Estimating in-plane velocity from an arbitrary radar return

1. A method comprising:

receiving, from a radar sensor associated with an autonomous vehicle, radar data comprising a Doppler value associated with a stationary object in an environment surrounding the autonomous vehicle;

determining, as an error, a difference between the Doppler value and a predicted value from a model for the radar data;

altering, as an updated model, one or more parameters of the model for the radar data to obtain an updated model, wherein:

the one or more parameters comprises an estimated velocity of the autonomous vehicle,

altering the one or more parameters of the model for the radar data is based at least in part on an asymmetrical loss function that comprises a first portion and a second portion,

the first portion comprises a square of the error,

the second portion is linearly proportional to the error

the second portion is based at least in part on an estimated velocity of the vehicle, and

altering the one or more one or more parameters comprises at least one of increasing or decreasing the estimated velocity to optimize a total error associated with the radar data;

determining a velocity of the autonomous vehicle based at least in part on the updated model; and

controlling the autonomous vehicle based at least in part on the velocity.

2. The method of claim 1 , determining the velocity comprises:

determining a lateral component of the velocity based at least in part on determining a shift of a sinusoid associated with the model; and

determining a longitudinal component of the velocity based at least in part on at least one of an amplitude of the sinusoid or a reference trajectory.

3. The method of claim 1 , wherein the estimated velocity is based at least in part on at least one of:

a previously-determined velocity;

a random sample consensus associated with the radar data;

sensor data comprising a measured velocity; or

a trajectory indicating a target velocity received from a computing system associated with the autonomous vehicle.

4. The method of claim 1 , further comprising:

determining an azimuthal offset associated with the updated model; and

calibrating the radar sensor by adjusting an orientation associated with the radar sensor based at least in part on the azimuthal offset.

5. The method of claim 1 , wherein the asymmetric loss function comprises at least one of a Huber loss function or a Cauchy loss function.

6. A system comprising:

one or more sensors;

one or more processors; and

a memory storing processor-executable instructions that, when executed by the one or more processors, cause the system to perform operations comprising:

receiving, from a radar sensor on a moveable platform associated with an autonomous vehicle, radar data comprising a plurality of Doppler values;

determining a model that fits the radar data, wherein:

determining the model is based at least in part on adjusting a parameter of the model based at least in part on output of an asymmetrical loss function comprising a first portion and a second portion,

the first portion is quadratic,

the second portion is linear; and

the second portion is based at least in part on at least one of an estimated velocity of the moveable platform or at least one of the plurality of Doppler values;

determining a velocity of the moveable platform based at least in part on the model; and

controlling the autonomous vehicle based at least in part on the velocity.

7. The system of claim 6 , wherein the parameter comprises an estimated velocity of the moveable platform, and

adjusting the parameter comprises at least one of increasing or decreasing the estimated velocity to optimize a total residual associated with the radar data, the total residual indicative of a difference between the estimated velocity and a measured velocity.

8. The system of claim 6 , wherein determining the velocity comprises:

determining a lateral component of the velocity based at least in part on determining a shift of a sinusoid associated with the model; and

determining a longitudinal component of the velocity based at least in part on at least one of an amplitude of the sinusoid or a reference trajectory.

9. The system of claim 6 , wherein the estimated velocity is based at least in part on at least one of:

a previously-determined velocity;

a random sample consensus associated with the radar data;

sensor data comprising a measured velocity; or

a trajectory indicating a target velocity received from a computing device.

10. The system of claim 6 , wherein the second portion penalizes the parameter inversely proportional to at least one of a measured Doppler value or a vertical field of view associated with the radar sensor.

11. The system of claim 6 , wherein the operations further comprise:

determining a phase offset associated with the updated model; and

calibrating the radar sensor by adjusting an orientation associated with the radar sensor based at least in part on the phase offset.

12. A non-transitory computer-readable medium storing processor-executable instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:

receiving, from a radar sensor associated with a moveable platform associated with an autonomous vehicle, radar data comprising a plurality of Doppler values;

determining a model that fits the radar data, wherein:

determining the model is based at least in part on adjusting a parameter of the model based at least in part on output of an asymmetrical loss function comprising a first portion and a second portion,

the first portion is quadratic,

the second portion is linear; and

the second portion scales is based at least in part on at least one of an estimated velocity of the moveable platform or at least one of the plurality of Doppler values;

determining a velocity of the moveable platform based at least in part on the model; and

controlling the autonomous vehicle based at least in part on the velocity.

13. The non-transitory computer-readable medium of claim 12 , wherein the parameter comprises an estimated velocity of the movable platform, and

adjusting the parameter comprises at least one of increasing or decreasing the estimated velocity to optimize a total residual associated with the radar data, the total residual indicative of a difference between the estimated velocity and a measured velocity.

14. The non-transitory computer-readable medium of claim 12 , wherein determining the velocity comprises:

determining a lateral component of the velocity based at least in part on determining a shift of a sinusoid associated with the model; and

determining a longitudinal component of the velocity based at least in part on at least one of an amplitude of the sinusoid or a reference trajectory.

15. The non-transitory computer-readable medium of claim 12 , wherein the estimated velocity is based at least in part on at least one of:

a previously-determined velocity;

a random sample consensus associated with the radar data;

sensor data comprising a measured velocity; or

a trajectory indicating a target velocity received from a computing device.

16. The non-transitory computer-readable medium of claim 12 , wherein the second portion penalizes the parameter inversely proportional to at least one of a measured doppler value or a vertical field of view associated with the radar sensor.

17. The non-transitory computer-readable medium of claim 12 , wherein the operations further comprise:

determining a phase offset associated with the updated model; and

calibrating the radar sensor by adjusting an orientation associated with the radar sensor based at least in part on the phase offset.