WEBVTT 1 00:00:00.010 --> 00:00:03.460 Reports are just coming in of an explosion 2 00:00:03.460 --> 00:00:07.560 at Liverpool Street Station here in London. 3 00:00:07.560 --> 00:00:12.070 The blast that ripped through Madrid's early 4 00:00:12.070 --> 00:00:14.070 morning commuter train sent blood and metal flying. 5 00:00:14.070 --> 00:00:17.070 Ambulances rushed more than a thousand wounded to city 6 00:00:17.070 --> 00:00:20.530 hospitals and the Government set up centers for 7 00:00:20.530 --> 00:00:23.090 people to identify the bodies of missing love ones. 8 00:00:23.090 --> 00:00:25.970 This horror struck again when a bomb exploded on a rush 9 00:00:25.970 --> 00:00:28.850 hour subway train near the European Union Headquarters. 10 00:00:28.850 --> 00:00:32.880 [Music] 11 00:00:32.880 --> 00:00:37.100 Since 2001 more than 1,900 attacks have 12 00:00:37.100 --> 00:00:42.100 been carried out against public transportation systems globally 13 00:00:42.240 --> 00:00:45.750 resulting in 4,000 deaths, 14,000 injuries. 14 00:00:45.750 --> 00:00:49.260 The attacks on metro stations, on rails 15 00:00:49.260 --> 00:00:54.260 I know what would happen God forbid if what we see happening in European nation, after European nation 16 00:00:54.590 --> 00:00:57.560 were to happen here in the United States. 17 00:00:57.560 --> 00:01:00.730 I don't want to be able to have people say I told you so. 18 00:01:00.730 --> 00:01:03.760 We need to get the job done and protect surface transportation. 19 00:01:03.760 --> 00:01:08.760 We are going to start seeing more attacks or attempted 20 00:01:09.400 --> 00:01:12.530 attacks on mass transit rail systems. 21 00:01:12.530 --> 00:01:15.810 It's too attractive a target for terrorist, because 22 00:01:15.810 --> 00:01:19.090 it's an opportunity to inflict a lot of casualties in 23 00:01:19.090 --> 00:01:22.380 spaces where people tend to gather in dense crowds and 24 00:01:22.380 --> 00:01:25.700 it's also an opportunity to inflict damage on critical infrastructure. 25 00:01:25.700 --> 00:01:29.770 If you look at the security posture that we have in place now, it's not sufficient. 26 00:01:29.770 --> 00:01:33.960 Because we are not performing the sort of detailed 27 00:01:33.960 --> 00:01:36.340 screening that takes place in aviation environments. 28 00:01:36.340 --> 00:01:39.400 Mass transit security is a global problem. 29 00:01:39.400 --> 00:01:43.460 We are no less vulnerable here than they are overseas. 30 00:01:43.460 --> 00:01:46.020 These are environments where you have a very high level of foot traffic. 31 00:01:46.020 --> 00:01:48.580 You have an unstructured crowd with many 32 00:01:48.580 --> 00:01:51.720 points of entry that are scattered over a large area. 33 00:01:51.720 --> 00:01:55.000 You have an environment that can readily bring in bags or heavy coats, many things 34 00:01:55.000 --> 00:01:58.150 to conceal a potentially dangerous item. 35 00:01:58.150 --> 00:02:01.290 This highlights the importance of developing tools 36 00:02:01.290 --> 00:02:04.400 that can assist the security staff on site, 37 00:02:04.400 --> 00:02:07.620 to ensure our systems stay as secure as possible. 38 00:02:07.620 --> 00:02:10.650 Our task is clear, we need a way to safely detect 39 00:02:10.650 --> 00:02:13.680 explosive threats on people and in their bags. 40 00:02:13.680 --> 00:02:15.750 and we have to do this without physically searching 41 00:02:15.750 --> 00:02:17.820 them or impacting their flow through the transit system. 42 00:02:17.820 --> 00:02:21.960 Currently there are no solutions to solve this challenging 43 00:02:21.960 --> 00:02:23.490 problem and this is why the Department of 44 00:02:23.490 --> 00:02:26.230 Homeland Security Science and Technology Directorate 45 00:02:26.230 --> 00:02:30.340 has a program to mitigate the explosive threat to surface transportation. 