on the distance histogram images to classify each picture element of each of the received images; and
track a dynamic object over time and cause an action by the autonomous agent if it is determined, based on a result of the distribution-based outlier analysis, that a distance between the dynamic object and the autonomous agent is less than a predefined distance.
12 . The non-transitory computer-readable medium of claim 11 , wherein the instructions further cause the processing circuitry to:
perform the distribution-based outlier analysis on the distance histogram images to classify each picture element of each of the received images as either an outlier picture element representing the dynamic object or a non-outlier picture element representing a static portion of the environment.
13 . The non-transitory computer-readable medium of claim 12 , wherein the instructions further cause the processing circuitry to:
identify, based on location information of the autonomous agent, any outlier picture elements representing a portion of the autonomous agent, and ignore any identified outlier picture elements.
14 . The non-transitory computer-readable medium of claim 12 , wherein the instructions further cause the processing circuitry to:
generate dynamic images comprising the outlier picture elements representing the dynamic object alone.
15 . The non-transitory computer-readable medium of claim 12 , wherein the instructions further cause the processing circuitry to:
generate a static image comprising the non-outlier picture elements representing the static portion of the environment alone, wherein the static image is a complement of a dynamic image comprising the outlier picture elements representing the dynamic object.
16 . The non-transitory computer-readable medium of claim 11 , wherein the distance histogram images are based on Gaussian distributions.
17 . The non-transitory computer-readable medium of claim 11 , wherein the instructions further cause the processing circuitry to:
convert the measured distance information into a cartesian space of the autonomous agent.
18 . The non-transitory computer-readable medium of claim 11 , wherein the processing circuitry is located in a cloud or at an edge.
19 . The non-transitory computer-readable medium of claim 12 , wherein the instructions further cause the processing circuitry to classify an individual picture element as an outlier picture element or a non-outlier picture element by comparing its Gaussian distribution against Gaussian distributions of other picture elements.
20 . The non-transitory computer-readable medium of claim 12 , wherein the instructions further cause the processing circuitry to:
fuse the outlier picture elements representing the dynamic object based on the received images from a plurality of sensors with different fields-of-view; and
identify the dynamic object based on the fused outlier picture elements.
21 . An apparatus, comprising:
an interface means for receiving images with measured distance information of an environment of an autonomous agent; and
processing means for:
generating distance histogram images over time, wherein the distance histogram images include the measured distance information in corresponding picture elements;
performing a distribution-based outlier analysis on the distance histogram images to classify each picture element of each of the received images; and
tracking a dynamic object over time and cause an action by the autonomous agent if it is determined, based on a result of the distribution-based outlier analysis, that a distance between the dynamic object and the autonomous agent is less than a predefined distance.
22 . The apparatus of claim 21 , wherein the processing means is for:
performing the distribution-based outlier analysis on the distance histogram images to classify each picture element of each of the received images as either an outlier picture element representing the dynamic object or a non-outlier picture element representing a static portion of the environment.
23 . The apparatus of claim 22 , wherein the processing means is for:
identifying, based on location information of the autonomous agent, any outlier picture elements representing a portion of the autonomous agent, and ignore any identified outlier picture elements.
24 . The apparatus of claim 22 , wherein the processing means is for:
generating dynamic images comprising the outlier picture elements representing the dynamic object alone.
25 . The apparatus of claim 22 , wherein the processing means is for:
generating a static image comprising the non-outlier picture elements representing the static portion of the environment alone, wherein the static image is a complement of a dynamic image comprising the outlier picture elements representing the dynamic object.