We seldom think about it at the time, but when we are driving a vehicle our senses are being bombarded with thousands of impressions every second, all of which influence the choices we make. Self-driving vehicles need to behave in the same way, reading and interpreting their surroundings to make the right decisions.
Marco Trincavelli holds a PhD in robotics and works as a developer at Scania. His job is to analyse and interpret the information captured by the sensors of vehicles. “I get data in from the sensors and try to go through it and create models that are robust and allow the computer to make the right decision.”
Positioning in underground mines
Scania has already developed functioning prototypes that can drive on unmarked unsealed roads and interact with other vehicles to avoid collisions. Positioning is still dependent on GPS, which doesn’t work in places such as underground mines.
“We are developing special positioning modules to allow for the vehicles to be positioned in mines without GPS,” says Trincavelli. “In order to be able to drive completely autonomously, you need positioning that’s accurate down to just two to three centimetres above the ground, and that’s not possible with GPS.”
Several technologies are available to give vehicles the ability to interpret their surroundings. Scania has chosen to focus on radar and cameras, setting itself apart from Google, for example. “They’re focusing on laser sensors,” says Trincavelli. “It’s not that we’re opposed to that kind of sensor, but we first want to see what can be achieved using the sensors that we have on vehicles already in production and then evolve to laser sensors when we feel it’s needed.”
Collaboration for autonomous vehicle technologies
It also comes down to being smart about the way the choice of technology is adapted to the application. “There may be some scenarios where a full sensor set-up isn’t necessary,” says Trincavelli. “If you can manage with the cheaper but more robust sensors that we have today, then that’s better than shoving in all of the sensors at once.”
The technology behind Scania’s autonomous vehicles is the result of collaborative work between Scania, Saab Technologies, and Autoliv. “Scania has developed the sensor functions and the decision-making capabilities,” says Trincavelli. “The computers and radar are commercial products that we buy in, and our stereo cameras are developed by Autoliv.”
As well as the hardware components, Saab Technologies has developed the overall system for controlling and monitoring all the vehicles in locations such as mines.
When assembling a complete system for autonomous vehicles, it’s crucial that all the different bits of the puzzle fit together properly. “Achieving integration between the modules is one of the most difficult things that we do,” says Trincavelli. “It’s like working with musicians in an orchestra; if one plays a bad note the whole thing sounds bad. All the different parts need to understand each other and work together.”