Generating data using radar observation model based on machine learning

1. An autonomous vehicle control system comprising one or more processors and one or more memories operably coupled with the one or more processors, wherein the one or more memories store instructions that, in response to the execution of the instructions by one or more processors, cause the one or more processors to perform the following operations:

generating a first path estimation of an object;

generating, based on the first path estimation of the object and first sensor measurement data, second sensor measurement data as output data of a machine learning model;

generating, based on the first path estimation of the object, predicted position of the object;

generating, based on the second sensor measurement data and the predicted position of the object, a second path estimation of the object; and

operating a vehicle based on the second path estimation.

2. The autonomous vehicle control system of claim 1 , further comprising:

obtaining, from one or more radar sensors, the first sensor measurement data.

3. The autonomous vehicle control system of claim 1 , wherein

the first path estimation is associated with a first time, and

the predicted position, the first sensor measurement data, and the second path estimation are associated with a second time that is later than the first time.

4. The autonomous vehicle control system of claim 3 , wherein generating the predicted position of the object comprises:

generating the predicted position of the object by using a motion model based on a time difference between the first time and the second time.

5. The autonomous vehicle control system of claim 1 , wherein the first sensor measurement data represents a plurality of radar points, each radar point being associated with at least one or more values of range, azimuth, range rate, or intensity.

6. The autonomous vehicle control system of claim 1 , wherein generating the second sensor measurement data by the machine learning model comprises:

generating a predicted bounding box of a first path represented by the first path estimation;

sampling radar points from the first sensor measurement data;

associating the sampled radar points with a side of the predicted bounding box of the first path;

encoding the sampled radar points into feature vectors; and

generating the second sensor measurement data as output data of the machine learning model which has the feature vectors as input data thereof.

7. The autonomous vehicle control system of claim 6 , wherein the feature vectors comprises one or more input channels, each input channel comprising a plurality of azimuth angle bins representing a field of view of a plurality of radar points.

8. The autonomous vehicle control system of claim 7 , wherein the plurality of azimuth angle bins of each input channel contain values representing one of maximum range, minimum range, maximum range rate, minimum range rate, or maximum intensity.

9. The autonomous vehicle control system of claim 1 , further comprising:

generating a proposal that represents a third path estimation of the object, which is different from the first path estimation;

generating, based on the proposal, a combined path estimation of at least the second path estimation and the third path estimation; and

performing autonomous control of the vehicle based on the combined path estimation.

10. The autonomous vehicle control system of claim 9 , wherein the proposal is generated without using a machine learning model.

11. The autonomous vehicle control system of claim 9 , wherein the proposal is generated based on the first sensor measurement data.

12. The autonomous vehicle control system of claim 1 , wherein the second sensor measurement data includes data representing a single radar point associated with a center of a bounding box of a path of the object.

13. The autonomous vehicle control system of claim 12 , wherein the second sensor measurement data comprises one or more output channels, each output channel comprising a plurality of azimuth angle bins representing a field of view of the single radar point.

14. The autonomous vehicle control system of claim 13 , wherein the plurality of azimuth angle bins of each output channel contain values representing one of weight, range, or range rate.

15. A method comprising:

generating a first path estimation of an object;

generating, based on the first path estimation of the object and first sensor measurement data, second sensor measurement data as output data of a machine learning model;

generating, based on the first path estimation of the object, a predicted position of the object;

generating, based on the second sensor measurement data and the predicted position of the object, a second path estimation of the object; and

operating a vehicle based on the second path estimation.

16. The method of claim 15 , wherein generating the second sensor measurement data by the machine learning model comprises:

generating a predicted bounding box of a first path represented by the first path estimation;

sampling radar points from the first sensor measurement data;

associating the sampled radar points with a side of the predicted bounding box of the first path;

encoding the sampled radar points into feature vectors; and

generating the second sensor measurement data as output data of the machine learning model which has the feature vectors as input data thereof.

17. The method of claim 16 , wherein the instructions, in response to the execution of the instructions by one or more processors, cause the one or more processors to perform:

generating a proposal that represents a third path estimation of the object, which is different from the first path estimation of the object;

generating, based on the proposal, a combined path estimation of at least the second path estimation and the third path estimation; and

performing autonomous control of the vehicle based on the combined path estimation.

18. An autonomous vehicle, comprising:

at least one of a steering system or a braking system; and

a vehicle controller comprising one or more processors configured to:

generate a first path estimation of an object;

generate, based on the first path estimation of the object and first sensor measurement data, second sensor measurement data as output data of a machine learning model;

generate, based on the first path estimation of the object, a predicted position of the object;

generate, based on the second sensor measurement data and the predicted position of the object, a second path estimation of the object; and

control operation of the at least one of the steering system or the braking system based on the second path estimation.

19. The autonomous vehicle of claim 18 , wherein in generating the second sensor measurement data by the machine learning model, the one or more processors are configured to:

generate a predicted bounding box of a path represented by the first path estimation;

sample radar points from the first sensor measurement data;

associate the sampled radar points with a side of the predicted bounding box of the first path;

encode the sampled radar points into feature vectors; and

generate the second sensor measurement data as output data of the machine learning model which has the feature vectors as input data thereof.

20. The autonomous vehicle of claim 18 , wherein the one or more processors are configured to:

generate a proposal that represents a third path estimation of the object, which is different from the first path estimation;

generate, based on the proposal, a combined path estimation of at least the second path estimation and the third path estimation; and

control operation of the at least one of the steering system or the braking system based on the combined path estimation.