Level 3: The tipping point for AD
Technology and Business
In this article, we look at how far autonomous driving has come in recent years and ask:
- Have we reached a bottleneck at autonomous level 3?
- What’s the impact of region on our progress at level 3?
- Is safety still a concern in the development of level 3 autonomy?
Whenever we discuss driverless technology, it is useful to keep the now well-established levels of autonomy at the front of our minds. Looking at the enormous scope for variation which exists on the scale of 0 to 5, where 0 includes no automation of function and 5 describes a system which can perform all driving functions independently, it’s important to remember that the terms ‘autonomous’ and ‘driverless’ don’t only apply to the highest levels.
Right now, it’s level 3 which is most desired, and closest to reality. But it’s also here that developers are facing something of a bottleneck. Inspired by our Chinese sister site, we’re taking a closer look at what’s happening at that level.
Are we stuck at level 3 autonomy?
Whether or not progress has halted at level 3, this stage of development, where automated driving can occur for short periods of time, has been a point of discussion recently. Vitally, the system behind level 3 automation needs to be able to recognise its own limits and notify a human driver to resume control in time. That poses a huge challenge which is proving difficult to overcome.
Last year Audi claimed that the A8 would become the first level 3 autonomous car in the world, but was never able to realise this goal as liability and regulation issues across Europe meant it was eventually released with just level two autonomy. Honda, however, are now set to be the first to mass produce cars with this level of autonomous capability. Before the end of March 2021, they plan to launch sales of a Honda Legend equipped with level 3 automated driving equipment.
Despite this pending success, the topic remains controversial for many. Some think that level 3 is merely a step in the road to level 5 and ultimately can, and will, be crossed as technology improves. Others believe it is more significant because, as the definite midpoint in driverless tech, it has huge potential for mass production and could be key to overcoming technological, logistical and legal problems.
See how different regions are working on autonomous travel in our round-up of 10 unexpected places taking on the driverless challenge
Xu Tao says that Continental overcomes this with a global autonomous driving research and development structure, with local teams in Germany, the United States, China, Japan and other regions for localised development and collaboration. On the premise of focusing on the core technology of autonomous driving, Continental will also focus on some special regional requirements, and integrate these special requirements into the global development and co-operation system. It’s not only regulation which is at fault for the bottleneck at level 3, Xu Tao explains that safety, as ever, is also at the center of consideration.
Is safety still a concern in the development of level 3 autonomy?
Of course, as the level of autonomous driving increases, the safety difficulties that need to be overcome are also increasing, which is one of the reasons why research and development at level 3 encounters bottlenecks.
While at this level an automatic driving system continues to perform driving tasks where it can, a driver must be able to take over the vehicle at any time. As the responsibility switches from a person to their vehicle and back again, the safety credentials of autonomous driving systems become more urgent.
Xu Tao explains, "this element of safety is embodied in the sense-plan-act' driving effect chain of autonomous driving." Sensing is to perceive the environment inside and outside the car; planning is to perform calculations based on perception data and make corresponding control decisions; action is to control the vehicle.
Level 3 autonomous driving puts forward higher requirements for the vehicle's perception, planning and action capabilities. Sensors must capture massive amounts of information, the ability of the computing unit to analyse data will also be crucial, as will measures to ensure driving safety.
As we progress beyond level 3, those concerns will only increase. Yet, with thorough and forward-thinking development at this stage, the path to higher levels of autonomy could be smoothed out. It seems clear that we will spend some time getting level 3 autonomy right, but once we clear the massive challenge of the system recognising its own limits on time, levels 4 and 5, and hence the real driverless dream, become a lot closer.
We’re always eager to hear what you think. Is level 3 the best we can hope for right now? Or is it the key to speeding up progress? Share your thoughts in the comments below.
5G expert Phil Cottom on the link between connectivity and driverless tech
Will everyone win in a driverless future? - An expert’s perspective
Submit your story
Become part of our autonomous revolution and submit your stories, images and videos
Stay up to speed with our weekly briefing. Enjoy autonomous driving content direct to your inbox