The viability of mass-market autonomous vehicles (AVs) hangs in the balance. The global market is expected to decline by more than 3% during 2020 as a result of COVID-19 and may take years to recover. High technology costs, reliability concerns, a lack of industry-wide standards and the lack of an appropriate regulatory framework mean the automotive industry has struggled to introduce autonomous features beyond luxury vehicles and into the mass market, while those same mass market brands are focusing their investment on electric vehicles. Despite these headwinds, the industry remains committed to crossing the chasm from Level 2 autonomy to either Level 3 or 4.
For Level 2 autonomy, the driver must constantly supervise autonomous functions, for example a driver’s hands must be on the wheel for lane keeping steering assist functions. Progress to Levels 3 and 4 will mean vehicles can operate completely autonomously, with all safety critical functions performed by the vehicle, although the driver must be ready to take over when needed at Level 3. This journey to Levels 3 and 4 is defined by a higher cost of AV technology, regulatory restrictions and lack of consumer trust, all of which combine to create a barrier to adoption that no amount of bullish optimism from Elon Musk seems likely to fix.
High technology costs, reliability concerns, a lack of industry-wide standards and the lack of an appropriate regulatory framework mean the automotive industry has struggled to introduce autonomous features beyond luxury vehicles and into the mass market
While Tesla has been promising a leap straight to Level 5, which would mean it avoids the challenge of ‘take back control’ interaction between human and machine, this would be better described as ‘Level 5 ready’. Such vehicles could theoretically be autonomous in all scenarios and driving modes, but the wider environment—including testing, regulatory compliance, driver trust and the influence of environmental factors—will not be ‘ready’ for several years.
So how can the industry unleash the huge potential of vehicle autonomy during the current storm?
Reducing costs
Sensing systems, and the AI tools that turn data into decision making, cannot yet navigate freely in all scenarios, locations and weather. For now, the most sophisticated autonomy will be in areas with benign weather, many testing miles and precise, 3D/HD maps. To increase the area in which AVs can operate it’s important to take a hybrid approach that uses real-world and simulated testing. This is not only prudent for reliability but is also more economically viable.
LiDAR systems remain the gold standard for accurate sensing. But despite recent cost reductions, LiDAR is too expensive for mass-market vehicles. In response, today’s state-of-the-art takes cheap, low-resolution depth sensors and cameras and applies a neural network to build a detailed depth point cloud—a 3D map of the environment around the vehicle—at a fraction of the cost of LiDAR.
In the same mould, we’re now seeing self-driving systems developed, trained and tested using simulations, while AI techniques are used to generate an endless supply of road environments, lighting conditions, sensor noise and damage, ensuring the technology is robust. Such creative approaches that apply cutting-edge technology to the major costs associated with autonomy will now eat away at the high costs seen to date and give suppliers a competitive advantage.
Look to logistics and deliveries
COVID-19 has affected supply chains for food and medicine, particularly in the ‘last mile’. This may provide the pressing urgency for AVs that’s lacking in the consumer vehicle market. Amazon’s recent purchase of Zoox is an effort to reduce its massive logistics overhead and ensure business continuity in the event of further pandemic spikes. In China, AVs have been used to deliver medication and disinfect roads. Such initiatives provide valuable road-testing time and improve consumer confidence in a self-driving future.
The ultimate goal is not just to deliver returns on the autonomy investments made to date, rather it’s to engage more drivers with autonomous features that improve safety and to unlock the huge efficiencies available in wider autonomy applications
Regulatory change is on the way
Level 3 and 4 autonomy is currently illegal in most markets, though regulatory change is beginning. Germany is building a legal framework to allow Level 4 vehicles to operate on its roads and as the leader of Europe’s automotive industry it’s likely that the EU will follow Germany’s lead.
US regulation varies from state to state, while the federal government regulates the construction and reliability of vehicles themselves, creating a complex environment for AV companies. The AV START Act attempts regulatory overhaul, but the legislation remains stalled, which could hinder the US automotive industry and its international competitiveness.
These then are the challenges that the automotive industry must address to ensure autonomy can become increasingly sophisticated and affordable. The ultimate goal is not just to deliver returns on the autonomy investments made to date, rather it’s to engage more drivers with autonomous features that improve safety and to unlock the huge efficiencies available in wider autonomy applications, such as logistics and deliveries.
The opinions expressed here are those of the author and do not necessarily reflect the positions of Automotive World Ltd.
Thomas Carmody is Associate Director, Head of Transport and Infrastructure at Cambridge Consultants
The Automotive World Comment column is open to automotive industry decision makers and influencers. If you would like to contribute a Comment article, please contact editorial@automotiveworld.com
