As part of the complete Jet Bridge Autonomy Package, the deep learning enhanced sensors guarantee docking in about 45 seconds, much less than the 2-3 minutes most airlines assign to the docking process. An additional benefit is that autonomous jet bridges are typically operating during thunderstorms where no staff is permitted on the bridge. This allows passengers to disappear if the aircraft could be stuck until the storm passes.
Human intrusion detection
Security at airports has always been a concern, with several areas on the land side, which could potentially allow for bad-intentional security threats. Alternatively, a curious child may slip through the security door/portal on a ticket counter barge conveyor or flat plate billing conveyor. Both events represent security violations and can lead to serious injuries to people unfamiliar with luggage handling conveyors due to equipment in the system.
The trick here is to allow different flows of checked luggage to move through portal openings, detecting when a person is about to slip through the security door/portal. In addition to posting security guards to watch a large portal, no real satisfactory solution has been implemented.
Currently, 3D sensors of the same color trained in neural networks with specific training to recognize humans and objects will perform tasks quickly, accurately and inexpensively. Neural network training included annotated images of people in a variety of colors, sizes and shapes. With parameterization, the sensor creates a field that is human-detected and enables various alarm outputs (such as local output or Ethernet messages). You can also parameterize neural networks to detect objects in your field of view. Therefore, the sensor can also be used to detect collisions in jet bridges and other types of ground service equipment (GSE).
Bag Classification/Hygiene Detection
Baggage Handling Systems (BHS) are designed to handle a very wide variety of items, but there are some items that can cause havoc and jam in the system, such as:
– Cylindrical duffel bags that tend to roll over the slope of the conveyor
– Handle causes jams around expansion/telescoped wheeled bags, conveyor turn or diverter systems
– Children's car seats are misplaced on totes rather than totes, but these are often clogged with the entrance to explosive detection systems (EDS) equipment
– Bag with long exposed carrying straps packed with conveyor transition
– Children's strollers are not placed in totes, making them difficult to transport or reuse due to their extremely irregular shape
– Large cubic box that can be jamed with explosive detection system (EDS) equipment
– Backpack or knapsack that can cause a tilt to fall or expose the strap
You can detect these problematic bags at the ticket counter or significantly improve the system's throughput and reliability before going too far in BHS.
To solve this application, hundreds of thousands of images were collected at international airports and divided into over 30 different categories. In this process, two-color 3D cameras collected images from different perspectives, allowing not only collection of images of colored bags, but also the size and shape of the bag. Annotated images were then fed to a deep learning platform to train advanced neural networks. After the first round of training, neural networks were deployed to the BHS system where solutions were tested and evaluated in live bags. The detection error was removed and the new neural network was trained again. This process was repeated until the results were satisfactory. Several systems have already been live operations at international airports.
