Autonomous

Self-driving cars are being tested by their own instructor ‘Synthia’

Published: June 24, 2016

At the Computer Vision Center in Barcelona, scientists have developed a smarter way to develop self-driving cars to match the level of human drivers. This innovative technology is called, Synthiaa virtual city simulation that can train driving AIs to recognize and handle different types of obstacles and situations, even in rain or deep snow.

Since the last few years, NVIDIA has entered the game of driverless cars by trying to make them safe enough and replace human drivers. However, with all these features being implemented, it would still take a substantial amount of time till you put your feet on the dashboard and let your Model S drive you to work.

synthia-dataset-self-driving-cars-7

Synthia consists of photo-realistic frames rendered from a virtual city and comes with precise pixel-level semantic
annotations for 13 classes, i.e., sky, building, road, sidewalk, fence, vegetation, lane-marking, pole, car, traffic signs, pedestrians, cyclists and miscellaneous. This would not only help autonomous cars to operate effortlessly but also make them self-sufficient to roam anywhere without a driver.

To begin with, researchers started by creating a realistic simulation of a not only a city and its surroundings, complete with pedestrians, cyclists and poorly parked buses, but also a complex weather system that includes rain, snow, and seasons. the researchers started by creating a realistic simulation of a not only a city and its surroundings, complete with pedestrians, cyclists and poorly parked buses, but also a complex weather system that includes rain, snow, and seasons.

After evaluating Synthia’s performance, scientists have now released the data with a public license for non-commercial use to get feedback and further improve the platform. Complete document with images can be accessed here- The Computer Vision Foundation (PDF).

Source: The Computer Vision Foundation (PDF)

Tags

Related Articles