The long road to autonomy
04 August 2020
Only a few years ago, the prospect of self-driving cars seemed tantalisingly close. But work on the complex technology looks to be ongoing. Autovista Group Daily Brief journalist Tom Geggus considers the future of the ′future of transport’.
It has not escaped the attention of major media outlets that 2020 was supposed to be the year of the self-driving car. The likes of Bloomberg, the New York Times and Vox have all pointed to the absence of autonomous vehicles on the roads.
News of the technology’s development saturated the last decade. Reports of Google secretively working on an autonomous car surfaced in 2010. In 2015, Toyota announced plans to launch automated driving technologies by around 2020. Two years later, Honda projected it would have highway-based self-driving capabilities by this year. Last year, Elon Musk estimated that by mid-2020, Tesla’s self-drive systems would allow drivers to not pay attention to the road.
So while consumers readied themselves for more free time behind the wheel, it appeared the technology encountered some roadblocks. For one, navigating even mid-level autonomous technology through complex and ill-defined legal systems appears to be an ongoing issue for developers.
In May, Audi announced it would not be featuring a Level 3 system in the new A8, as it had planned to only a few years earlier. The carmaker pointed to issues with legal structures, explaining none currently exist for its system. More recently, Tesla was banned from using advertisements in Germany that reference its ′Autopilot’ system as a fully-autonomous driving experience. A court in Munich decided the adverts, which stated Tesla vehicles had ′full potential for autonomous driving’ and ′Autopilot inclusive’ were ′misleading’.
Goggo Network is looking to develop a legal and engineering framework of European autonomous mobility networks. One of the company’s founders, Martin Varsavsk recently appeared in an online Move seminar. He said that Europe is currently ′greatly behind’ in terms of autonomous-vehicle regulation, which was interesting as it represents the world’s largest transportation market in monetary terms.
However, this is a claim the European Automobile Manufacturers’ Association (ACEA) would likely dispute. According to its latest industry pocket guide, Europe leads the world when it comes to self-driving vehicles, responsible for a third of all global patent applications. Even if this is the case, more defined regulations would only stand to benefit those filing patents as well as those developing the resulting technology.
Fortunately, there has been some recent progress. In June, the United Nations announced some 60 countries had reached a mobility milestone. These nations agreed to adopt the first binding international regulation on Level 3 vehicle automation. Entering into force in January 2021, it will apply to automated lane-keeping systems (ALKS) in passenger cars. It sets out performance-based requirements that must be met by manufacturers before ALKS-equipped vehicles can be sold within the countries. For example, under the regulation, non-essential displays will need to be automatically suspended when the driver has to take back control.
The coronavirus conundrum
Coronavirus (COVID-19) has proven itself to be a double-edged sword when it comes to self-driving systems. On the one hand, manufacturers find themselves having to make sweeping cuts to tackle the financial implications of months of shutdowns. In April, self-driving ride-sharing company Lyft announced it was cutting its workforce by 17%, equalling 982 employees. It stated this move was made to reduce operating costs, in light of the ongoing economic challenges posed by the pandemic. Reports of redundancies at Zoox also surfaced, with 120 people estimated to have lost their jobs in the same month.
On the other hand, social distancing looks to remain a cornerstone of COVID-19 spread prevention. This means self-driving systems could be more important than ever, particularly where last-mile deliveries are concerned. In San Francisco, self-driving car company Cruise helped make contactless food deliveries to vulnerable communities. The UK also saw autonomous drop-offs, as miniaturised delivery vehicles from Starship Technologies transported food across the town of Milton Keynes.
Collaboration is key
Leaner budgets can have an upside too. Manufacturers that already invested large sums in self-driving systems are less likely to just throw in the towel. Instead, there is the potential to turn others in the industry, forming agreements and collaborations. In June, Volvo partnered with Waymo, while Amazon signed an agreement to acquire Zoox. More recently, Ford announced it would work more closely with Intel’s autonomous arm, Mobileye. These collaborative efforts will not only help share costs but knowledge, spurring the advance of autonomous technology.
AImotive is one of the largest independent teams in the world, developing automated driving technology. Working with partners like PSA Group, Volvo Cars and Samsung, the company focuses on three main industry challenges: software, processing hardware, as well as development and validation tools. Speaking with Autovista Group, Gabor Pongracz, product manager for AImotive, outlined how important collaboration has become key to the development of autonomous technology.
′In the last couple of years, OEMs and tier ones have realised the complexity and the challenge of bringing automated driving to the roads is probably much larger than initially anticipated,’ he said. ′It’s like this monolithic approach turned out not to work. Different companies have different strengths and weaknesses and it makes sense to combine these in order to overcome the challenges that we face.’
An automated driving platform has a complex ecosystem, he explained. But this leaves experts like AImotive to pioneer advances in software, allowing other companies to focus on their own strengths. From sensors to processing hardware and analysis software, simply developing self-driving systems can be challenging enough, let alone making the technology commercially viable. This is where industry collaboration comes into its own.
An autonomous accessory?
While some might consider self-driving technology a feature relatively exclusive to high-end luxury cars, its potential contribution to safety cannot be understated. In 2019, scientists at the UK’s Transport Research Laboratory estimated the expected mix of automated and non-automated vehicles on the roads by 2040 could lead to 22% fewer collisions.
Pongracz explained that autonomous drive solutions must first be safe themselves, meaning they must comply with strict safety standards. Secondly, self-driving systems are capable of relieving the driver from potentially tedious or fatiguing tasks, like long stretches of highway driving, allowing them to be more rested for urban areas.
′Just imagine that you have a long daily commute to your work. You can let your car drive in the traffic jam on the highway, and you can relax, read something and when it comes to the city where simply the automation is not yet possible, you can take over the driving task and focus focused on the driving itself,’ he said.
Evidence of autonomous technology creating safer driving conditions is appearing. In Q2, Tesla recorded ′one accident for every 4.53 million miles (7.29 million kilometres) driven, in which drivers had Autopilot engaged.’
′For those driving without Autopilot and without our active safety features, we registered one accident for every 1.56 million miles driven,’ the carmaker went on to say. ′By comparison, NHTSA’s most recent data show that in the United States, there is an automobile crash every 479,000 miles.’
Level 5 finale
When people dreamed of self-driving cars a decade ago, they probably imagined being chauffeured around by a robot car, leaving them to sit back and enjoy the entire ride without having to once lift a finger. Today this would be understood as Level 5 autonomy. In other words, a vehicle capable of performing all driving functions at all times.
With the likes of Audi struggling to establish a lower Level 3 system, it might seem implausible that anyone would be close to Level 5. But this is exactly what Elon Musk, CEO of Tesla, claimed in July.
′I remain confident that we will have the basic functionality for level five autonomy complete this year,’ Musk told the World AI Conference in Shanghai. ′I think there are no fundamental challenges remaining for level five autonomy.’ Tesla was approached by Autovista Group to expand on Mr Musk’s comments; however, the carmaker failed to do so prior to publication.
Discussing the advancement of autonomous technology, Pongracz identified manufacturers having to be more conservative with how they mapped out their autonomous roadmaps.
′If we simply look back on some statements, where it was expected that you would have Level 4, Level 5 robot taxis roaming around the cities in 2020, that’s clearly not the case’, he said. ′Essentially, all of the industry players have pushed back the timelines and created more conservative roadmaps for automation.’
So, while we may be waiting a while longer for advanced self-driving systems, this is where companies like AImotive and a sense of general cooperation can be a saving grace. By sharing the complex tasks that come with developing what could potentially be a lifesaving system, hopefully, it will not be too long until autonomous vehicles properly take to the roads.