20 Oct LIDAR: Delaying the Release of the Autonomous Car?
The LIght Detection And Ranging system, or more fondly known as LIDAR, has quickly became an essential component of the autonomous car. LIDAR is the eyes of the driverless car, and operates similar to the well-known RADAR system. The LIDAR system generates pulsed laser beams that reflect on hitting an obstruction. These reflections are read by a sensor to draw a 3-d map of the surrounding. Waymo, Toyota and Uber have employed LIDAR as their driverless car’s topography surveillance system.
But not Tesla, who have opted to install cameras for navigation, instead. Considering its benefits, LIDAR should have been a no contested choice, right? This question made us venture into some of LIDAR’s drawbacks. Here they are:
Bulky in Size
If you have ever seen Google or Uber’s driverless car on the road, the one thing you cannot fail to notice is the large contraption mounted on the car’s roof. That giant dome is the LIDAR system. The entire system is responsible for imaging the car’s surrounding topology, making accuracy imperative. For this very reason, some manufacturers install more than one LIDAR, coupled and placed together, making the entire system extremely bulky in size. The LIDAR must also have unobstructed 360 degree view, which is why it must be mounted on the roof of the car. This unnecessary increase in weight could affect the car’s overall performance, not to mention the hideous look.
Lidar systems are extremely expensive, each device costing anywhere from thousands to tens of thousands of dollars. The price considerably shoots up considering the fact that more than one LIDAR system is needed per car. Additionally, to cover LIDAR’s flaws, manufacturers also need to install other mapping equipment like Cameras and RADAR, bringing the overall surveillance cost to a big number. Driverless Taxi owners might have to charge customers extra to cover this expense. Elon Musk of Tesla claims that optical sensors and RADARs alone are sufficient to perform the task of topography mapping, which is why they haven’t installed LIDAR in their autonomous car.
Google’s current LIDAR system weighs over 80Kgs and each sensor costs over US$70,000.
Degraded Performance in Dense Weather Conditions
A LIDAR system alone cannot draw a complete, coloured picture of the surrounding topography, like a camera does, but instead maps out a point diagram of the surrounding. The system cannot recognise colour or contrast. These factors do not affect the autonomous car’s navigation in regular light and weather conditions, where it works flawlessly. A LIDAR’s accuracy, however, seems to falter in conditions of heavy snow or fog, because the laser pulses find it hard to pierce through dense air. This creates a major problem for the autonomous car that relies on LIDAR only. Manufacturers are currently employing LIDARs along with Cameras and RADAR systems to ensure accuracy at all times.
Only time will tell which of the two choices: The one made by manufactures like Waymo and Uber to implement LIDAR, or Tesla’s choice of optical sensors and RADAR over LIDAR, is optimal. LIDAR developers like Velodyne, who manufacture the system Google uses, are working towards a smaller and cheaper version of a fully operating LIDAR system. Until the size and operational flaws are addressed, however, autonomous cars will continue to carry the bulky machine on its roof and will couple it with cameras and RADAR for visual sensing.