46 00:02:30.340 --> 00:02:34.590 The surface transportation explosive threat detection program is working in collaboration with 47 00:02:34.590 --> 00:02:38.620 the surface transportation system end users and our stakeholders, as well as our partners 48 00:02:38.620 --> 00:02:42.670 at John Hopkin's applied physics laboratory and 49 00:02:42.670 --> 00:02:44.745 MIT Lincoln labs to close the surface transportation 50 00:02:44.745 --> 00:02:46.820 security gap through innovative technologies 51 00:02:46.820 --> 00:02:49.040 that are designed to work together and provide a 52 00:02:49.040 --> 00:02:52.060 layered approach to security throughout the transportation system. 53 00:02:52.060 --> 00:02:55.080 Basically we are taking a curb to platform approach to security. 54 00:02:55.080 --> 00:02:59.150 Curb to platform sensing is the idea that we are 55 00:02:59.150 --> 00:03:01.250 going to distribute sensors at various points from the 56 00:03:01.250 --> 00:03:03.350 parameter of the facility to the platform. 57 00:03:03.350 --> 00:03:07.590 So from the time you step out of the cab to the time you 58 00:03:07.590 --> 00:03:10.770 get on a train or a bus, there are sensors that are 59 00:03:10.770 --> 00:03:13.800 keeping track of subjects in the system. 60 00:03:13.800 --> 00:03:15.830 This does a few things for us. 61 00:03:15.830 --> 00:03:20.220 First, it allow us to get multiple looks of a subject for better detection. 62 00:03:20.220 --> 00:03:24.230 Second, it makes it more difficult for a subject to evade the sensing systems. 63 00:03:24.230 --> 00:03:28.250 If a subject knew that all of our sensors were say only 64 00:03:28.250 --> 00:03:30.290 at the front door of a facility than he knows, he only 65 00:03:30.290 --> 00:03:33.320 needs to defeat the sensors at that spot. 66 00:03:33.320 --> 00:03:36.700 And finally, it recognizes that there are many 67 00:03:36.700 --> 00:03:40.870 security environments that we must protect in a mass transit system. 68 00:03:40.870 --> 00:03:45.070 Here at S&T were developing a number of tools, among them a video algorithm capable 69 00:03:45.070 --> 00:03:49.120 of automatically detecting bags left behind and tagging individuals who left those bags. 70 00:03:49.120 --> 00:03:53.150 Were also developing a suite of video forensic tools that allow 71 00:03:53.150 --> 00:03:57.220 video surveillance systems to work more effectively and efficiently. 72 00:03:57.220 --> 00:04:01.410 Today we are at New Carrollton Metro Station, 73 00:04:01.410 --> 00:04:04.460 we are testing some of our camera algorithms by 74 00:04:04.460 --> 00:04:08.470 placing bags in a variety of locations. 75 00:04:08.470 --> 00:04:13.470 If you have a fairly well commercial algorithm that alerts maybe only 60 times 76 00:04:13.860 --> 00:04:17.930 a hour, once a minute. When you multiply that by the 50,000 or so 77 00:04:17.930 --> 00:04:22.100 video camera in a mass transit system the number of false alerts is actually quite overwhelming. 78 00:04:22.100 --> 00:04:26.160 It's inconceivable that even a fully staffed operation 79 00:04:26.160 --> 00:04:30.170 center is going to be able to field that many events over a course of a day. 80 00:04:30.170 --> 00:04:32.190 We are looking at a false alert rate that's appreciably 81 00:04:32.190 --> 00:04:34.210 lower than whats commercially available 82 00:04:34.210 --> 00:04:38.540 and basically baked into the infrastructure and the video 83 00:04:38.540 --> 00:04:42.880 analytics and the video management systems that presently exist. 