Object identification with thermal imaging
A plurality of thermal images forward of a vehicle are collected. Thermal data in the plurality of thermal images is normalized based on an ambient air temperature to generate a plurality of normalized thermal images. The plurality of normalized thermal images are input to a machine learning program trained to output an identification of an object based on the ambient air temperature and a risk of collision between the vehicle and the object. A vehicle component is actuated based on the identification of the object and the risk of collision with the object.
1. A system, comprising a computer including a processor and a memory, the memory storing instructions executable by the processor to:
collect a plurality of thermal images forward of a vehicle;
normalize thermal data in the plurality of thermal images based on an ambient air temperature to generate a plurality of normalized thermal images;
divide each of the plurality of normalized thermal images into a plurality of subregions that include respective pluralities of pixels, each subregion having a respective infrared intensity threshold based on thermal values of the pixels in the subregion;
assign one or more of the pixels in a respective subregion to a region of interest based on the thermal values of the pixels in the respective subregion and the infrared intensity threshold for the respective subregion;
input the plurality of normalized thermal images to a machine learning program trained to output an identification of an object in the region of interest based on the ambient air temperature and a risk of collision between the vehicle and the object; and
actuate a vehicle component based on the identification of the object and the risk of collision with the object.
2. The system of claim 1 , wherein the object is an animal, and the risk of collision is based on predicted trajectory of the animal, the predicted trajectory of the animal based on a difference of thermal data between a specified set of pixels in a first normalized thermal image and a second specified set of pixels in a second normalized thermal image.
3. The system of claim 1 , wherein instructions further include instructions to determine a distance to the object and a predicted trajectory of the object based on the normalized thermal images and to input the distance and the predicted trajectory to the machine learning program, the machine learning program being trained to output the risk of collision with the object based on the distance and the predicted trajectory.
4. The system of claim 3 , wherein the instructions further include instructions to predict a trajectory of the vehicle and to input the predicted trajectory of the vehicle to the machine learning program, the machine learning program being trained to output the risk of collision with the object based on the predicted trajectory of the object and the predicted trajectory of the vehicle.
5. The system of claim 1 , wherein the instructions further include instructions to identify the object based on the thermal data in the plurality of normalized thermal images indicating a temperature exceeding the ambient air temperature.
6. The system of claim 1 , wherein the instructions further include instructions to identify a region of interest in each normalized thermal image based on one or more previously determined object properties.
7. The system of claim 1 , wherein the instructions further include instructions to actuate a brake based on the risk of collision with the object.
8. The system of claim 1 , wherein the instructions further include instructions to provide an output to a display screen based on the risk of collision with the object.
9. The system of claim 1 , wherein the instructions further include instructions to generate a bounding box enclosing the object.
10. The system of claim 1 , wherein the instructions further include instructions to identify a current location of the vehicle and to output the identification of the object based on the current location of the vehicle.
11. The system of claim 10 , wherein the instructions further include instructions to identify the current location of the vehicle as one of an urban environment or a rural environment, to output the identification of the object as a pedestrian when the current location is the urban environment, and to output the identification of the object as an animal when the current location is a rural environment.
12. A method, comprising:
collecting a plurality of thermal images forward of a vehicle;
normalize thermal data in the plurality of thermal images based on an ambient air temperature to generate a plurality of normalized thermal images;
dividing each of the plurality of normalized thermal images into a plurality of subregions that each include respective pluralities of pixels, each subregion having a respective infrared intensity threshold based on thermal values of the pixels in the subregion;
assigning one or more of the pixels in a respective subregion to a region of interest based on the thermal values of the pixels in the respective subregion and the infrared intensity threshold for the respective subregion;
inputting the plurality of normalized thermal images to a machine learning program trained to output an identification of an object in the region of interest based on the ambient air temperature and a risk of collision between the vehicle and the object; and
actuating a vehicle component based on the identification of the object and the risk of collision with the object.
13. The method of claim 12 , wherein the object is an animal, and the risk of collision is based on predicted trajectory of the animal, the predicted trajectory of the animal based on a difference of thermal data between a specified set of pixels in a first normalized thermal image and a second specified set of pixels in a second normalized thermal image.
14. The method of claim 12 , further comprising determining a distance to the object and a predicted trajectory of the object based on the normalized thermal images and inputting the distance and the predicted trajectory to the machine learning program, the machine learning program being trained to output the risk of collision with the object based on the distance and the predicted trajectory.
15. The method of claim 13 , further comprising predicting a trajectory of the vehicle and inputting the predicted trajectory of the vehicle to the machine learning program, the machine learning program being trained to output the risk of collision with the object based on the predicted trajectory of the object and the predicted trajectory of the vehicle.
16. The method of claim 12 , further comprising identifying the object based on the thermal data in the plurality of normalized thermal images indicating a temperature exceeding the ambient air temperature.
17. The method of claim 12 , further comprising actuating a brake based on the risk of collision with the object.
18. The method of claim 12 , further comprising identifying a current location of the vehicle and outputting the identification of the object based on the current location of the vehicle.