En route to safer, more reliable autonomous driving
The development of autonomous driving systems is currently a focus of research for the automotive industry. An EU-funded project has moved work forward in this area by developing an advanced driver-assistance system that can work safely and reliably in all weathers.
© jan_S, #186469727, source:stock.adobe.com 2020
Current driver-assistance systems work well in good conditions. However, in heavy rain, snow or fog, the sensors in these systems do not provide enough information for safe driving.
As the world moves gradually towards fully autonomous driving systems where the car is in complete control it is essential that the sensors and related technologies deliver reliable information and decision-making that can cope with different conditions, as well as the erratic behaviour of other road users.
The EU-funded ROBUSTSENSE project has successfully tackled these issues by developing an advanced driver-assistance platform. The project team, which drew in 15 partners from five European countries, provided a range of expertise in sensors and data processing.
Our platform is equipped with specialised technologies, including software algorithms specifically implemented to cope with adverse weather, and a newly developed LiDAR sensor for extreme conditions, explains Werner Ritter, ROBUSTSENSE project coordinator. Our modular system is based on layers that relate to data and information flow within an intelligent and robust sensor array that reacts to real-world situations. It manages diversity and complexity while dealing with uncertainties on the road.
Reading the road
A sensor layer constantly scans the environment to assess driving conditions and the state of the road. This data helps determine if vehicle speed needs adjusting. A fusion layer then combines the collected information in a way that allows the system to see the entire scene including weather conditions, the presence of pedestrians, and the number, size, and movement of other vehicles.
With the scene complete, an understanding and planning layer ensures the vehicle makes all the right moves. For example, the ROBUSTSENSE platform can deal effectively with other road users behaviour if the system is unsure, the vehicle will slow down in readiness to react before speeding up when the situation has been resolved.
The platform can also monitor its own performance and reliability by using a unique self-assessment system. If a sensor or camera is dirty or partially covered by snow, the system knows that this input is less reliable and makes the necessary adjustments.
The development of a LiDAR sensor with a greater range was another key breakthrough. LiDARs measure distance very accurately by using lasers. ROBUSTSENSE managed to increase the LiDAR wavelength to 1 550 nm (nanometres) from a standard maximum of 905 nm, giving the new system more time to make decisions especially in fog.
On the right track
The ROBUSTSENSE technologies have been successfully demonstrated in a number of different commercially available cars. The testing shows that our system has the ability to determine road surface conditions and can cope with non-compliant behaviour by other road users, Ritter adds. It can make autonomous driving adjustments and detect pedestrians in fog. The projects results could also find applications beyond the automotive sector. For example, the manufacture of LiDARs with an enhanced range could improve detection and measurement in areas such as land and marine mapping.
Meanwhile, the project software and networks for optical sensors could be of value in areas such as original equipment manufacturing as well as the development of ICT infrastructure and robotics.
ROBUSTSENSE received EU funding from the Electronic Components and Systems for European Leadership Joint Undertaking (ECSEL JU) worth 3 348 357€ as well as 3 404 968€ from national funding authorities in Germany, Austria, Italy, Spain and Finland.