to the next round of the australian open, but there are concerns about his fitness. he needed medical treatment in the second round with what appeared to be a problem with his lower back or leg. he won in straight sets, but was not at his best. this is his first event since wimbledon. his next opponent is chung hyeon, who's become the first south korean man to reach the last 16 in melbourne. he knocked out the fourth seed alexander zverev in five sets, to get there. defending champion roger federer is also into the last 16, beating richard gasquet in straight sets. the world number two and five—time winner in melbourne has yet to drop a set so far in this year's tournament. it is looking ominous. the match between two former winners was rather one—sided, with angelique kerber knocking out maria sharapova. kerber won in straight sets and she's now the only grand slam champion left in the women's draw. there was huge relief for world number one simona halep, who survived an epic battle, with the unseeded american lauren davis, to reach the fourth round.

she saved three match points, before eventually winning, 15—13, in the decider — that set alone lasted two hours and 22 minutes, they were on court for nearly four hours. ijust kept i just kept believing ijust kept believing that it's not over, even if she is leading a little bit more. i gave everything i had today. actually, i'm really proud that i could stay there and win it. it was not easy at all, she played great. but the men's doubles champions of two years ago, are out, jamie murray and bruno soares, lost in three sets, to the indian pair of leander paes, and purav raja. but britain's dom inglot, is through. he and new zealand's marcus daniell, beat the french duo, benoit paire, and hugo nys. the day's football action is under way and we've already had a goal in the first of the scottish cup fourth

round matches. aberdeen up against st mirren thanks to the penalty early on from adam rooney. that is i7 early on from adam rooney. that is 17 minutes old now, that match. chelsea's new signing ross barkley is on the bench for today's game at brighton — it's the early fixture, and they're just kicking off there. obviously goalless. six seconds gone. english golfer matt fitzpatrick shot a stunning round at the abu dhabi championship, to take the clubhouse lead. he sank io birdies in a round of 63. if it weren't for one solitary bogey, he'd have equalled the course record. he's is—under par, alongside rory mcilroy, two shots behind leader thomas pieters, with ross fisher in between. that's all the sport for now. it's now time for click. this week, cars, bars and a police ride. siren.

i'm on my way to a reported incident, on one of las vegas‘s busiest highways. on your channel, 76, accident on rainbow and 95. with the last rain falling overfour months ago, the oily roads mixed with fresh water have become a lethal recipe for disaster. in the driving seat is sergeantjuan arias from nevada highway patrol. he's using waycare, software that alerts him to an incident as soon as it soon as it's reported by someone calling 911 as it's reported by someone calling 911 or through driving apps like waze and it provide him with details and the best route to get to the scene. it'll tell me the location, what kind of accident, if it's debris,

how long it's going and if there's any responders that are assigned to the call that are on their way. it constantly updates him on the situation, as it develops. having a robust system in place doesn'tjust help with weather—related collisions. with our route 91 shooting that we had at mandalay, for the portion that we handled, int the highway patrol, is really getting the public off the highway strip as quickly as possible, closing off the freeways, so we could have the critical resources, fire, medical, ambulances, to get people to the hospital and get there quickly. in 2017, 15,000 crashes were attended to, with over 300 people dying on average in each year in road accidents in nevada. getting emergency services to the scene as quickly as possible is critical. camera 17. the system has been running through the regional transportation

commission's traffic management centre for the past three months. now, because we're getting information in through so many different data streams, not just the dispatchers, but we're getting it through social media. we're getting it through things like the waze apps. people are tagging them in, as they're driving. because all of this is happening so quickly, we might have already sent out all of that information and had everybody in this room aware before the first 911 call comes in. so we're talking about possibly ten to 15 minutes of improvement in response time for some of these incidents. that's major, when you're dealing with traffic incidents. waycare pulls in data from various sources, traffic signals, in—vehicle sensors, tv cameras, and information from driving apps. it factors in things like what day of the year it is, the time of day and the weather. responding to incidents rapidly is one thing, but the point is to be able to predict incidents,

