Autonomus Vehciles will reduce accidents by 93%. Julia Markovich, Transportation Analyst with the Conference Board of Canada, stated “A system of auto mobility became entrenched in the 20th century. Approaches to government, policy and design were all based around the car. We enforced that system of auto mobility,” said Markovich. “The culture is now moving into a new era.“

As 90% of accidents are due to human error this is an easily achievble % in the present environment. 
What will alter this?
As Autonomus Vehcilces are introduced there will be an imediate decline in the number of accidents. In fact, we alreaady harvested this in the mining industry. It is proven that the 400 tonne Autonomus Pilberra trucks have reduced accidents. The only accidents that happen are the smaller non autonomus trucks trying to cut through before the driverless trucks come through.
So the figures are allready becoming murky as we have already strated the reduction.
Vehicels will be fit for need. They will deliver indivduals, couples and bus loads on a fit for use basis. Light rail, driverless cars and busses will run around in complete understanding on what “other traffic” is doing.
Due to the transition from human driven to autonomous a spike will happen with humans wanting to interact, both maliciously and in error with the units. To reduce this dedicated lanes, tracks and areas are the best line of defence, but costly.
So between now and 2025, how will you react?

Self Driving cars – How do they Drive

There are many devices and integrated items that ensure that a self driving car drives safely and within the boundaries that we set on to it. Through my Self Driving Car Engineer NanoDegree I have learnt many of the ways of controlling Autonomous Vehicles through the use of sensors utilising robotics to enable the car to navigate and drive accordingly.

A newer approach, that I have been concentrating on, is the use of Deep Learning capabilities for the cars to mimic human driving behavior. Both of these, robotics and neural networks, together enable the car to be safe and reliable.

Initially lets look at video capabilities to stay within the lines.

Using Python Code I developed the vehicle to understand where the lines where in the road.
1  Yellow to white – This was to capture any yellow shades to be processed in B&W
2  Grayscale – The modification of the image to a shades of grey
3  Gaussian blur – Slight blur to remove ‘noise’ from the image
4  Canny – With a low and high thresholds

5  Region of Interest – Selecting two areas that will only be reviewed
6  Hough – Lines detecting the lines intersections and outputting an array of endpoints
7  Drawlines – Left and right slopes defined and lines drawn

Why are these so important. Well this video shows the output of the lane lines project.

An Autonomous Car can not drive on this alone, far from it. It is very basic and will only work within the boundaries that we have provided.

Next I will show some basic Neural Networks and how they adopt the learning of street signs.