Neuromorphic foreign object detection
A neuromorphic foreign object debris (FOD) detection system includes a FOD processing system and a neuromorphic sensor. The FOD processing system includes a FOD controller including a trained artificial intelligence machine learning (AIML) model representing an area of interest. The neuromorphic sensor has a field of view (FOV) containing the area of interest and is configured to output pixel data in response to FOD appearing in the FOV. The FOD controller detects the FOD is present in the area of interest in response to receiving the pixel data, and generates an alert signal indicating the presence of the FOD.
1. A neuromorphic foreign object debris (FOD) detection system comprising:
a FOD processing system including a FOD controller including a trained artificial intelligence machine learning (AIML) model representing an area of interest; and
a neuromorphic sensor in signal communication with the FOD processing system, the neuromorphic sensor having a field of view (FOV) containing the area of interest and configured to output pixel data in response to FOD appearing in the FOV,
wherein the FOD controller detects the FOD is present in the area of interest in response to receiving the pixel data, and generates an alert signal indicating the presence of the FOD.
2. The FOD detection system of claim 1 , wherein the AIML model is implemented as a deep convolutional neural network (DCNN) model.
3. The FOD detection system of claim 1 , wherein the AIML model is implemented as a deep convolutional auto encoder (CAE) model.
4. The FOD detection system of claim 1 , wherein the neuromorphic sensor includes an event camera including a plurality of pixels, each pixel configured to realize a local change in brightness and to output local pixel data indicating the local change in brightness as it occurs, and wherein the pixel data indicates corresponding to at least one localize pixel among the plurality of pixels that realizes the local change in brightness while excluding pixel data corresponding to remaining pixels that do not realize the local change in brightness.
5. The FOD detection system of claim 1 , wherein the area of interest includes an aircraft engine.
6. The FOD detection system of claim 5 , wherein the neuromorphic sensor is located remotely from the aircraft engine and the FOV contains the aircraft engine.
7. The FOD detection system of claim 5 , wherein the neuromorphic sensor is coupled to an aircraft engine and the FOV contains the aircraft engine.
8. The FOD detection system of claim 5 , wherein the neuromorphic sensor is disposed inside the aircraft engine.
9. A method of performing neuromorphic foreign object debris (FOD) detection comprises:
generating training data;
training an artificial intelligence machine learning (AIML) model using the training data depicting an area of interest;
monitoring the area of interest using a neuromorphic sensor;
outputting pixel data from the neuromorphic sensor in response to FOD appearing in the area of interest; and
generating an alert indicating a presence of FOD in the area of interest in response based on the pixel data.
10. The method of claim 9 , wherein the AIML model is implemented as a deep convolutional neural network (DCNN) model.
11. The method of claim 9 , wherein the AIML model is implemented as a deep convolutional auto encoder (CAE) model.
12. The method of claim 9 , wherein the area of interest includes an aircraft engine.
13. The method of claim 12 , wherein the neuromorphic sensor is located remotely from the aircraft engine and the FOV contains the aircraft engine.
14. The method of claim 12 , wherein the neuromorphic sensor is coupled to an aircraft engine and the FOV contains the aircraft engine.
15. The method of claim 12 , wherein the neuromorphic sensor is disposed inside the aircraft engine.
16. The FOD detection system of claim 4 , wherein the FOD controller detects the FOD is present in the area of interest in response to receiving the output local pixel data from at least one pixel among the plurality of pixels.
17. The method of claim 9 , further comprising detecting, by at least one local pixel among a plurality of pixels included in the neuromorphic sensor, a local change in brightness independent from a change in brightness among remaining pixels included in the plurality of pixels;
analyzing, by a FOC controller, the the pixel data which indicates the local change in brightness of the at least one local pixel corresponding to at least one local pixel among the plurality of pixels while excluding pixel data corresponding to the remaining pixels that do not realize the local change in brightness; and
determining the presence of FOD in the area of interest based on the local change in brightness of the at least one local pixel indicated by the pixel data.