before they happen so the responders can be better prepared and in the right location. using deep learning, what we do is we look at the historical data, run it through algorithms to develop patterns that are merging. and tie it to what's happening now on the road. and by doing that, we can look forward in time, to identify where these incidents are likely to occur. unfortunately, waycare wasn't able to predict this one. you can see the trooper‘s taking pictures. it looks like it's the rear. you see how she was spinning out. she did a full 180 and struck right here. being able to foresee accidents here could really save lives. the hope is that as the data gets more sophisticated, the predictions will become more accurate. every day, we get more and more evidence about what causes, what triggers an incident and the artificial learning gets smarter and smarter and more capable. for nevada now, though, the initial results are promising. they get there faster,

we clear it faster and that means less secondary accidents. and if you think about it, secondary accidents have basically 18% of secondary accidents are fatalities. so, we're reducing the fatalities on the road. and, of course, the goal is to prevent accidents altogether and richard taylor and lara lewington have been looking at some in—car technologies that may help make that a reality. at ces, as you might expect, there's a lot of interest in self—driving cars and it's pretty clear that we're on a one—way street towards full autonomy. but that does still seem to be a way off, although we don't know exactly how far. in the meantime, though, there is plenty of innovation to be seen, before we reach our final destination. unsurprisingly, the move towards autonomated driving is focused largely on safety, with hyundai creating a system to intervene

when we need it the most. with a combination of biometric sensors in the seat, they're tracking heart rate and a low—resolution camera, which is tracking yourfacial movements. the reason it's low resolution, is so that the refresh rate is a lot quicker. so, if there's a problem, if you've lost concentration or you're drifting off to sleep, then the car can quickly react. to autonomously be moved off the road to a safe spot. and the basic premise of this technology could be available in just a year. meanwhile, nissan has a different, even more futuristic twist on biometrics using my grey matter. the idea of this system is really to provide an interaction between man and machine, between my brain and the ai. the concept here with nissan is that even in a world of autonomous vehicles, there will be roles for humans to play.

after all, a lot of people do find driving quite a positive experience. it can interpret the signals coming from the human and actually enhance the ride. this so—called brain—to—vehicle tech currently involves wearing this bizarre—looking electrode—studded helmet to capture my brain activity and interpret the signals as much as half a second before my muscles do. so, as i'm about to, say, change lane or hit the brakes, it'll initiate the action for me, giving me a smoother ride, and yet still allowing me a sense of control. they do need to sort out that helmet, though. laughter. oh dear, i'm not driving very well here! yet what we can't hide away from is the fact that when full autonomy does come to pass, it's not simply about cars. this is yamaha's concept motorbike. a self—driving racing vehicle that should be able to do speeds of over 120 mph, although not on actual roads, you'd hope.

but whatever the form of autonomous vehicle, it'll need to interact safely with pedestrians and cyclists too, a challenge that ford are hoping to overcome in their vehicles. initially, cyclists will have to be seen by the vehicles and we are building perception into our autonomous vehicle, that allows it to detect the cyclists, objects, understand their intent and to ensure we can be safely navigating in same space. and ford are just one of the big brands that have called on the help of nvidia, whose processes, combined with intelligence software, can make the environment around the vehicles safer. for example, using lidar sensors to alert the driver who is about to open a car door onto a cyclist. and ai is fuelling other experiences inside the car, too. speech—recognition specialists nuance power many of today's in—car interactions and they showed off how they'll looking in the future, as well.

today, we think about the assistant as something we interact with using voice, but we can add other modalities. of course, we have the screen, we have touch, but maybe we can use gestures and in this specific prototype, we introduced eye tracking. to help the assistant understand, what am i, as a driver, looking at and then i can ask questions about my environment. if i see a coffee shop in front of me, i canjust ask a question about it. what is the user rating of this copy shop? of this coffee shop? car: starbucks coffee has a user rating of three stars. so, the other part of the system is that there are microphones placed in different parts of the car, which means the ai can respond, according to where the different passengers are. so, here on the passenger seat i can say, "hello, dragon, i'm cold". car: 0k, raising the temperature in zone two to 71.0 degrees... there's definitely a trend towards making ourjourneys more

