The Road to Self-Driving Vehicles

In the last month or so I have had friends and colleagues asking me about self-driving cars and whether completely autonomous level 5 self-driving cars are possible. This was largely due to Elon Musk's statement "  I think there are no fundamental challenges remaining for level 5 autonomy". To answer this question let me first start with the history of self-driving cars.

It turns out that the concept of a driverless vehicle dates to the 1940, when General Motors was the largest car maker in the world. During a world expo GM proposed this idea, where a family in a car could be sitting around a table talking to each other and playing games while the vehicle is driving itself on the highway. So, the idea of self-driving vehicle started many decades ago.

There are many research institutes / carmakers in the world that have been working on a practical model of driverless cars from the mid nineteen eighties.

Making of a self-driving car requires one to solve multiple problems like self-driving at high speeds of 70 to 100 miles an hour, taking of exit/entry ramp, merging and changing of lanes, driving in urban and sub-urban settings, know status of traffic lights, understanding of work zones, recognize pedestrians and bicyclists, maintain requisite distance, change speeds if need be, stay away form cones, barrels and drums, and many more. There are many possible problems and how well they can be solved is another question for another day.

First, let us understand why we need an autonomous vehicle. In other words what is the use case and return on investment of building such a technology driven vehicle. The answer is as follows.

There are two million road accident deaths every year given pre corona situation. All these fatalities and accidents add up to an economic cost of half a trillion dollars every year. Humans get easily distracted and if we can take humans out of the equation of driving; fatalities, injuries, and accidents will drop dramatically over time. Hence the holy grail of having a self-driving car is to bring life fatalities to zero.

Self-driving cars will also help in productive utilization of travel time, enhanced elderly quality of life, enhanced quality of life for the physically and mentally challenged. The impact on environment due to pollution created by automobiles are detrimental. In case of all the vehicles being self-driven, there will be coordination among them and hence they will travel in harmonised speed making better and faster forward progress uniformly reducing commute time and benefiting the environment with lesser travel times resulting in lesser pollution spread. DUI of alcohol and other substances within or outside the law of the land will have no significance from an accident standpoint.

Development of a self-driving autonomous vehicle system requires to have a parallel infrastructure built within the existing infrastructure or beside it. The infrastructure must be clearly defined and the governance well laid. It is like having car for transport, trains for transport and airplanes for transport, with completely different governance.

Given the road conditions today and the numerous possibilities a self-driving car may encounter, it will call upon scientist to build various algorithms one after the other endlessly. The self-driving cars must develop perception, cognition, and action. The degree of perception, cognition and action can be developed to any level but even close to 100%. A judgmental error is inevitable and hence a completely autonomous level 5 self-driving car is not a possibility given the conditions and the technology existing today. The best way to reap the benefits of a self-driving vehicle is to go ahead and build a parallel commute infrastructure with standardization and well-defined governance.