this report from our transport correspondent, tom burridge, contains some flashing images. not a comfortable wait at heathrow. nearly 100 flights cancelled in a testing 2a hours. our first flight from manchester to heathrow was delayed by three hours, so we missed our connecting flight to rio and we can't fly out to rio until tomorrow. with stormy weather last night, passengers boarded flights which didn't take off and some like jess sims headed to another airport. they have our bag, so we are off to the wedding with nothing to wear. but hopefully we'll be able to find something at gatwick. but at gatwick and other airports in the south and midlands, weather problems made worse by a damaged radar at air traffic control. since we got to naples airport last night, we had absolutely no information. no help, no staff anywhere. hopefully, after last night's escapades which was basically after waiting nearly five hours, about to board the plane and being told all the food was spoiled and we weren't allowed on the plane. tonight, things at heathrow are getting back to normal. but at europe's busiest airport, virtually every take—off and landing slot is full. so a fairly minor issue can cause significant disruption. loudspeaker: we would like to apologise for any inconvenience this may cause you. 0n the railways, the chaos of last night calmed. but with overhead cable still damaged, few trains on the midlands mainline and tricky journeys. i'm headed to harrogate, north yorkshire. i should have gone from sandy in bedfordshire up through york, it should have taken me an hour and a half or two hours. it's going to take me... well, i don't know, five or six hours according to this. services cancelled too on the eurostar today — not the ideal start to her weekend. in around ten minutes, it's the film review here on bbc news. but first, here's click. it's the hardest problem in technology and one that could change everything. this week we're in arizona, the self driving capital of the world, to ride in the latest robotic vehicles, meet the people who are not happy to see them... ..and to find out what happens when the tech goes wrong. when california told uber that there were going to be some new regulations that they needed to adhere to, governor ducey, in a public announcement, said... california may not want you, we want you to know that arizona does. we are a state that is open for business. we are a state that welcomes business and new people and technology. 0ur governor, governor ducey, had basically opened wide the arms of our state to welcome them there and it was a no—brainer. chandler is a hub where autonomous vehicles are growing and having more miles mapped on our roads than any place else in the galaxy. it's kind of great to be the centre of that. well, chandler, i believe, was chosen because it has very wide streets, very clean streets, they're on a grid, and the people would be very accepting of this kind of technology. and so it was that the technology that will one day change our society, our landscape, and our lives, found a home in arizona. the makers of self driving cars have flocked to the town of chandler, which has fast become the industry's testbed. today, i'm taking a ride in one of the most prolific types of self driving cars here, made by google subsidiary waymo. now, self driving cars come kitted with loads of sensors so they can see in every direction, and sometimes in ways that we can't with our eyes. on top we've got loads of normal cameras looking in every direction. and the fact that there's loads of them means that they can judge distances by seeing how different objects move in relation to each other. now, there's also radar, four of those, one on each corner, and these spinning things, these are the interesting things, these are lidar sensors. there are five round and a big one in the black bump on the top, which can see three football pitches ahead and behind. right, let's go for a ride. 0oh, we have our safety driver. 0k. and away we go. computer: please remember to buckle your seatbelt. we've just pulled out in front of quite a fast moving car there. we made it. i call that quite a human manoeuvre. companies like waymo are experimenting whether their technology can avoid hitting you. i mean, that's the experiment. when you go out on the streets, when you cross on the crosstalk, and there's the waymo, waymo is actively testing whether or not its cars can avoid an accident, and avoid an accident with you if you happen to be on the roads. and some people also are clearly offended by that notion. and, last year, the fears of the community became a reality. a self driving uber vehicle failed to detect her crossing an empty road at night and the safety driver failed to hit the brakes. it was the first case of a pedestrian being killed by a self driving car. the uber was, the vehicle was a volvo again. it was a self driving vehicle. it was in the autonomous made at the time. and our investigation did not show at this time that there were significant signs of the vehicle slowing down. the uber vehicle hit elaine herzberg at 30 mph. —— 38 mph. this was a huge moment for the burgeoning industry, which led to uber having to immediately halt their self driving programme. so what exactly happened and whose fault was it? we went to the site of the crash in tempe, arizona, with the news editor of the phoenix new times, ray stern, to find out more about the incident. ok, it's on the other side. she took her bike from this area, walked it across this lane, and then entered this lane. the uber vehicle was in this