0:00:00.691,0:00:01.969 I was only nine 0:00:01.993,0:00:06.916 when my grandfather first described to me[br]the horrors he witnessed six years earlier 0:00:06.940,0:00:09.844 when human stampedes killed 39 people 0:00:09.868,0:00:11.884 in our hometown of Nashik, India. 0:00:12.461,0:00:15.750 It was during the 2003 Nashik Kumbh Mela, 0:00:15.774,0:00:18.746 one of the world's largest[br]religious gatherings. 0:00:19.213,0:00:22.943 Every 12 years,[br]over 30 million Hindu worshippers 0:00:22.967,0:00:24.546 descend upon our city -- 0:00:24.570,0:00:27.410 which is built[br]only for 1.5 million people -- 0:00:27.434,0:00:29.077 and stay for 45 days. 0:00:29.601,0:00:32.672 The main purpose[br]is to wash away all their sins 0:00:32.696,0:00:35.021 by bathing in the river Godavari. 0:00:35.807,0:00:38.045 And stampedes may easily happen 0:00:38.069,0:00:41.346 because a high-density crowd[br]moves at a slow pace. 0:00:42.002,0:00:46.311 Apart from Nashik, this event happens[br]in three other places in India, 0:00:46.335,0:00:48.054 with varying frequency, 0:00:48.078,0:00:50.847 and between 2001 and 2014, 0:00:50.871,0:00:56.369 over 2,400 lives have been lost[br]in stampedes at these events. 0:00:57.362,0:00:59.219 What saddened me the most 0:00:59.243,0:01:03.758 is seeing people around me[br]resigning to the city's fate 0:01:03.782,0:01:07.607 in witnessing the seemingly[br]inevitable deaths of dozens 0:01:07.631,0:01:08.908 at every Kumbh Mela. 0:01:09.448,0:01:10.878 I sought to change this, 0:01:10.902,0:01:13.877 and I thought, why can't I try[br]to find a solution to this? 0:01:13.901,0:01:15.389 Because I knew it is wrong. 0:01:15.701,0:01:19.153 Having learned coding at an early age[br]and being a maker, 0:01:19.177,0:01:20.845 I considered the wild idea -- 0:01:20.869,0:01:22.057 (Laughter) 0:01:22.081,0:01:23.466 [Makers always find a way] 0:01:23.490,0:01:26.555 I considered the wild idea[br]of building a system 0:01:26.579,0:01:29.041 that would help regulate[br]the flow of people 0:01:29.065,0:01:32.054 and use it in the next Kumbh Mela in 2015, 0:01:32.078,0:01:35.683 to have fewer stampedes[br]and, hopefully, fewer deaths. 0:01:36.177,0:01:38.648 It seemed like a mission impossible, 0:01:38.672,0:01:40.440 a dream too big, 0:01:40.464,0:01:42.645 especially for a 15-year-old, 0:01:42.669,0:01:46.208 yet that dream came through in 2015, 0:01:46.232,0:01:49.925 when not only did we succeed 0:01:49.949,0:01:52.901 in reducing the stampedes[br]and their intensity, 0:01:52.925,0:01:55.102 but we marked 2015 0:01:55.126,0:01:59.133 as the first Nashik Kumbh Mela[br]to have zero stampedes. 0:01:59.596,0:02:04.934 (Applause) 0:02:04.958,0:02:07.444 It was the first time in recorded history 0:02:07.468,0:02:10.214 that this event[br]passed without any casualties. 0:02:10.888,0:02:12.038 How did we do it? 0:02:12.623,0:02:15.654 It all started when I joined[br]an innovation workshop 0:02:15.678,0:02:18.354 by MIT Media Lab in 2014 0:02:18.378,0:02:20.347 called the Kumbhathon 0:02:20.371,0:02:24.775 that aimed at solving challenges[br]faced at the grand scale of Kumbh Mela. 0:02:25.784,0:02:30.292 Now, we figured out to solve[br]the stampede problem, 0:02:30.316,0:02:32.480 we wanted to know only three things: 0:02:32.504,0:02:35.243 the number of people, the location, 0:02:35.267,0:02:38.726 and the rate of the flow[br]of people per minute. 