84 00:04:42.880 --> 00:04:46.010 The FOVEA program allows us to be able 85 00:04:46.010 --> 00:04:50.160 to identify a package that's been left behind and then 86 00:04:50.160 --> 00:04:53.410 to figure out who left it behind and then 87 00:04:53.410 --> 00:04:56.570 start to track that person and to determine 88 00:04:56.570 --> 00:04:59.080 whether or not this is a threatful situation and 89 00:04:59.080 --> 00:05:03.350 then to deploy proper resources to keep the people safe. 90 00:05:03.350 --> 00:05:07.720 So we are helping make existing video surveillance 91 00:05:07.720 --> 00:05:11.850 systems more efficient, more effective by giving users tools to help them get through video faster. 92 00:05:11.850 --> 00:05:15.910 One of the tools within FOVEA is a tool that we call "Jump Back," and it let's 93 00:05:15.910 --> 00:05:20.170 a user highlight an abandoned object, simply draw a box around it and Jump Back 94 00:05:20.170 --> 00:05:24.200 will go back to when that object first appeared. And from there the user can investigate 95 00:05:24.200 --> 00:05:28.250 what are the circumstances around it. So for instance it's jumped back 96 00:05:28.250 --> 00:05:32.450 to when this person left the bag. We can actually begin to 97 00:05:32.450 --> 00:05:36.510 bookmark the person in the video and follow them as they move throughout the station. 98 00:05:36.510 --> 00:05:40.510 So here we are actually piecing together information from different video cameras. 99 00:05:40.510 --> 00:05:44.730 Now once a user actually follows the person throughout 100 00:05:44.730 --> 00:05:48.780 the station and understands where they come from and where they have gone to 101 00:05:48.780 --> 00:05:52.930 they can then reconstruct that video and stitch all of those pieces together 102 00:05:52.930 --> 00:05:55.000 into one final video. 103 00:05:55.000 --> 00:05:58.120 So one task that often takes a really long, 104 00:05:58.120 --> 00:06:01.240 long time is just simply reviewing videos. 105 00:06:01.240 --> 00:06:05.640 So imagine you need to review 12 hours of video from the night before. You can watch it and fast forward, 106 00:06:05.640 --> 00:06:09.820 but we found that you can only speed that up by a factor of 10 before things start flying by too fast. 107 00:06:09.820 --> 00:06:12.080 So we developed a tool called "Summarization." 108 00:06:12.080 --> 00:06:17.080 The video Summarization functionality in FOVEA is a very important tool for us. 109 00:06:18.090 --> 00:06:22.230 What allows us to scan hours on hours of video in just a short period of time. 110 00:06:22.230 --> 00:06:26.480 We had a incident here at Amtrack where it took us 111 00:06:26.480 --> 00:06:30.620 two full man days to identify exactly what happened. When we ran it through the tool, was able to do 112 00:06:30.620 --> 00:06:34.710 it in just a matter of hours. FOVEA acts as a force multiplier. 113 00:06:34.710 --> 00:06:38.990 The analytics is just a simple, great way to leverage 114 00:06:38.990 --> 00:06:43.100 the infrastructure and the man power to get the 115 00:06:43.100 --> 00:06:45.220 security that we require and FOVEA allows me to do that. 116 00:06:45.220 --> 00:06:47.260 We're also investing in stand-off threat detection 117 00:06:47.260 --> 00:06:49.400 systems using complimentary parts of 118 00:06:49.400 --> 00:06:53.640 the electromagnetic spectrum. These systems will serve as 119 00:06:53.640 --> 00:06:56.940 a key component of the layered screening concept. 120 00:06:56.940 --> 00:07:01.940 This is the Centimeter Wave Sensor that we are developing here at APL as apart of our 121 00:07:02.180 --> 00:07:06.520 layered threat detection approach. These would be incorporated 122 00:07:06.520 --> 00:07:10.520 into the walls, you'd be going down the hallway. People can just walk in front of 123 00:07:10.520 --> 00:07:14.530 of this and this images them, but at such a 124 00:07:14.530 --> 00:07:18.560 blurry level, that most people would not consider this to be particularly invasive. 