They have a lot of sensors, software that has maps loaded, plus software that can recognize new obstacles in the road or moving into the path.Can someone explain to me, as if I’m six years old, how a driverless car works? Does it get a signal from a satellite? Does it get a signal from a cellular phone tower? How does it know where it is and where it’s going? How does it avoid road hazards? How does it drive on a snowy road in the winter?
sorry for juvenile questions, but I’m unclear what technology guides a driverless vehicle.
Those are all very good questions.
Lets back up 30 years, when the concept of a self driving car included some sort of wire embedded in the roadway. It was a neat idea, but not practical.
Now lets look at today. Google claims its cars have driven a gazillion miles with no accidents, is safer than a human driver, etc. But what has Google really done? First, a platoon of 4-5 cars goes down the road and uses video and LIDAR scans to scan the road. Then a team of people on computers do LIDAR Extraction and reduce the size of the point clouds to something manageable. This extracted data is then loaded into a ‘self driving car’ as a map. The self driving car is also equipped with LIDAR, and compares what it sees to what is in the map. Without this process, the ‘self driving car’ would not even recognize that a traffic signal is at an intersection, would not expect it, and would not be looking to determine what color it is...which means if a new signal goes up, look out.
So, what makes this any different than a car being guided by a wire embedded in the pavement?
Absolutely nothing.
Add to it the obvious problems of rain hurting visibility, snow covering landmarks and lines, freezing rain covering lenses, etc...and frankly UBER drivers have nothing to worry about. There will not be driverless cars in the next 50 years.
So what’s going on...why the push? Well, the embedded cable idea - its been adapted to factory floors and guides robots quite well. The technology was developed in the name of automobiles, and may have even gotten DOT grant money, but was always meant for the factory floor.
What Google is doing will work on the factory floor too. But its also good for other industries. An example is the rail industry. A part of the Positive Train Control (PTC) mandate includes near continuous monitoring of the track for anything that may ‘foul’ it - such as a sign post that gets tipped over and becomes too close to the track. A railroad that fails to identify and correct this in a timely manner will be ‘penalized’...and that penalty will be doled out in a slower speed allowed in a certain railroad subdivision for a specified period of time - easily enforced through PTC. So the railroads need something that can build a 3 dimensional map of their entire system, and then rapidly scan the system often, in order to recognize anything that has changed in the map...exactly what Google is doing with their cars.
Other examples would be earthmovers in a quarry, farming, strip mining, construction.
Google will certainly utilize the technology they are developing, but they won’t be utilizing it on our roads and highways.
They have a lot of sensors, software that has maps loaded, plus software that can recognize new obstacles in the road or moving into the path.GPS to know where it is (already in most cars). Traction control to handle snow (already in most cars). Cameras for lane control (going in more and more cars). It’s amazing how close to self driving your car already is. Little bit more data processing on the cameras with a longer range, a bit of LIDAR and boom, be a passenger in your own car during your commute.