0:02:39.241,0:02:43.370 So we started to look for technologies[br]that would help us get these three things. 0:02:44.091,0:02:48.345 Can we distribute radio-frequency tokens[br]to identify people? 0:02:48.369,0:02:51.782 We figured out that it would[br]be too expensive and impractical 0:02:51.806,0:02:53.742 to distribute 30 million tags. 0:02:54.545,0:02:57.926 Can you use CCTV cameras[br]with image-processing techniques? 0:02:57.950,0:02:59.997 Again, too expensive for that scale, 0:03:00.021,0:03:02.899 along with the disadvantages[br]of being non-portable 0:03:02.923,0:03:05.741 and being completely useless[br]in the case of rain, 0:03:05.765,0:03:08.147 which is a common thing[br]to happen in Kumbh Mela. 0:03:09.120,0:03:11.445 Can we use cell phone tower data? 0:03:11.806,0:03:14.273 It sounds like the perfect solution, 0:03:14.297,0:03:16.070 but the funny part is, 0:03:16.094,0:03:18.839 most of the people[br]do not carry cell phones 0:03:18.863,0:03:20.347 in events like Kumbh Mela. 0:03:20.887,0:03:24.601 Also, the data wouldn't have been[br]granular enough for us. 0:03:24.625,0:03:27.078 So we wanted something that was real-time, 0:03:27.102,0:03:30.419 low-cost, sturdy and waterproof, 0:03:30.443,0:03:32.854 and it was easy to get[br]the data for processing. 0:03:33.672,0:03:36.028 So we built Ashioto, 0:03:36.052,0:03:38.137 meaning "footstep" in Japanese, 0:03:38.161,0:03:42.346 as it consists of a portable mat[br]which has pressure sensors 0:03:42.370,0:03:44.861 which can count the number[br]of people walking on it, 0:03:44.885,0:03:47.317 and sends the data over the internet 0:03:47.341,0:03:49.888 to the advanced data analysis[br]software we created. 0:03:50.483,0:03:53.771 The possible errors,[br]like overcounting or double-stepping, 0:03:53.795,0:03:56.445 were overcome using design interventions. 0:03:56.803,0:04:00.329 The optimum breadth of the mat[br]was determined to be 18 inches, 0:04:00.353,0:04:02.592 after we tested many different versions 0:04:02.616,0:04:05.203 and observed the average[br]stride length of a person. 0:04:06.111,0:04:09.016 Otherwise, people might step[br]over the sensor. 0:04:09.040,0:04:11.612 We started with a proof of concept[br]built in three days, 0:04:11.636,0:04:14.510 made out of cardboard and aluminum foil. 0:04:14.534,0:04:15.535 (Laughter) 0:04:15.559,0:04:17.933 It worked, for real. 0:04:17.957,0:04:20.647 We built another one[br]with aluminum composite panels 0:04:20.671,0:04:22.309 and piezoelectric plates, 0:04:22.333,0:04:25.904 which are plates that generate a small[br]pulse of electricity under pressure. 0:04:26.959,0:04:30.228 We tested this at 30 different[br]pilots in public, 0:04:30.252,0:04:33.974 in crowded restaurants,[br]in malls, in temples, etc., 0:04:33.998,0:04:35.632 to see how people reacted. 0:04:36.720,0:04:38.743 And people let us run these pilots 0:04:38.767,0:04:45.243 because they were excited to see localites[br]work on problems for the city. 0:04:45.680,0:04:48.799 I was 15 and my team members[br]were in their early 20s. 0:04:49.736,0:04:54.227 When the sensors were colored, 0:04:54.251,0:04:57.092 people would get scared[br]and would ask us questions like, 0:04:57.116,0:04:59.323 "Will I get electrocuted[br]if I step on this?" 0:04:59.347,0:05:00.348 (Laughter) 0:05:00.372,0:05:04.744 Or, if it was very obvious that it was[br]an electronic sensor on the ground, 0:05:04.768,0:05:06.261 they would just jump over it. 0:05:06.285,0:05:07.361 (Laughter) 0:05:07.385,0:05:10.449 So we decided to design[br]a cover for the sensor 0:05:10.473,0:05:13.530 so that people don't have to worry[br]what it is on the ground. 