The future of the autonomous vehicle core technology: Q&A with Hakan Sivencrona, DelphiAdd bookmark
Advanced -driver assistance technologies such as automate lighting and adaptive cruise control have been around for years. But Håkan Sivencrona, the Functional Safety Manager / Lead System Engineer at Delphi, isn’t looking behind him.
He’s looking at what technologies are making their way into the autonomous vehicle of tomorrow. He’s looking at new features as well as those currently on the docket that need more improvement.
Progress has been gradual; it hasn’t yet kicked into high gear. Gradual improvements have been made with each and every model since the first roll out of the camera systems that were able to detect high beams. Today’s technologies have evolved to adaptive cruise control and fully autonomous braking systems, he says. Several manufacturers are looking at traffic jam assist, which means that the system will keep track of the lanes and barriers in lower speeds.
“[It’s important to] understand what the driver really sees, how the driver will act, and what could really be a supportive functionality,”
This is an example of autonomous behavior even though it has not reached fully autonomous capability, he points out. Håkan believes that concept improvements will evolve with time and at with every model, adding that, “sometimes you need to go back and improve the platforms.”
For instance, Håkan believes counter measures forsteering and braking failure should be tested. For instance, torque vectoring simulates a side wind that would cause the vehicle to start to skid. This was crucial in determining drivers’ reactions.“[It’s important to] understand what the driver really sees, how the driver will act, and what could really be a supportive functionality,” he says.
t’s clear that passenger safety is of utmost important to vehicle manufacturers as technologies advance. Likewise, accident avoidance is a main safety concern for drivers. Luckily, several core technologies are being implemented into the autonomous car blueprint to improve upon its safety features.
Global development in this area is progressing nicely, he explained, “In Europe, I think it will be fairly easy for OEMs to have a best effort because we know that these new products will save lives.” While vision-based systems and myriad radar capabilities currently exist, hebelieves these core capabilities need to be revisited around the world. This involves validation of these systems, he says.
“How can vision be ensured to also fulfill the autonomous drive?” Håkan asked. “My firm belief is without fault injection it will not be possible to validate these systems.” Lane-keep assist was one of the industry’s key highlights of 2015, he says, even though it is not fully safe. In the future, he foresees the immersion of a vision-based assist that is supported by radar.
Håkan is mum on more specifications for the autonomous vehicle in 2016, however. He predicts that expensive redundancies will continue, but when the industry learns to use the implicit redundancy that is in the vehicle system itself, overall costs will decrease. The road toward a fully autonomous car is long and winding. A rush into the autonomous arena needs to be followed up by precision in safety and technology.
“We will use the cannon in the beginning. But it’s only a rifle that is needed in the end,” he said.
Presented in association with the Autonomous Vehicles Summit