125 00:07:18.560 --> 00:07:22.570 It's only the very large things. That you know, "Hey that looks like a 126 00:07:22.570 --> 00:07:26.650 rifle, that looks like an explosive." So then an operator 127 00:07:26.650 --> 00:07:30.700 will see the people walking by the array and the large 128 00:07:30.700 --> 00:07:35.060 false colored, red blob that says, "he is concealing something." 129 00:07:35.060 --> 00:07:39.110 The power and the frequencies we're mitting are the 130 00:07:39.110 --> 00:07:43.220 same as put out by your Wi-Fi. This will not harm you in 131 00:07:43.220 --> 00:07:47.580 any way, these are lower than exposures that you would get from your own cell phone. 132 00:07:47.580 --> 00:07:51.680 Our concept is you use a low cost system like a Centimeter Wave system as the 133 00:07:51.680 --> 00:07:55.880 initial layer of defense, identifying potential items of interest. 134 00:07:55.880 --> 00:08:00.270 Then we can use the Millimeter Wave system as our next layer, providing a more detailed view 135 00:08:00.270 --> 00:08:04.360 of potential threat items. Both layers working in a safe way 136 00:08:04.360 --> 00:08:07.660 that maintains the traveling publics privacy. 137 00:08:07.660 --> 00:08:10.660 This system is a prototype stand off of microwave imager. 138 00:08:10.660 --> 00:08:14.720 Its a panel that's placed on a wall or in a corridor or perhaps on the ceiling. 139 00:08:14.720 --> 00:08:18.900 The idea here is if you come in proximity of the unit with 140 00:08:18.900 --> 00:08:23.080 a concealed bomb, a rifle, it's going to be able to see it. 141 00:08:23.080 --> 00:08:27.110 Microwave sensing can compliment FOVEA by proving information that FOVEA doesn't have. 142 00:08:27.110 --> 00:08:31.510 So you can image a scenario, where FOVEA alerts us to 143 00:08:31.510 --> 00:08:35.510 the fact that somebodies dropped a bag and walked away. 144 00:08:35.510 --> 00:08:39.840 That's all we know. We don't now if the bags empty or if we need to call in a bomb 145 00:08:39.840 --> 00:08:43.990 tech to check things out. Microwave sensing can tell us if the bags empty 146 00:08:43.990 --> 00:08:48.130 or if its full of metal or perhaps if it has some other suspicious item inside. 147 00:08:48.130 --> 00:08:52.160 That will allow us to tailor our response to save resources, 148 00:08:52.160 --> 00:08:56.410 save time and keep everybody safe. 149 00:08:56.410 --> 00:09:00.460 A passenger enters a subway system, there's a microwave scanner that's able to detect 150 00:09:00.460 --> 00:09:04.470 something suspicious inside that persons bag. 151 00:09:04.470 --> 00:09:07.760 The system then will queue through video surveillance an 152 00:09:07.760 --> 00:09:12.760 Italian track process where that person is tracked through multiple cameras in order to gather additional observations 153 00:09:12.850 --> 00:09:17.020 through other scanners then all that information is assembled and entered into 154 00:09:17.020 --> 00:09:21.200 a model that can decide whether there's enough evidence to queue a security personnel to intervene. 155 00:09:21.200 --> 00:09:25.490 So the layer sensing architecture would be a system that's constantly 156 00:09:25.490 --> 00:09:29.540 updating the risk level that it observes. So from the moment somebody steps into 157 00:09:29.540 --> 00:09:33.730 a facility and starts collecting observations, it's going to have some assessment of what 158 00:09:33.730 --> 00:09:38.150 it thinks the risk level is. But as that person proceeds through the subway system, 159 00:09:38.150 --> 00:09:41.200 additional observations are collected. 