Autonomous vehicles (AV) are highly impressive pieces of kit. The potential these vehicles hold to revolutionise mobility, even with today’s early maturity levels, is simply remarkable. However, while the AV looks great on paper, ensuring that in practice these vehicles can adequately perceive and position themselves within their environments is an area of huge interest and concern.
Perhaps the most discussed area is that of LiDAR. Having been the talk of the proverbial town in the early part of the decade, enthusiasm around the technology has begun to slow. Many remain convinced it is needed in one form or another to enable autonomy, but others continue to worry over the size of the equipment, its cost and its lifecycle.
Level 3, 4 and 5 AVs are highly unlikely to see any wide-scale deployment for several years, even decades. As such, the small wins made today must be celebrated
The alternatives are also somewhat flawed. The low resolutions associated with radar mean much hope is being placed on the next step: imaging radar. However, this technology is still very early on in its lifecycle. Cameras on their own are also not enough to enable autonomy.
However, these concerns are not halting development. In unison, these technologies are able, to a certain degree, to counter their flaws. Some even believe that through the use of clever machine learning techniques that these flaws could be cured entirely. More work is needed, however.
On top of this is the issue of cybersecurity. While the Jeep Cherokee hack of 2015 did not involve an AV, it did highlight significant security concerns that remain today. For instance, in June 2019, sensor security company Regulus Cyber demonstrated it was possible to trick a Tesla Model 3 into taking an incorrect turn. While these researchers were not able to gain access to the vehicle’s core driving systems as with the Cherokee hack, the fact they were able to override the Model 3’s driving decision making sets a significantly worrying tone.
All is not doom and gloom, though. Current limitations aside, it must be remembered that these technologies are still at the very beginning of their operational lives. Level 3, 4 and 5 AVs are highly unlikely to see any wide-scale deployment for several years, even decades. As such, the small wins made today must be celebrated.
Even by the most conservative estimates, the LiDAR space alone is already worth more than US$1bn
The arrival of 5G is one such win. With countries across the world beginning to roll out their 5G networks, a whole new array of positioning tech possibilities could be enabled. It could also open the door for more complex connected solutions which could plug the gaps left by the LiDAR, radar, camera combination. High definition mapping techniques and offerings are also improving. While they may be thought to be the last resort should an AVs main sensors fail, these maps are still an essential part of AV perception foundation.
With each day that passes, more time and money is being invested in AV positioning tech. Even by the most conservative estimates, the LiDAR space alone is already worth more than US$1bn. Even areas that could be considered niche are receiving significant interest. Tactile sensing, which refers to the use of pre-existing in-vehicle sensors to record vehicle parameters such as weight, RPM and tyre wear as well as road conditions, could be worth more than US$4bn by 2023.
Which such interest in this space, the industry is setting itself up for an extremely productive decade.
‘Special report: The current state of autonomous vehicle positioning technology’ is available to download and read now
