#2- Technological

1. Types of technology found in autonomous vehicles:

- Radar: Accident-prevention systems trigger alerts when they detect something in a car's  blind spot.
- Lidar: A rooftop ranging system comprised of laser paints a 360-degree picture of the car's surroundings that is accurate to within 2 centimeters.
- Stereo vision: 2 wind-shield mounted cameras build a real-time 3D image of the road ahead, spoting hazards like pedestrians and animals.
- Lane guidance: Cameras mounted behind the rear-view mirror recognize lane markings, spoting te contrast between the road surface and boundary lines.
- Infrared camera: 2  infrared headlamps extend your vision at night without blinding other drivers. The  signature of the infrared beam is detected by a camera, which displays an illuminated image on the dashboard.
- GPS: GPS is accurate to within 1.9 meters. With GPS covering the macro location of a vehicle, smaller on-deck cameras can recognize smaller details like red lights, stop sign, and construction zones.

2. Closer look at sensors technology:

When designing cars, it’s in the automaker’s best interest to push hard for cost optimization, without sacrificing safety.

Due to this strategy, the camera is currently the most prominently used sensors in today’s cars, covering functions like emergency braking.

At the same time, the goal of reaching level 4 truly autonomous vehicles by 2020 (Levels of Driving Automation, SAE international standard J3016) is putting emphasis on the power of adding lidar and radar to the mix. Let’s take a look at what makes each sensor special and how it brings the automobile closer to full autonomy.

1.       Lidar is the master of 3D mapping. Lidar, short for light detection and ranging, is a technology that measures distance using laser light. The technology can scan more than 100 meters in all directions, generating a precise 3D map of the car’s surroundings. This information is then used by car to make intelligent decisions about what to do next. The problem with lidar is that they generate a large amount of data and are still quite expensive for OEMs to cheaply implement.

2.       Radar is the master of motion measurement. Radar, short for radio detection and ranging, is a sensor system that uses radio waves to determine the velocity, range and angle of objects. Radar is computationally lighter than a camera and uses far less data than a Lidar. While less angularly accurate than lidar, radar can work in every condition and even use reflection to see behind obstacles. Modern self-driving prototypes rely on radar and lidar to “cross validate” what they’re seeing and to predict motion.

3.       Cameras are the master of classification and texture interpretation. By far the cheapest and most available sensor (but not the cheapest processing), cameras use massive amounts of data (full HD means millions of pixel or Megabytes at every frame), making processing a computational intense and algorithmically complex job. Unlike both lidar and radar, cameras can see color, making them the best for scene interpretation.

Clearly all three sensors bring advantages to automakers building the next wave of connected cars but after breaking down the pros and cons of each option, we may infer some predictions.

Due to higher cost, lidar might remain a premium option for the time being as OEMs figure out the cost structure of their self-driving cars. The move toward level 4 fully autonomous cars (SAE, 2014) will require essential safety-proven technology but the journey through levels 2-3 will find lidar lagging behind in take rate.

Radar is a proven technology increasingly becoming more efficient for the autonomous car. The new RFCMOS technology recently introduced to the market will allow smaller, lower power, efficient sensors that fit right into the OEM cost reduction strategy. This will also make radar more complementary to cameras as the “dynamic duo”.

Cameras are the cheapest sensor of the three and will likely remain the volume leader the near term. Their future will be strongly dependent on the development of the software algorithms controlling the self-driving car and how it can process the massive amount of data generated. The introduction of potentially “free” algorithmic approaches from companies like Google may help change the speed of adoption completely.

REFERENCES

http://www.eetimes.com/author.asp?section_id=36&doc_id=1330069
https://www.pinterest.com/pin/327848047850806932/