Autonomous vehicle (AV) developers around the world have varying approaches for testing and deployment, but almost unanimously agree on one thing: LiDAR is a prerequisite for more advanced levels of automation. Suggestions that LiDAR may in fact hinder the performance of an AV would have come as a surprise for most stakeholders in the industry.
Even for Tesla’s outspoken Chief Executive, the notion that LiDAR sensors are a “fool’s errand” has proven one of Elon Musk’s more divisive arguments. At Tesla’s Autonomy Day held in April 2019, Musk blasted not only LiDAR, but also HD maps. A combination of camera sensors and a ‘neural network’ that is trained by data gathered from its vehicles would ensure that AVs do not become “brittle” and unable to adapt to certain situations. In essence, the car will perceive and understand its surroundings in the same way as a human.
There is an important distinction within the ‘LiDAR versus camera’ argument. LiDAR proponents—essentially all major automakers and AV developers such as Waymo—are not against camera sensors. LiDAR, they say, will simply be
