Exploring the Intersection of Computer Vision and Transportation

From self-driving cars to self-landing planes and beyond

Computer Vision and Transportation 

Computer vision is like human vision, except it’s through the eyes of AI. In the world of AI, computer vision can quickly process images and identify what it’s staring at, except with higher accuracy, because eyes don’t play tricks on machines. In the world of transportation that translates into allowing cars to be self-driving because they are capable of identifying if there’s a car, human, road sign, or something else in their path. Due to the advancements in computer vision, we see success in advanced car features, flight assistance in the aerospace industry, and for robots used in outer space.  

Computer Vision X Vehicles 

When it comes to computer vision in vehicles the most commonly thought of feature is a fully self-driving car, but it actually plays a significant role in several other aspects including: 

  • Autonomous driving
  • Parking assistance 
  • Lane assistance 
  • Distracted driver 
  • Blind spot monitoring 

While only a select few cars have full autonomous driving support, features like parking assistance are becoming standard in newer car models across the industry. One of the newer technologies is parallel parking assistance. Parallel parking assistance with computer vision utilizes cameras and algorithms to aid drivers in parking their vehicles in tight spaces. The computer vision system evaluates the images captured by the cameras to identify the position of the car in relation to other objects, estimate the size of the parking space, and determine the best trajectory for the car to follow during the parking process. The system then provides visual and auditory guidance to the driver, such as displaying lines or emitting beeps to indicate the distance between the car and other objects. Some systems can even take over control of the steering and brake systems to park the car automatically. This technology makes parking easier and safer. 

According to Forbes, computer vision can help create smart transportation systems in its entirety through accident prevention and traffic management. In their article How Computer Vision Can Create Smart Transportation Systems they talk about 3 attributes every smart city must have 

  • Livability 
  • Workability 
  • Sustainability 

The sustainability part is where computer vision comes into play. For accident prevention, smart sensors and smart cameras located in busy places and major roads can send updates to drivers to let them know if they should take an alternative route and or adjust their driving to prevent a potential accident.  

The same logic applies for traffic management, smart sensors can indicate when alternative routes are needed because one area is backed up with traffic. There is also a second prominent use case and that’s for emergency service vehicles so they can find the quickest path to their destination and provide life saving services in a timely manner.  

Appen X Airbus 

In essence, airplanes are just flying cars, so logically they can benefit from similar computer vision technologies like vehicles on the road do. We recently worked with Airbus to help annotate their flight paths and update their AirSense program to increase the accuracy of estimated flight and arrival times for commercial flights. Our skilled annotators were able to train a computer vision model to identify the start and end points a trip, so the average travel time could be accurately counted.  

Other advances in the aerospace industry are getting an upgrade with computer vision. Planes are using Instrument Landing System (ILS) to help pilots land planes without human assistance. Yet fewer than 1,000 airports have ILS, which means other technology needs to be implemented for self-landing planes to exist.  

Airbus and Wayfinder have partnered up to find a solution for this very scenario. They are using smarts sensors, actuators, and computer vision to train planes to land safely on the runway through a process called deep learning. To expediate the process of gathering enough data to train the self-landing model correctly, synthetic data was generated from X-Plane. While research and model training are still being done, soon the phrase “does anyone know how to land this plane” will just be a thing in the movies. 

Computer Vision Explores the Great Unknown 

Space, the final frontier to be discovered. With the help of computer vision and AI, it is making the undiscoverable, discoverable. DLR a German space agency developed CIMON a robot that was sent into space to help assists astronauts with various tasks.  

Of course, getting to another planet in itself is no easy feat. SpaceX has created an AI powered auto-pilot system to help astronauts safely reach their destination and use as little of their resources as possible. This was first seen on the SpaceX Falcon 9 rocket. 

As mentioned in our Smart Solutions for a Greener Futuer blog, AI can help heal the environment, the same is true for space. There are currently an estimated 330 million pieces of space debris, including 36,500 objects bigger than 10cm, such as old satellites, spent rocket bodies and even tools dropped by astronauts orbiting Earth. A Swiss company called ClearSapce is using AI powered cameras to help pick up all the debris in space to make travel easier and safer.  

The Future is Computer Vision 

Computer vision is the vision of the future. Thanks in part to smart sensors and countless hours of annotated data being poured into machine learning models, our roads, skies, and outer space are safer places to be. From self-driving cars to self-flying spaceships, the risk of accidents from human error is significantly reduced. People and machines can get to their destinations safely, and in a reasonable amount of time too, thanks to AI mapping out the best routes to take. Computer vision will continue to advance in the coming years and even more automation will make the world a bit easier and safer to navigate.  

Website for deploying AI with world class training data