Driverless cars — will it work on Indian roads?
The excitement of driverless cars is all pervading. The anticipated efficiency, redundancy of jobs, loss of emotional connect between the car and the driver and/ or passengers, and even the need to define “death algorithm” — who the car be programmed to kill in case of an emergency; make people sit up and notice this phenomenon.
While a lot is being discussed on this in the developed world, an often repeated question in India is — will driverless car work here?
The general consensus is no. The reasons showcased are pretty obvious — the relative abundance of cheap labor (hence drivers), the social antipathy to the potential job loss, the sub-optimal adoption of technology in the society (so the resistance to trust the same in case where life may be at stake), need for high context human interfaces, absence of regulations, etc. But the most commonly referred constraint is — the sheer madness of Indian roads, especially in the cities. This is so unique that no algorithm can be defined for this.
So the question is, can the Indian roads be modeled? Are they really that big a chaos what we think it to be? Let us look deeper with the aid of some public statistics:
1. India is considered as one of the most unsafe of roads — with approximately 140K deaths and 1.2 million injuries a year.
2. However, if we look deeper into this, a large number of these accidents involving fatalities are not involving the cars (they are mostly public transport and two-wheelers). Even among the accidents involving the cars, a significant number are traced to rash & negligent driving, being in inebriated states or wrong & ambitious judgments — at signals, overtaking or just making a smart but dangerous move.
3. If we look at the top six metros accounts for 3.2% of fatalities with 3.6% of the population, and that too accounting for 17.6% of the country’s four-wheeler population.
Even the following graph does portray a relatively safe picture for India roads in these cities, save for Chennai.
So looks the traffic in India is not as bad as we may feel. So looks there are a bunch of underlying logic which are playing, and maybe if we are able to code them driverless car may actually work in India.
What will be the key challenges in building coding? What will be the new areas over and above the codes for the international market? I bucket them in into the following six:
ü Dynamism of the road availability — Indian road can quickly change from available to block to dug up to one way or diverted or some other unexpected state. No algorithm can define this. It has to be picked up from notifications, media reports or social media updates. Also, this needs to verify constantly on a real-time basis — ideally with updates from users.
ü Traffic signals — Though technology is available, in most places the traffic police (if deployed at a signal) will prefer to personally operate it. As a result, it follows no logic which is can be predicted, neither is it synchronized to overall traffic movement. Most of often, people sense the next change looking at the policeman who is changing the traffic lights.
Also, there is a propensity by the drivers to avoid traffic rules whenever there is an opportunity. Drivers behind may create a ruckus to move ahead even if the signal is red — such distractions should not send out false alarms and to be eliminated at the source.
ü Pedestrian psychology — Indians have a very low sense of safety, and almost all feel protected by the sheer low probability of something happening to them.
Many will do jaywalking, will use phones, show hands and expect traffic to stop; or simply expect the driver to be scared and careful. This will not change in the near future, and there is no clear cut way to avoid this. However, blaring horn whenever someone is within a defined zone and moving in a set of predetermined directions will help to significantly reduce this risk.
ü Psychology of the other drivers on the road — Indians is always in a hurry, though most times for reasons which are not known even to us. Also, the resources are limited and the rich-poor divide its stark. This brings in significant aggression from many to the road. The weak compliance system, coupled with actual convictions makes it worse. The best way to address this is defensive driving — never over speed even if the road is available, never overtake unless there is an unspoken agreement with other car and most of all, do not do anything which may look smart to the other person. In short, let’s not allow the other driver to get the goat.
ü Animal psychology — This probably the simplest of the lot. The three animals most seen are the dogs (which protect themselves well unless the speed is high), the cow/ buffalo (usually with a guide and in a group), and animals of burden (they are slow and more manageable). Low-speed city driving will help address this issue adequately.
ü Curiosity — Indians are curious, and express the same without much constraint. Be ready for close inspection, sudden adulation or even being accosted by a group on the road. Well, this is fun, so long it doesn’t become a safety issue.
The few driverless cars which with the first ones to hit the road should probably look at using texts and diagrams on the body in English, Hindi and local language — to explain the entire concept of driverless cars and how it works. Hopefully, it will help address a significant part of the attention.
References:
https://en.wikipedia.org/wiki/List_of_cities_in_India_by_population
Road Accidents in India — 2015, Ministry of Road Transport & Highways
http://www.mapsofindia.com/top-ten-cities-of-india/top-ten-highest-no-of-car-ownership.html
World Resources Institute (http://www.wri.org/)
