Vehicle detection of missing body parts

1. A computer, comprising a processor and a memory, the memory storing instructions executable by the processor such that the computer is programmed to:

input to a trained machine learning program, (1) a sensor fusion error that measures a statistical correlation of data received from a radar sensor and a second sensor in a vehicle, (2) a radar detection range, (3) an amount of reflection from a radar radome, (4) weather data, and (5) aerodynamic data that measures an aerodynamic drag opposing a vehicle motion; and

output from the trained machine learning program a determination concerning a presence of the radar radome.

2. The computer of claim 1 , wherein the second sensor includes a camera sensor or lidar sensor, and the instructions further include instructions to:

receive data from the second sensor; and

apply the trained machine learning program to the received second sensor data, wherein the machine learning program is further trained to output the determination concerning the presence of the radar radome based on the received second sensor data.

3. The computer of claim 1 , wherein the instructions further include instructions to:

receive vehicle collision data specifying whether a collision occurred; and

apply the machine learning program to the received vehicle collision data, wherein the machine learning program is further trained to output the determination concerning the presence of the radar radome based on the received vehicle collision data.

4. The computer of claim 1 , wherein the instructions further include instructions to:

upon determining a lack of the radar radome, store a diagnostic code in a computer memory;

adjust one or more calibration parameters of the radar sensor;

determine an updated sensor fusion error based at least in part on the adjusted one or more calibration parameters; and

stop adjusting the one or more calibration parameters upon determining that the updated sensor fusion error is less than a specified error threshold.

5. The computer of claim 4 , wherein the instructions further include instructions to increase at least one of an operating frequency or an operating power level of the radar sensor upon determining, based on the weather data, a presence of moisture on the radar sensor.

6. The computer of claim 4 , wherein the instructions further include instructions to adjust an operating parameter of a radar sensor blockage detection algorithm based on the weather data.

7. The computer of claim 1 , wherein the instructions further include instructions, upon determining that (i) based on stored diagnostic code the radar radome is missing and (ii) based on the machine learning program output, the radar radome is present, to output a message, to a user device, including an instruction to adjust one or more calibration parameters of the radar sensor.

8. The computer of claim 1 , wherein the instructions further include instructions to deactivate a vehicle autonomous operation upon determining that the radar radome is missing.

9. The computer of claim 1 , wherein the instructions further include instructions to determine the radar detection range based on a number of objects detected by each of the radar sensor and the second sensor, in a plurality of ranges from the vehicle.

10. The computer of claim 1 , wherein the instructions further include instructions to train the machine learning program by applying a set of training data including the inputs and an expected output including one of “the radar radome is missing” or “the radar radome is present” for each set of the training data.

11. The computer of claim 1 , wherein the instructions further include instructions to determine the weather data including data concerning snow, rain, or wind.

12. The computer of claim 1 , wherein the instructions further include instructions to determine the aerodynamic drag of the vehicle based on data including a vehicle speed, a vehicle fuel consumption, an engine torque, and a road slope.

13. A method, comprising:

inputting to a trained machine learning program, (1) a sensor fusion error that measures a statistical correlation of data received from a radar sensor and a second sensor in a vehicle, (2) a radar detection range, (3) an amount of reflection from a radar radome, (4) weather data, and (5) aerodynamic data that measures an aerodynamic drag opposing a vehicle motion; and

outputting from the trained machine learning program a determination concerning a presence of the radar radome.

14. The method of claim 13 , further comprising:

receiving data from the second sensor; and

applying the trained machine learning program to the received second sensor data, wherein the machine learning program is further trained to output the determination concerning the presence of the radar radome based on the received second sensor data.

15. The method of claim 13 , further comprising:

upon determining a lack of the radar radome, storing a diagnostic code in a computer memory;

adjusting one or more calibration parameters of the radar sensor;

determining an updated sensor fusion error based at least in part on the adjusted one or more calibration parameters; and

stopping adjusting the one or more calibration parameters upon determining that the updated sensor fusion error is less than a specified error threshold.

16. The method of claim 15 , further comprising increasing at least one of an operating frequency or an operating power level of the radar sensor upon determining, based on the weather data, a presence of moisture on the radar sensor.

17. The method of claim 15 , further comprising adjusting an operating parameter of a radar sensor blockage detection algorithm based on the weather data.

18. The method of claim 13 , further comprising, upon determining that (i) based on stored diagnostic code the radar radome is missing and (ii) based on the machine learning program output, the radar radome is present, outputting a message, to a user device, including an instruction to adjust one or more calibration parameters of the radar sensor.

19. The method of claim 13 , further comprising training the machine learning program by applying a set of training data including the inputs and an expected output including one of “the radar radome is missing” or “the radar radome is present” for each set of the training data.

20. The method of claim 13 , further comprising determining the aerodynamic drag of the vehicle based on data including a vehicle speed, a vehicle fuel consumption, an engine torque, and a road slope.