1. Will these cars “see” and “react to” the environment, fulfilling the promise of saving millions of lives?
2. What are the gaps or roadblocks in current AV and ADAS technologies?
Autonomous vehicles “see” and “react to” the environment through a combination of sensors, each possessing their strengths and weaknesses. Thus far, car companies have been trying the find the perfect recipe of sensors to achieve this end of goal of vehicle situational awareness, but have yet to find the right mix. So what are the gaps or roadblocks with these sensors?
LIDAR, or light detection and ranging, use lasers as a means to measure distances between objects, in this case a vehicle and its surroundings. This is combined with software and AI to create predictive behavior of elements in the environment and, theoretically, allow the vehicle to drive itself.
It is the current popular choice for 3D environment sensing, as it has the highest resolution of all existing sensor tech. However, this resolution comes with a hefty price tag. Velodyne LIDAR, popularized by its use on Waymo vehicles (formerly Google’s Self-driving Car Project), has been successful bringing their price of ≈$80,000 to around $7,500, but this is still unrealistic for the automotive industry.
If you’ve followed the AV emergence, you’ve likely heard of Quanergy, a company touting a $250.00 solid state LIDAR unit. While this could be highly disruptive, it raises a few questions:
• If the unit is solid-state, can it provide a 360° field of view? If not, how many $250.00 units are necessary for a complete system?
• Can these, or any LIDAR units, perform in common weather such as rain, snow, fog, or dust?
1. It’s expensive. Plain and simple.
2. Because of its inherent reliance on light, particles from weather disturb LIDAR sensors, preventing an accurate reading of the environment.
RADAR, or radio detection and ranging, works in a similar manner as LIDAR, but it relies on radio waves as a means to detect distances between objects instead of light. Waves transmit outward, bounce of elements in the environment, and are then picked up by a receiver. From the time between transmission and reception, distance can be derived. Similarly, to LIDAR, through a dash of software and/or AI, vehicle situational awareness is possible.
Existing RADAR is directional and used for things like automatic braking and lane guidance. It has lower resolution than LIDAR, and because it is directional, it has a limited field of view. The strengths of RADAR are that it is cost-effective for auto-makers at a price of $150.00 - $200.00, it is highly reliable, and can operate at long and short ranges (100-200 meters).
RADAR can perform in any weather or lighting condition, making it a prime candidate for use in the final iteration of AVs.
Pros: High resolution
Cons: Expensive, unreliable in adverse weather conditions
Pros: Highly reliable, long and short range, cost-effective
Cons: Limited field of view, lower resolution
It will be interesting to see which sensor technology becomes the market leader for situational awareness. Will LIDAR overcome its reliability issues? Can RADAR be expanded to 360°? Experts suspect sensor fusion or the use of multiple sensors to cross verify each other. Either way, we are at the cusp of a technological breakthrough and it will be a very different world in 5 years.