0:05:13.554,0:05:15.530 So after some experimentation, 0:05:15.554,0:05:18.364 we decided to use an industrial sensor, 0:05:18.388,0:05:21.357 used as a safety trigger[br]in hazardous areas 0:05:21.381,0:05:22.532 as the sensor, 0:05:22.556,0:05:24.637 and a black neoprene rubber sheet 0:05:24.661,0:05:25.903 as the cover. 0:05:26.374,0:05:29.191 Now, another added benefit[br]of using black rubber 0:05:29.215,0:05:32.842 was that dust naturally[br]accumulates over the surface, 0:05:32.866,0:05:35.516 eventually camouflaging it[br]with the ground. 0:05:35.540,0:05:40.652 We also had to make sure that the sensor[br]is no higher than 12 millimeters. 0:05:41.206,0:05:43.435 Otherwise, people might trip over it, 0:05:43.459,0:05:45.364 which in itself would cause stampedes. 0:05:45.792,0:05:49.579 (Laughter) 0:05:49.603,0:05:50.965 We don't want that. 0:05:50.989,0:05:52.778 (Laughter) 0:05:52.802,0:05:56.881 So we were able to design a sensor[br]which was only 10 millimeters thick. 0:05:57.469,0:05:59.709 Now the data is sent[br]to the server in real time, 0:05:59.733,0:06:01.381 and a heat map is plotted, 0:06:01.405,0:06:04.748 taking into account[br]all the active devices on the ground. 0:06:04.772,0:06:09.057 The authorities could be alerted[br]if the crowd movement slowed down 0:06:09.081,0:06:12.216 or if the crowd density[br]moved beyond a desired threshold. 0:06:12.992,0:06:17.984 We installed five of these mats[br]in the Nashik Kumbh Mela 2015, 0:06:18.008,0:06:20.398 and counted over half a million people 0:06:20.422,0:06:22.104 in 18 hours, 0:06:22.128,0:06:26.505 ensuring that the data was available[br]in real time at various checkpoints, 0:06:26.529,0:06:28.405 ensuring a safe flow of people. 0:06:28.429,0:06:31.874 Now, this system, eventually,[br]with other innovations, 0:06:31.898,0:06:35.050 is what helped prevent stampedes[br]altogether at that festival. 0:06:35.585,0:06:38.458 The code used by Ashioto during Kumbh Mela 0:06:38.482,0:06:41.894 will soon be made publicly available,[br]free to use for anyone. 0:06:41.918,0:06:43.895 I would be glad if someone used this code 0:06:43.919,0:06:46.131 to make many more gatherings safer. 0:06:46.878,0:06:48.804 Having succeeded at Kumbh Mela 0:06:48.828,0:06:52.740 has inspired me to help others[br]who may also suffer from stampedes. 0:06:52.764,0:06:55.573 The design of the system[br]makes it adaptable 0:06:55.597,0:06:57.591 to pretty much any event 0:06:57.615,0:07:01.411 that involves an organized[br]gathering of people, 0:07:01.435,0:07:05.745 and my new dream is to improve,[br]adapt, and deploy the system 0:07:05.769,0:07:11.238 all over the world to prevent loss of life[br]and ensure a safe flow of people, 0:07:11.262,0:07:14.049 because every human soul is precious, 0:07:14.073,0:07:16.451 whether at concerts or sporting events, 0:07:16.475,0:07:18.362 the Maha Kumbh Mela in Allahabad, 0:07:18.386,0:07:19.837 the Hajj in Mecca, 0:07:19.861,0:07:21.758 the Shia procession to Karbala, 0:07:21.782,0:07:23.266 or at the Vatican City. 0:07:23.893,0:07:25.770 So what do you all think, can we do it? 0:07:25.794,0:07:26.945 (Audience) Yes! 0:07:26.969,0:07:28.120 Thank you. 0:07:28.144,0:07:29.145 (Cheers) 0:07:29.169,0:07:32.439 (Applause)