How Do Self-Driving Vehicles See the World and Orientate? or How Do Self-Driving Vehicles Work? 8 How Do Self-Driving Vehicles See the World and Orientate? or How Do Self-Driving Vehicles Work? 9 How Do Self-Driving Vehicles See the World and Orientate? or How Do Self-Driving Vehicles Work? 10

The World Through the Eyes of Self-Driving Vehicles

#Neural Networks #Machine Learning

For someone, a self-driven vehicle or self-flying drone moving without human intervention is magic, but we at Evergreen know what technologies and software are needed for his correct operation, orientation in space and safe traffic on the road.

How Do Self-Driving Vehicles See the World Around Them?

A self-driven vehicle or a self-flying drone sees the world around using special sensors that it is equipped with. Thanks to them, it recognizes objects on the way, determines the distance to them, moves smoothly and without hitches, understands his location on the route.

Sensors a Self-Driven Vehicle Is Equipped with

  • Computer stereo vision іs the extraction of 3D data from digital images. 3D information can be extracted by examining the relative positions of objects by comparing information from two simultaneously working cameras.
  • Gyro-Stabilizer is smooth and operates seamlessly.  Gyros provide extremely smooth stabilization: soft movement of the machine, reduction of a roll mode.
  • LIDAR (Light Detection and Ranging) is the primary device for the self-driven vehicle’s orientation. Lidar is a laser scanner operating on the radar principle. As a device, lidar is an optical rangefinder that uses the phenomenon of light reflection and its dispersion in transparent and translucent environments, processes information about distant objects using active optical systems. The coherent laser radiation rays (some models are equipped with more than 60  lasers) form 3D or 2D picture of the world.
  • Global Positioning System (GPS, Glospace) A self-driving vehicle determines its location on the route with the help of GPS. Іt is a radio system that uses information from navigation satellites that are needed for drawing up a route, calculating the time required for the journey, determining the three-dimensional position of the vehicle.

All this data, as well as SLAM, simultaneous localization, mapping, are used by the car in autopilot mode for orientation in space, object recognition around itself and to avoid accidents.

As for the self-flying drone, in addition to the standard equipment that makes the flight possible, the developers use an autopilot module with tilt sensors, an accelerometer, and an electronic compass for orientation in space.

Neural Networks to Identify Objects from Evergreen

Modern self-driven vehicles and self-flying drones are already able to navigate well in space, avoid accidents and exchange information. However, there are plenty of tasks that can and should be taught for greater efficiency.

At Evergreen, we are developing neural networks and solving Computer Vision (CV) tasks suitable for robots and applicable for autonomous and self-driven vehicles. In practice, this means that your self-driven vehicle will not only be able to orient in space and not to crash into anything but will be able to determine which object is in front of it during movement, scan its parameters and analyze the situation observed from the camera and carry out an assigned task. Likewise, robots-loaders are getting the job done in large warehouses.

Do you want to add intelligence to your robot, train it and add useful for your business skills or use computer vision machines to automate processes? Feel free to contact us. We love to solve non-trivial tasks.

The images used in this article are taken from open sources and are used as illustrations.
Do you want to discuss your project or order development?