Systems and methods for determining an object type and an attribute for an observation based on fused sensor data
This document discloses system, method, and computer program product embodiments for controlling a vehicle. For example, the method includes: receiving an observation probability distribution function associated with a target object that was detected by sensor(s) of an autonomous vehicle (AV); identifying a target attribute associated with the target object; detecting a target attribute value associated with the target attribute; and issuing vehicle control instruction(s) that cause AV to adjust driving operation(s) using a future behavior of the target object predicted based on an attribute probability distribution function that defines a probability that the target attribute is actually present for the target object based on probability distribution function(s), wherein the attribute probability distribution function comprises: a probability value associated with the target attribute being present for the target object; and a probability value associated with the target attribute not being present for the target object.
1. A method comprising:
receiving an observation probability distribution function associated with a target object that was detected by one or more sensors of an autonomous vehicle;
identifying a target attribute associated with the target object;
detecting a target attribute value associated with the target attribute; and
issuing one or more vehicle control instructions that cause the autonomous vehicle to adjust one or more driving operations using a future behavior of the target object predicted based on an attribute probability distribution function that defines a probability that the target attribute is actually present for the target object based on a plurality of probability distribution functions, wherein the attribute probability distribution function comprises:
a probability value associated with the target attribute being present for the target object; and
a probability value associated with the target attribute not being present for the target object, wherein:
the plurality of probability distribution functions comprises:
a first probability distribution function representing a probability of the autonomous vehicle detecting an object having one or more of object labels based on historical object type detections;
a second probability distribution function defining a probability of the autonomous vehicle detecting the target attribute based on historical attribute detections; and
a third probability distribution function defining a probability of the target attribute actually being present for the target object based on the target attribute value.
2. The method of claim 1 , wherein the second probability distribution function comprises a probability distribution function over an event that the target object is detected as having the target attribute, wherein the second probability distribution function is conditioned on all observations made through a current time.
3. The method of claim 1 , wherein the third probability distribution function is based on:
a set of combinations of attributes and object types; and
for each combination, an associated likelihood value representing a likelihood of the object type of the combination having the attribute of the combination.
4. The method of claim 1 , wherein the historical object type detections are organized in a tree structure.
5. The method of claim 1 , wherein the attribute probability distribution function is based on a sum of a product value across a target object label and the target attribute value, wherein the product value comprises a product of the first probability distribution function, the second probability distribution function, and the third distribution function.
6. The method of claim 1 , wherein the one or more sensors are associated with one or more object type labels, and wherein the observation probability distribution function comprises a detection likelihood value associated with each of the one or more object type labels.
7. A system comprising:
a memory; and
a processor coupled to the memory and configured to:
receive an observation probability distribution function associated with a target object that was detected by one or more sensors of an autonomous vehicle;
identify a target attribute associated with the target object;
detect a target attribute value associated with the target attribute;
execute one or more vehicle control instructions that cause the autonomous vehicle to adjust one or more driving operations using a future behavior of the target object predicted based on an attribute probability distribution function that defines a probability that the target attribute is actually present for the target object based on a plurality of probability distribution functions, wherein the attribute probability distribution function comprises:
a probability value associated with the target attribute being present for the target object, and
a probability value associated with the target attribute not being present for the target object, wherein:
the plurality of probability distribution functions comprises:
a first probability distribution function representing a probability of the autonomous vehicle detecting an object having one or more of object labels based on historical object type detections;
a second probability distribution function defining a probability of the autonomous vehicle detecting the target attribute based on historical attribute detections; and
a third probability distribution function defining a probability of the target attribute actually being present for the target object based on the target attribute value.
8. The system of claim 7 , wherein the second probability distribution function comprises a probability distribution function over an event that the target object is detected as having the target attribute, wherein the second probability distribution function is conditioned on all observations made through a current time.
9. The system of claim 7 , wherein the third probability distribution function is based on:
a set of combinations of attributes and object types; and
for each combination, an associated likelihood value representing a likelihood of the object type of the combination having the attribute of the combination.
10. The system of claim 7 , wherein the historical object type detections are organized in a tree structure.
11. The system of claim 7 , wherein the attribute probability distribution function is based on a sum of a product value across a target object label and the target attribute value, wherein the product value comprises a product of the first probability distribution function, the second probability distribution function, and the third distribution function.
12. The system of claim 7 , wherein the one or more sensors are associated with one or more object type labels, and wherein the observation probability distribution function comprises a detection likelihood value associated with each of the one or more object type labels.
13. A non-transitory computer-readable medium that stores instructions that is configured to, when executed by at least one computing device, cause the at least one computing device to perform operations comprising:
receiving an observation probability distribution function associated with a target object that was detected by one or more sensors of an autonomous vehicle;
identifying a target attribute associated with the target object;
detecting a target attribute value associated with the target attribute; and
issuing one or more vehicle control instructions that cause the autonomous vehicle to adjust one or more driving operations using a future behavior of the target object predicted based on an attribute probability distribution function that defines a probability that the target attribute is actually present for the target object based on a plurality of probability distribution functions, wherein the attribute probability distribution function comprises:
a probability value associated with the target attribute being present for the target object; and
a probability value associated with the target attribute not being present for the target object, wherein:
a first probability distribution function representing a probability of the autonomous vehicle detecting an object having one or more of object labels based on historical object type detections,
a second probability distribution function defining a probability of the autonomous vehicle detecting the target attribute based on historical attribute detections, and
a third probability distribution function defining a probability of the target attribute actually being present for the target object based on the target attribute value.
14. The non-transitory computer-readable medium of claim 13 , wherein the second probability distribution function comprises a probability distribution function over an event that the target object is detected as having the target attribute, wherein the second probability distribution function is conditioned on all observations made through a current time.
15. The non-transitory computer-readable medium of claim 13 , wherein the third probability distribution function is based on:
a set of combinations of attributes and object types; and
for each combination, an associated likelihood value representing a likelihood of the object type of the combination having the attribute of the combination.
16. The non-transitory computer-readable medium of claim 13 , wherein the attribute probability distribution function is based on a sum of a product value across a target object label and the target attribute value, wherein the product value comprises a product of the first probability distribution function, the second probability distribution function, and the third distribution function.
17. The non-transitory computer-readable medium of claim 13 , wherein the one or more sensors are associated with one or more object type labels, and wherein the observation probability distribution function comprises a detection likelihood value associated with each of the one or more object type labels.