System and method for vehicle defect detection
A defect detection system includes one or more route optical sensors configured to generate image data depicting an underbody of a vehicle that is on a route. The defect detection system also includes a controller configured to input the image data generated by the one or more route optical sensors into a first machine learning algorithm that determines an area of interest in the image data. The area of interest contains equipment of interest. The controller is configured to input the image data that is within the area of interest into a second machine learning algorithm that detects a defect on the equipment of interest. The controller is configured to perform at least one responsive action based on a type of defect that is detected.
1 . A defect detection system comprising:
one or more route optical sensors disposed along a route and configured to generate image data depicting an underbody of a vehicle that is on the route; and
a controller comprising one or more processors and operably connected to the one or more route optical sensors, the controller configured to input the image data generated by the one or more route optical sensors into a first machine learning algorithm that determines an area of interest in the image data, the area of interest containing equipment of interest, the controller configured to input the image data that is within the area of interest into a second machine learning algorithm that detects a defect on the equipment of interest, the controller configured to perform at least one responsive action based on a type of defect that is detected;
wherein the second machine learning algorithm is configured to detect the defect on the equipment of interest as a fluid leak in response to a fluid of the vehicle interacting with a leak visibility enhancement feature, wherein the leak visibility enhancement feature is one of a channel provided along an exterior surface of the equipment of interest or a paint or chemical configured to at least one of react with the fluid or enhance a color contrast between the fluid and the equipment of interest surrounding the fluid.
2 . The defect detection system of claim 1 , wherein the first machine learning algorithm is configured to determine the area of interest in the image data that contains a traction motor gear case as the equipment of interest.
3 . The defect detection system of claim 2 , wherein the second machine learning algorithm is configured to detect at least one of a fluid leak on the traction motor gear case or a missing locking element on a valve of the traction motor gear case as the defect.
4 . The defect detection system of claim 1 , wherein the controller is configured to generate cropped image data that contains the image data within the area of interest without including the image data that is outside of the area of interest, and the controller inputs only the cropped image data into the second machine learning algorithm.
5 . The defect detection system of claim 1 , wherein the controller is configured to generate a notification message as one action of the at least one responsive action, the notification message identifying the type of the defect that is detected, the controller further configured to control a communication device to communicate the notification message to at least one of the vehicle or a remote control system.
6 . The defect detection system of claim 1 , wherein the one or more route optical sensors are secured along a surface of the route at a middle section of the route, so that the vehicle that is on the route passes directly above the one or more route optical sensors as the vehicle moves along the route.
7 . The defect detection system of claim 1 , wherein the second machine learning algorithm is configured to determine a severity of the defect that is detected, and the controller is configured to perform the at least one responsive action based on both the type of the defect and the severity of the defect.
8 . The defect detection system of claim 1 , wherein at least one of the first machine learning algorithm or the second machine learning algorithm is an artificial neural network.
9 . The defect detection system of claim 1 , wherein the controller is configured to perform the at least one responsive action by one or more of scheduling maintenance for the vehicle, disabling the equipment that has the defect, or reducing a duty cycle of the equipment that has the defect.
10 . The defect detection system of claim 1 , wherein the leak visibility enhancement feature provided on the underbody of the vehicle and depicted in the image data that is analyzed by the second machine learning algorithm.
11 . The defect detection system of claim 1 , further comprising a wayside optical sensor configured to generate second image data depicting an exterior surface of the vehicle disposed on the route, the controller configured to analyze the second image data generated by the wayside optical sensor to detect one or more identifiers that are displayed on the vehicle and identify the vehicle based on the one or more identifiers, the controller further configured to associate the defect that is detected with an identity of the vehicle.
12 . A method comprising:
obtaining image data depicting an underbody of a vehicle on a route, the image data generated by one or more route optical sensors disposed along the route;
inputting the image data into a first machine learning algorithm that determines an area of interest in the image data, the area of interest containing equipment of interest;
inputting the image data that is within the area of interest into a second machine learning algorithm that detects a defect on the equipment of interest; and
performing at least one responsive action based on a type of defect that is detected,
wherein the second machine learning algorithm is configured to detect the defect on the equipment of interest as a fluid leak in response to a fluid of the vehicle interacting with a leak visibility enhancement feature, wherein the leak visibility enhancement feature is one of a channel provided along an exterior surface of the equipment of interest or a paint or chemical configured to at least one of react with the fluid or enhance a color contrast between the fluid and the equipment of interest surrounding the fluid.
13 . The method of claim 12 , further comprising generating a cropped image that includes only the image data within the area of interest, wherein inputting the image data into the second machine learning algorithm comprises inputting the cropped image.
14 . The method of claim 12 , wherein performing the at least one responsive action includes generating a notification message that identifies the type of the defect that is detected, the method further comprising controlling a communication device to communicate the notification message to at least one of the vehicle or a remote control system.
15 . The method of claim 12 , further comprising generating the image data via one or more route optical sensors secured along a surface of the route at a middle section of the route, the one or more route optical sensors having a field of view that is directed upward towards the underbody of the vehicle that is on the route.
16 . The method of claim 12 , further comprising determining a severity of the defect that is detected, and the at least one responsive action is performed based on both the type of the defect and the severity of the defect.
17 . The method of claim 12 , wherein performing the at least one responsive action comprises one or more of scheduling maintenance for the vehicle, disabling the equipment that has the defect, or reducing a duty cycle of the equipment that has the defect.
18 . The method of claim 12 , further comprising analyzing second image data generated by a wayside optical sensor and depicting an exterior surface of the vehicle that is on the route to detect one or more identifiers displayed on the vehicle;
identifying the vehicle that is on the route based on the one or more identifiers; and
generating a notification message as one action of the at least one responsive action, the notification message generated to associate the defect that is detected with an identity of the vehicle.
19 . A defect detection system comprising:
one or more route optical sensors disposed along a route and configured to generate image data depicting an underbody of a vehicle that is on the route; and
a controller comprising one or more processors and operably connected to the one or more route optical sensors, the controller configured to input the image data generated by the one or more route optical sensors into a first machine learning algorithm that determines an area of interest in the image data, the area of interest containing a traction motor gear case of the vehicle,
the controller configured to input the image data that is within the area of interest into a second machine learning algorithm that detects a defect on the traction motor gear case, the defect including at least one of a fluid leak on the traction motor gear case or an absent locking element on a valve of the traction motor gear case in response to a fluid of the vehicle interacting with a leak visibility enhancement feature, wherein the leak visibility enhancement feature is one of a channel provided along an exterior surface of the traction motor gear case or a paint or chemical configured to at least one of react with the fluid or enhance a color contrast between the fluid and the traction motor gear case surrounding the fluid, the second machine learning algorithm further configured to determine a severity of the defect, and
the controller configured to perform at least one responsive action based on both a type of defect that is detected and the severity of the defect.