Half of the population on earth is living in the city, by 2050, the proportion will reach 66%. Overcrowded streets Sensor are increasingly becoming worrisome, but the Indian road may help us find a solution.

   When we were in the crowded streets of Bangalore, India, the rickshaw driver Raju explained to me the rules of the road, he said: 'In 60 seconds, you must consider 70 options.' Lived in India for two and a half years later, I understand What he said. In India, every inch of the road will not be wasted. If there is a gap, there will be a locomotive to fill it up. The car's bumper will collide with each other, overtaking frequently occurs, the indicator and the rearview mirror have become optional attachments, become dispensable. Most of the drivers seem to rely more on the nearby vehicles to keep the whistle to judge, they are like an echolocation system. These are the unique ways for Indian drivers to deal with the madness of the road, and they are used to predicting those unpredictable things. 'We expect anything to happen,' Raju explains, and we can only make predictions, and we are always looking forward to the front of the car turning left or right, and even suddenly braking. This 'driving philosophy' has a complex cultural and historical roots, but it is also the product of the rapid growth of urban population and car ownership. Government data show that India's car ownership increased almost fourfold between 2000 and 2013. This growth rate exceeds the speed of the new infrastructure, so that the public can not adapt.

   India is not a case, in China, Brazil and the Philippines and other countries, there are urban congestion and car ownership surge similar situation. But for many technology companies and researchers, India's chaotic path makes it a perfect test ground. The overcrowding of the city is a complex and subtle problem. There is no panacea that can be solved. It needs a comprehensive plan to solve all aspects of infrastructure, income inequality and energy supply. But for the roads of these cities, technical experts around the world are pursuing a grand idea: unmanned cars have brought to ensure smooth traffic and help us optimize the journey. Their point of view is that intelligent robotic cars can communicate with each other, better perceive obstacles and create more organized traffic. Most of this technology fuel metering valve is developed for the orderly streets of the West. After experiencing the traffic conditions in Delhi, India will be the last place to deploy unmanned vehicles. However, this has prevented the design firm Tata Elxsi, which is developing an unmanned system that can retrofit any car and has begun testing the prototype on a test track near Bangalore's headquarters.

   Road tests are still very far away, they have a clear understanding of the challenges facing. 'The driving behavior is very precarious,' says Nitin Pai, director of marketing and strategy at Tata Elxsi. 'The rules are not rules, and more often serve as a guideline.' This is the reason to study the Indian road, Help us come up with the best idea. Unmanned vehicles rely heavily on machine learning, that is, artificial intelligence (AI) uses massive amounts of data to train their abilities, such as identifying vehicles and predicting their trajectories. So far, these cars have been in the United States San Francisco and Pittsburgh and other places on the road to test, in China and Japan and other countries have also been tested. However, most of the research is carried out in the West, because the driver's behavior there is considered to be predictable, the road has a clear sign, the test is more reliable. 'These are the basis for your establishment of an unmanned vehicle system,' Nitin Pai said. 'The Tata Elxsi team has trained their systems with third-party data collected by researchers Temperature Sensor at the Karlsruhe Institute of Technology in Germany. This is done on the German road, using a car equipped with a high-resolution camera, GPS and laser radar sensor. Among them, the lidar can use the laser and fly sound to carry out very sensitive distance measurement.

   Rajesh Kumar, head of tactical initiatives at Tata Elxsi, says they have developed algorithms that use these data, but the effects of machine learning are closely related to the data you provide. If we want to solve the road problems of overcrowded developing countries, we need to obtain data from these environments. There are many obstacles on the Indian road that do not appear on the German road, which means that AIs who receive German data training will not be able to identify all the objects encountered by unmanned vehicles traveling on the Indian road. In India, motorcycles and electric cars are more common, and cars compete with non-traditional road users. Raju laughed and said, 'There will be people, cattle, monkeys and elephants on the road, and they are also part of our traffic, so we always need to be alert to them.' All of this can help unmanned car technology change Better and more secure, and in the absence of Indian road data in the case is impossible to achieve. If unmanned vehicles are to travel on Indian roads, it is important to capture these differences, but it is also a data gold mine for researchers who want to improve the technology for other developing countries. As a result, these researchers have several drivers full of sensors every week, wandering around the Whitefield in Bangalore to build a local data set of high-definition video, lidar data and high-precision GPS data.
  In the United States or the United Kingdom work, does not mean that it can also play a role in our city. These urban environments are completely different, and the situation in India is much more complicated. So we need local solutions. 'David's lab has built a camera and sensor network in a commuter corridor in Chennai, where low-power wireless communications technology transfers data to IIT's temporary traffic control room to help them start building models of Indian traffic. Said: 'In the Western traffic, the vehicles are arranged in an orderly manner, during the congestion, they have set aside the smallest interval. In India, any space you have will be occupied by the vehicle. 'Many of David's students are trying to simulate the two-wheeled vehicles moving around the vehicle, and they compare this to the flow of liquid through the sandbag. At the same time, one of their models has been put into use through the city bus's GPS data Predict the arrival time. David said that Europe also used a similar approach, but they tend to set a uniform speed on the sub-part of the route, but in India you can not assume that everything is uniform or use any predictions A method ABS Sensor of monitoring traffic is the use of 'induction coils.' A wire is laid under each lane, and when the vehicle passes, the magnetic field of the wire is disturbed, causing the signal to be To the counter.However, this method only works when the vehicle is in strict accordance with the lane, but in India, this situation rarely occurs.Therefore, David's team to establish a system, even when the vehicle at the same time, Can also accurately detect the type of eight vehicles.

   David believes that technologies such as bus forecasting systems can reduce the burden by encouraging people to use public transport. In the passenger survey, they found that many people prefer to take the bus, but simply can not rely on them. David's team is also working on road planning systems to optimize the use of roads. Their traffic forecasting system will intelligently control traffic lights according to traffic flow, rather than using timers as they are now. All technologies can only provide a supporting role, the best way to solve Pressure Switch the traffic dilemma should also focus on the human body. Some traffic engineers think that trying to change the habits of Indian drivers may be a better way, but David thinks it is beyond their control. Like most things in India, it is easier to go with the flow.