160 00:09:41.200 --> 00:09:46.200 The system is going to update that risk level. Because we can design panels in different shapes 161 00:09:46.310 --> 00:09:50.330 and sizes, that gives us the opportunity to place them on walls near the entry way or near the subway platforms. 162 00:09:50.330 --> 00:09:54.430 We can place it on poles, in some cases on ceilings. 163 00:09:54.430 --> 00:09:58.450 We can even potentially integrate it into the fare gates that people walk through as they enter the system. 164 00:09:58.450 --> 00:10:02.670 Were developing this technology for use in 165 00:10:02.670 --> 00:10:06.730 the surface transportation environment, because it's our most difficult security challenge. 166 00:10:06.730 --> 00:10:11.030 However we'll also have applications to other parts of the Homeland Security critical 167 00:10:11.030 --> 00:10:15.070 infrastructure and can provide enhanced security at airports at bus and ferry terminals. 168 00:10:15.070 --> 00:10:19.270 It can also be used to monitor events that take place at 169 00:10:19.270 --> 00:10:22.440 stadiums, convention centers, schools and to enhance 170 00:10:22.440 --> 00:10:25.680 security where we have unstructured crowds. 171 00:10:25.680 --> 00:10:30.680 The mass transit environment presents daunting security challenges and we have very well-trained and dedicated 172 00:10:31.730 --> 00:10:35.860 law enforcement and security staff on-site, however 173 00:10:35.860 --> 00:10:40.090 they can't be at all places at all times and that's why it's 174 00:10:40.090 --> 00:10:40.340 critical that we develop tools like this, that can act as a 175 00:10:40.340 --> 00:10:44.430 critical that we develop tools like this, that can act as a 176 00:10:44.430 --> 00:10:48.630 force multiplier and queue suspicious situations and help them ensure security in our mass transit facilities. 177 00:10:48.630 --> 00:10:52.710 The reduction in time is what I'm looking forward to. 178 00:10:52.710 --> 00:10:56.950 What used to take us 20 hours, takes us 10 minutes 179 00:10:56.950 --> 00:11:01.360 and then your getting real time information. So imagine that a Officer is a scene, they want 180 00:11:01.360 --> 00:11:05.540 to know who they should be looking for. This system allows us to tell them that almost instantaneously. 181 00:11:05.540 --> 00:11:09.670 Amtrack has a very good relationship with S&T. 182 00:11:09.670 --> 00:11:13.700 We volunteer to be a test bed for the video surveillance analytics technology their testing 183 00:11:13.700 --> 00:11:18.000 and we're very proud to do it. We're very satisfied and happy 184 00:11:18.000 --> 00:11:22.030 with the performance of it so far and we will be rolling this technology out to our larger 185 00:11:22.030 --> 00:11:26.160 stations for a more detailed pilot, here shortly in the future. 186 00:11:26.160 --> 00:11:30.370 This operational environment is by far the most challenging from an explosive threat detection 187 00:11:30.370 --> 00:11:34.430 standpoint and a focused long-term investment as well as a coordinated 188 00:11:34.430 --> 00:11:38.440 program working with TSA and our surface transportation stakeholders, 189 00:11:38.440 --> 00:11:42.680 will allow us to continue to make the breakthroughs we need in achieving our 190 00:11:42.680 --> 00:11:46.750 goal of better protecting the nations surface transportation systems and the travelling public. 191 00:11:46.750 --> 00:11:50.810 The treat isn't gone, people are plotting right now! 192 00:11:50.810 --> 00:11:55.160 Right now against this country. 193 00:11:55.160 --> 00:12:00.160 [Music]