Search the internet and other sources and list the different types of technology that may be found in autonomous vehicles (include at least one more option than those listed above). Describe each technology in 1-2 sentences, and include at least one web link.
  • ·         LiDAR: Using laser in order to scan the environment, it is a suitable and more precise method of reading the surroundings as it provides a detailed 3D representation of the world. Usually a point-cloud is extracted for further analysis from the system, which then can be used to detect object shapes and so on [1].
  • ·         GPS: GPS systems, widely used in navigation, is a system that triangulates your position in the world and retrieves you the coordinates you are currently at. It employs satellites to calculate the position. Usually the relative position is provided, where a reference base-station is used to accurately calculate the position[2].
  • ·         Camera + infrared: Systems like Microsoft Kinect incorporates the synthesis between a 2D camera image and infrared 3D scanning to sense the environment, enabling tracking of a persons joints and position from the local camera coordinate system. As such, a vehicle would be able to also identify and track behaviors of pedestrians using similar techniques. I have myself worked with the Intel R200 camera, which provides similar capabilities as the Kinect. It seems to be a very viable and cheap solution to using LiDAR scanners. However, as I’ve come across, it has its drawbacks. Mainly that the camera we used was “blinded” when out-doors[3].
  • ·         Machine learning: While this techniques includes, usually, all of the aforementioned technologies, it still enhances them enough to be considered as its own unique system. I go into more detail below[4].  

Select one type of technology and look at it in greater detail. How is it currently being used? What are its benefits and drawbacks? What is its future potential? Don’t forget to include references to any sources you may have used.

I cannot find any commercial usage of this technology in cars on sale today. However, much testing within the field is done by most automotive manufacturers. As of now, usual systems “learn” by guessing and confirming the results. As such, systems will be unreliable in their infancy. But as more systems start learning and sharing their memory, it has the potential to be at least as good as human drivers in differentiating between objects. Furthermore, even the US government vehicle safety regulators deemed current systems to be as good as human drivers [5]. Therefore, it is safe to say that such systems will become highly implemented in the coming years, even in “normal” cars.





Comments

  1. It is interesting that you are considering that autonomous vehicles may "share their memories" and learn from each other in a large network. That could have a huge impact on the speed of improvement of the algorithms and behaviours. Well done. - The CDIO Academy Team

    ReplyDelete

Post a Comment