lane. and itjust kept staying in this lane even though the pedestrian is here. it should have swerved. it had time and place to swerve, but it didn't. so before she made it to the sidewalk it impacted her. i absolutely would have seen elaine as she started to cross the road. and i would have braked for her. most reasonable drivers would have. and, in fact, any driver who was paying attention would not have hit elaine herzberg. in order to entice uber and other companies into arizona, governor ducey relaxed regulations, which meant companies faced no requirement to disclose anything about their programmes, including crashes. basically the governor invited uber in. that was one problem. they were operating here without any real transparency in terms of what they were actually doing, when the vehicles were in autonomous mode, what their criteria were for it. and so the vehicles were doing whatever they wanted. and uber had free reign. tempe police called the crash "entirely avoidable" after investigations found that the safety driver was watching television on her phone at the time of the fatal incident. ms vasquez could still face charges of vehicular manslaughter. she looked down, they estimate, 160 times during the circuit that she was doing. the evidence showed that she was streaming the view, which is a tv show, on her phone at the exact time of the impact. so what exactly went wrong with uber‘s self driving technology on that night? it can't really be to do with poor visibility, can it? one of the messed up things about the whole incident has been the video released by uber after the accident. and if you've seen this video, it looks like this street is very dark and then at last second the woman on the bike suddenly pops out of the darkness — right before the impact. in fact, this area is not as dark as this video shows. this drive—through at night follows the same route as the uber vehicle. it shows that the street lighting makes the road clearly visible far into the distance. the new york times reported that uber were not living up to expectations before the crash. as of march 2018, uber was struggling to meet the targets of 13 miles per intervention in arizona. as a comparison, gm—owned cruise reported to california regulators that they went more than 1200 miles per intervention and waymo said that their california test cars went an average of nearly 5600 miles before driver intervention. reports said that the uber vehicle actually detected elaine herzberg six seconds before the crash, but the perception system got confused, classifying her as first an unknown object, then as a vehicle, and finally as a bicycle. those volvos came from the factory with an accident avoidance system, one of these new semiautonomous systems that a lot of the new cars have. 1.3 seconds before impact, the self driving system realised emergency braking was needed. however, uber had disabled the emergency braking system on the volvo to prevent conflict with the self driving system. nevertheless, prosecutors have determined that uber were not criminally liable in the death. if uber hadn't disabled the technology that potentially the vehicle would have detected the pedestrian even without the uber autonomous technology, just with the volvo technology and stopped the vehicle. but uber disconnected that because apparently the vehicle was being a little too jerky in its motions, and it didn'tjibe correctly with the autonomous vehicle system that uber had in there. a safety driver supervising an imperfect system should ensure its overall safety. however, that only works if they're paying attention. with self driving cars being tested live on busy streets, accidents are inevitable. so this may not be the last incident that we see on the road to a driverless future. but the number of accidents involving self driving cars is very low for the millions of miles of testing that have taken place. while the advancements we've seen in the last few years are more than impressive, getting a computer to fully understand the real world, and drive safely through it, will be a monumental achievement. they're not saying it's done, because it's not. this not a solved problem, it is a hard problem, it is many years before you can buy a car that has no steering wheel and you can say, "i'll have the car with no windscreen," and it has the same functionality as your car now. to start with, they will have subhuman capacity and superhuman capacity in other things. subhuman in their ability to reason, about all the extraordinary things that can happen on a road that has nothing to do with driving. superhuman in their ability to concentrate and never ever, ever get distracted. to see in ways that humans don't see with radar and laser, to sense distance, extraordinary things and above all, the ability for these vehicles to share and acquire competencies, not because of their own experience but because of the experience of all the other vehicles everywhere else in the world, that's an extraordinary thing and that is the compelling reason why these vehicles are coming. they will be better than us, because there is nothing in our evolutionary history that makes us good at controlling 1.5 tons of metal at 70 miles an hour. here in arizona, i've seen the benefits of — and the resistance to — the idea of the machines taking over another part of our lives. so i think the question is not if or when this will happen, but will we let it? hello and welcome to the film review on bbc news. to take us through this week's cinema releases, jason solomons. so, jason, what do we have this week?