The self-driving car space is exciting. New tech companies join the race every year to become the first company to produce a self-driving car. The big names are Apple, Uber, Google, GM, Tesla, and many more. But the interesting part is that the technology used by Tesla is fundamentally different from the other.
LIDAR
Most companies use LIDAR – Laser Detection and Ranging. The LIDAR module is mounted in the centre of the roof, and emits millions of laser beams per second in all directions, and these beams reflect back towards the car. The four Radar sensors receive this reflected light and the system on board creates a 3D model of the space around the car. Using this data, the computer is able to drive without assistance.
A lot of computer systems are involved as well: writing code for interpreting the data from the sensors and then taking decisions to steer the car takes years to develop. At first, all self driving cars’ mission was to collect enough data to gather a data set to draw conclusion from. Machine Learning, which recently emerged, was and is key to this technology.
Its Features
LIDAR enables a self-driving car (or any robot) to observe the world with a few special super powers:
- Continuous 360 degrees of visibility – Imagine if your human eyes allowed you to see in all directions all of the time
- Insanely accurate depth information – Imagine if, instead of guessing, you could always know the precise distance (to an accuracy of ±2cm) of objects in relation to you
By firing off millions of beams of light per second, the measurements from the LIDAR sensor enable a visualization of the world that is truly 3D. You can infer the exact measurement of any object around you (up to around ~60m, depending on the sensor)
Traction of LIDAR in Self-Driving Cars
In plain words: Mapping! LIDAR allows you to generate huge 3D maps (its original application!), which you can then navigate the car or robot predictably within. Autonomous vehicles get to “see” their surrounding, and are able to view the bounds of a lane, a traffic signal or any obstacle ahead of time. This type of technolgy is exactly what today’s self-driving car needs and thus LIDAR took off.
All LIDAR has evolved, it has allowed the generation of more detailed 3D maps. As such, its use case in autonomous vehicles has been made- mot only does it allow the car to get a sense of where it is in its surrounding but also allows it to track obstacles such as people or trees.
The Tesla Way
In contrast to all such technology is Tesla. Tesla’s founder Elon Musk is very critical about LIDAR. Forbes Magazine reports, At Tesla’s Autonomy Day last month, Elon Musk formalized his war on LIDAR for robocars. In recent times, he has repeatedly deprecated it, but in the investor day he came out swinging:
LIDAR is a fool’s errand. Anyone who relies on LIDAR is doomed. Expensive sensors that are unnecessary.
Elon Musk
Tesla, among the majors, stands alone in using just cameras and radar. Tesla’s choice has simple origins–there is no robocar-capable LIDAR that can be purchased today for a production automobile.
Once you solve vision, [LIDAR]’s worthless
Elon Musk
Solving vision, so that it tells you everything LIDAR tells you, as Musk describes, is a very tall order. Tesla’s technology is putting just eight cameras (three in the front, two on each side and one in the back) plus a front radar and some ultrasonic sensors for close-in work. All this requires much more processing and interpretation power. And Tesla has done it. By making Tesla cars mass-produced, they have been able to gather data – so much data – and thus fed it into their ML. By this, they have been able to produce an algorith that seems reliable. They have already deployed semi-autonous driving tech to their cars, and Elon Musk says that all their present cars would have to ability to “self-drive” by 2020.
They are all going to dump LIDAR, that’s my prediction, mark my words
Elon Musk
He’s probably right–some day. What nobody knows is when that day is. We know it’s not today, and we can guess it’s probably before 2030. But we want to build cars today. We do know that cheap LIDAR will be here quite soon.
This means that Tesla is making a big bet. Forbes reports, they are betting the vision breakthrough will be soon. They are hoping they will be the ones to bring it about, using their custom neural network hardware and the huge amount of data they can gather from the growing fleet of Tesla cars roaming the roads, under their control from a software standpoint.
The others are taking a safer bet, that the breakthrough is further away and that the cheap LIDAR is more predictably on the way.
MORE POSTS ABOUT THIS TOPIC IN THE FUTURE
Tech Decoder 