award-en-03

How do autonomous vehicles deal with harsh weather?

start

Automated vehicles

Autonomous driving is already a reality, and holds the promise of safer roads and enhanced mobility.

We have created this resource to provide you with information about automated vehicles, their architecture and challenges of autonomous logistics.

This course has been developed within the european project AWARD (All Weather Autonomous Real logistics operations and Demonstrations), bringing together industrials and researchers all over Europe.

info

Next

Previous

How do autonomous vehicles deal with harsh weather?

Adapting the software

Guaranteeing safety

Discover the challenges of data processing

Explore the challenges of safe autonomous driving

Adapting the hardware

Explore what is done do adapt the sensors mechanically

Take your time and feel free navigating the contents.There is no particular order in discovering the different units, and you can skip those parts that you are already familiar with.

If you want to delve into a topic, just click on the "+ info" button, and explore the additional ressources.

The challenges

Discover why harsh weather conditions are a big challenge

Next

Previous

The challenges

Navigating in flood, rain, snow or fog, can be quite a challenge even for experienced drivers.The blurred vision of the environment makes it sometimes difficult to recognize shapes and potential danger, and the vehicle control is more delicate. The same applies for automated vehicles!Click on the different icons to discover why these weather conditions are challenging for AVs.

info

video

🌨️​

🌫️​

⛈️

Next

Previous

Adapting the software

Guaranteeing safety

Discover the challenges of data processing

Adapting the hardware

Explore what is done do adapt the sensors mechanically

Explore the challenges of safe autonomous driving

The challenges

Discover why harsh weather conditions are a big challenge

skip to the end

How do autonomous vehicles deal with harsh weather?

Feel free to explore this part again!

Next

Previous

Adapting the hardware

The presence of water or snowflakes in the air seriously impacts the functionning of the sensors, and in particular the lidars and video cameras.Here is an autonomous Tow Tractor that has been tested in an airport in Oslo. You will discover more about it in Episode 6.Click on the picture to zoom in and examine the different devices you can see on the vehicle.

info

info

Next

Previous

Adapting the software

Guaranteeing safety

Discover the challenges of data processing

Adapting the hardware

Explore the challenges of safe autonomous driving

Explore what is done do adapt the sensors mechanically

The challenges

Discover why harsh weather conditions are a big challenge

skip to the end

How do autonomous vehicles deal with harsh weather?

Feel free to explore this part again!

Next

Previous

Adapting the software

When driving under rain, a human driver is able not focusing on the small droplets on the windshield and concentrate on the road ahead.The biggest challenge of autonomous driving in harsh weather conditions lies here : how to filter the information and not misunderstand the signals coming from the sensors ?

video

info

Click on the icons to discover the main focus of current research and development on the subject.

Next

Previous

Adapting the software

Guaranteeing safety

Discover the challenges of data processing

Adapting the hardware

Explore what is done do adapt the sensors mechanically

Explore the challenges of safe autonomous driving

The challenges

Discover why harsh weather conditions are a big challenge

skip to the end

How do autonomous vehicles deal with harsh weather?

Feel free to explore this part again!

Next

Previous

Guaranteeing safety

The challenges of autonomous driving in harsh weather conditions still have to be resolved, in order to be able to guarantee safety of the different road users at all times.The different enhancements of the sensors have to be tested in real-life conditions, as you will discover in Episodes 4 to 7. Beforehand, they are tested in strongly controlled environments. These tests are ran on specific platforms, able to simulate dense fog and heavy rain. Discover an example of such platform on the next page!

info

info

Next

Previous

contact

Discover PAVIN - Simulation platform for adverse climate conditions in France

Next

Previous

Adapting the software

Guaranteeing safety

Discover the challenges of data processing

Explore the challenges of safe autonomous driving

Adapting the hardware

Explore what is done do adapt the sensors mechanically

The challenges

Discover why harsh weather conditions are a big challenge

skip to the end

How do autonomous vehicles deal with harsh weather?

Feel free to explore this part again!

Next

Previous

go to the course page

discover AWARD

go to the next episode

go to the episodes review

You have completed this episode!

What do you want to do next ?

Previous

Improving the signal

In order to enable the sensors to get better at "ignoring" the noise they may perceive in snowy, rainy, foggy or dusty conditions, a great effort is made today to enhance the perception of the sensors.Some of the solutions focus on amplifying the received backchannel signal (for instance, when the laser "hits back" the lidar).Other researchers focus on the development of "filters". For example, algorithms are developed to "de-rain" the signal for the video cameras, to increase the performance of the lidar in foggy conditions, or to allow autonomous vehicles recognizing the path when the road is covered in snow.

Rain

Heavy rain has a strong impact on the lidar system and the video cameras.The raindrops reflect the laser signal: as a result, the lidar has a reduced recognition of the distance and the perceived shapes are distorted. The video cameras have a degraded signal as well, which makes the recognition less reliable.

Data fusion

The signals collected by the vehicle are of very diverse natures: images, light waves, radio waves, GPS signals... In order to help the built-in computer to process the information, algorithms are developed to fuse all the information, just like a human brain would do.This is a big challenge for autonomous driving, and it particuarly stands out in adverse weather conditions. As a matter of fact, the signal is impaired and the "filters" just add more complexity.

Other solutions have been explored in the AWARD project, in particular the infrared cameras. The tests have shown the benefits of these devices in harsh weather conditions.Find out more in these scientific papers :

info

info

Fog

The presence of small water droplets, that constitute fog, has great impact on the lidar system. As a matter of fact, it impacts the laser trajectory and distorts the received signal. The lidar can, thus, "see" an obstacle, even though there is none, and has trouble detecting the distances.The degraded visibility also impacts video cameras, making it difficult to recognize the surroundings correctly.

The different solutions have to stand the test of simulation, in order to be validated.Find out more about the simulation in this scientific paper :

info

In order to be able to "correctly see" the environment, the different sensors have to be protected. The goal is to prevent water, snow, dust or mud covering the cameras, lidars and radars.A first solution is, thus, to install protections over and beneath the sensors. These protections are similar to mudguards and visors. Other solutions consist in applying a heating cover over the lidars.You can zoom in and look closer at the different sensors placed on the vehicle to see how they are protected.

Snow

In a snowfall or with snow on the road, the lidar system and the video cameras can struggle quite a lot.Similarly to the raindrops, snowflakes deviate the trajectory of the laser beam. Furthermore, a snowfall can be perceived as strong noise, because of the falling particles. The lidar has, thus, trouble interpreting correctly the received signals.The video cameras struggle recognizing road markings, if they are covered by snow. This makes it difficult for the autonomous vehicle to orient itself in its environment.