AI News, Drones learn to navigate autonomously by imitating cars and bicycles

Drones learn to navigate autonomously by imitating cars and bicycles

Integrate autonomously navigating drones Researchers of the University of Zurich and the National Centre of Competence in Research NCCR Robotics developed DroNet, an algorithm that can safely drive a drone through the streets of a city.

Designed as a fast 8-layers residual network, it produces two outputs for each single input image: a steering angle to keep the drone navigating while avoiding obstacles, and a collision probability to let the drone recognise dangerous situations and promptly react to them.

Powerful artificial intelligence algorithm Instead of relying on sophisticated sensors, the drone developed by Swiss researchers uses a normal camera like that of every smartphone, and a very powerful artificial intelligence algorithm to interpret the scene it observes and react accordingly.

'This is a computer algorithm that learns to solve complex tasks from a set of 'training examples' that show the drone how to do certain things and cope with some difficult situations, much like children learn from their parents or teachers,' says Prof.

Autonomous high flying drones learn to navigate by watching traffic below

In countries where commercial drone delivery is permitted beyond line-of-sight (hint: not the US), autonomous drones still have a big blind spot: Urban areas.

The use of cameras instead of higher-cost sensors was intentional, a way of ensuring that the algorithm can be used on a wide variety of platforms in production today.

The smarts of the system come from the powerful artificial intelligence algorithm, which uses deep learning to interpret images and direct drones to react.

'This is a computer algorithm that learns to solve complex tasks from a set of 'training examples' that show the drone how to do certain things and cope with some difficult situations, much like children learn from their parents or teachers,' says Scaramuzza.

The DroNet algorithm teaches drones to navigate city streets like cars

By observing and learning how cars and bicycles react to the dynamic environment of a city street, the DroNet algorithm lets the drones recognize static and moving obstacles, triggering it to slow down and avoid crashes.

“Instead of relying on sophisticated sensors, DroNet only requires a single camera — very much like that of every smartphone — on a drone.” Most of today’s drones use GPS to navigate, which is great if they’re traveling above buildings but complicated if they are flying at low altitudes in densely populated streets.

So, in order to teach the drone to navigate city streets safely, Scaramuzza and his team collected data from cars and bicycles in urban settings, and fed that data into the DroNet algorithm, which used the data to learn street etiquette —

Drones Learn to Navigate Autonomously By Imitating Cars and Bicycles

Researchers at the University of Zurich (UZH) and the National Center of Competence in Research Robotics in Switzerland have developed the DroNet algorithm, which enables drones to fly by themselves through the streets of a city and in indoor environments.

DroNet generates two outputs for each single camera input image, including a steering angle to keep the drone navigating while evading obstacles, and a collision probability so the drone can identify and promptly respond to dangerous situations.

DroNet Algorithm Learns From Traffic to Navigate City Streets

Street-level drone operations in urban centers are far from becoming a reality.

The team has harnessed machine learning to develop DroNet, a convolutional neural network “that can safely drive a drone through the streets of a city.”

“We have developed an algorithm that can safely drive a drone through the streets of a city and react promptly to unforeseen obstacles, such as other vehicles and pedestrians,” said Davide Scaramuzza, head of the University of Zurich’s Robotics and Perception Group.

But given the amount of data required and the risk it could pose to people and road users, that’s not really an option.

So far the system has been able to drive drones autonomously while following learned rules of the road, such as staying in your lane, avoiding pedestrians and other basic things we take for granted.

Quadcopter Navigation in the Forest using Deep Neural Networks

We study the problem of perceiving forest or mountain trails from a single monocular image acquired from the viewpoint of a robot traveling on the trail itself.

Watch This AI Drone Teach Itself How To Fly Through Trial And Error

Researchers are crashing drones on purpose in the name of artificial intelligence. Check it out: SUBSCRIBE FOR THE LATEST VIDEOS ...

Atlas, The Next Generation

A new version of Atlas, designed to operate outdoors and inside buildings. It is specialized for mobile manipulation. It is electrically powered and hydraulically ...

Learning Models of Language, Action and Perception for Human-Robot Collaboration

Robots can act as a force multiplier for people, whether a robot assisting an astronaut with a repair on the International Space station, a UAV taking flight over ...

Nature Inspired Robot Swarm Cooperative Retrieval

Dr Neil Vaughan - Nature Inspired Robot Swarm Cooperative Retrieval - Alife 2016. Abstract: In nature, bees and leaf-cutter ants communicate to improve ...

MIT 6.S094: Introduction to Deep Learning and Self-Driving Cars

This is lecture 1 of course 6.S094: Deep Learning for Self-Driving Cars taught in Winter 2017. Course website: Lecture 1 slides: ..

Mapping the Planet with Computer Vision

Learn from our Inception program member, Jan Erik Solem, CEO and Co-Founder of Mapillary, on how they are using computer vision to map the entire planet ...

2018 Isaac Asimov Memorial Debate: Artificial Intelligence

Isaac Asimov's famous Three Laws of Robotics might be seen as early safeguards for our reliance on artificial intelligence, but as Alexa guides our homes and ...

A Compilation of Robots Falling Down at the DARPA Robotics Challenge

Lots of robots fell over at the DARPA Robotics Challenge Finals. Learn more: ...

Robots: Crash Course Computer Science #37

Today we're going to talk about robots! Robots are often thought as a technology of the future, but they're already here by the millions in the workplace, our ...