enjoyable, as well as safer. toyota have even updated their happiness—tracking concept car, aiming for a more pleasurable journey and even suggesting where you might want to go. for anyone who needs their car to tell them. car: since you are a foodie, i will tell you something interesting. there are many options around union square, from casual dining to michelin stars. high—end restaurants as well as popular cafes. do you like it? yes. sounds good. that was a bit of fun, but i didn't need the al to tell me that i was ready for dinner. sushi? yeah. let's go. welcome to the week in tech. it was the week that ford announced it would invest £8 billion in electric cars, in the next five years. a flaw in a vr porn app called sin vr left 20,000 users' names and email addresses exposed. and hackers managed to steal $100,000 worth of cryptocurrency by hijacking a server that hosts a web—based wallet

for virtual currency. it was a busy week for cryptocurrency, as bitcoin encountered its biggest daily crash in four months. it's thought that fears over increased regulation, particularly in asia, were behind the downturn. a contraceptive app previously heralded as effective as the pill has been criticised by a swedish hospital for a number of unintended pregnancies they say were linked to the app. the company behind the app natural cycles have defended the product, saying that as with any form of contraception it isn't100% effective. they are now launching an internal investigation, however. and i bet you didn't expect the latest nintendo offering to include, well, a whole lot of cardboard. the latest add—ons for the switch console are cardboard packs, turning the controllers into a fishing rod, a motorbike and even a robot suit. gimmick or brilliant? finally, a rescue with a difference. a drone was used to save two swimmers in trouble off the coast of new south wales in australia.

lifeguards were being trained to use the rescue drone when practice suddenly became reality and the drone was launched, dropping a flotation device to the teenagers. the whole rescue took just 70 seconds. in this trendy part of downtown las vegas, these passengers are waiting to hop on a special kind of ride. for the past two months, french autonomous vehicle company navya has been offering free bus rides to the public. admittedly, it doesn't travel far, itjust does a loop around the block with one stop at a doughnut shop. at least they're getting a taste of the future! down the road, though, i'm waiting to catch a more private ride, which i've booked on an app. as if by magic, the door opened! laughter. the team was still ironing out a few issues, shall we say.

i think this is the first genuinely autonomous vehicle i've been in where there really is no driver and there really is no place for a driver. there's just safety man, here. that's it. safety man has an xbox one controller down by his side. navya is certainly not alone in this space. uber and google's waymo have been battling it out for some time to become the first fully—autonomous cab—sharing service. self—driving cars use a lot of sensors to be able to navigate the road safely. perhaps one of the most important is lidar, this is what enables the car to judge its surroundings. the design of these sensors is at the heart of waymo and uber‘s court case. navya's car is no different. it too uses lidar to look around. interestingly, what it's not doing

is looking at the traffic lights to judge what colour they are. they've fitted special sensors to each traffic lights and those sensors talk to the car. that doesn't sound very scalable to me. that sounds like you wouldn't be able to put this sort of technology on the open road, without fitting every single traffic light in the us with these sensors. it is much morejust for predetermined routes for this kind of shuttle vehicles. while i've been riding around in this particular smart vehicle, dave lee has been up in reno, not that far away, looking at a system that's making use of data collected by vehicles like this to help an entire city to move more smoothly. there's been great strides made in self—driving technology, over the past decade or so, but the thing about autonomy is that it's often tested in bright, clear conditions.

the real world is much more distracting. sirens. in fact, it's notjust darkness that's difficult for existing autonomous technologies. whether it's through rain, snow, orjust far up ahead on the road, there's a lot self—driving vehicles struggle to see. important work is taking place at the university of reno, nevada, that's attempting to solve that problem, making autonomy more intelligent. and it all begins here, on the corner of 15th and virginia. so, at that corner we have a lidar sensor. that lidar sensor used to be on the autonomous vehicle. but we moved it from the vehicle to the intersection, so it can track each pedestrian here, each vehicle here. what kind of things is it picking up? is it recognising who people are or not? no, it only recognises this as a pedestrian

or this is a vehicle. it does not recognise who the person is. think of this intersection as providing more eyes to an autonomous vehicle. it could detect a threat and communicate that to a car heading in its direction, telling it to slow down, beware. so what these sensors are doing, in essence, is giving autonomous cars more eyes on the road. yes. so theyjust know more about what's coming up ahead. exactly, so no black spots. part of the same programme is this connected car. a modified lincoln that can not only drive itself around, but also communicate with other vehicles and components in the city, signalling its intentions to others. the hardware that you see is pretty similar to what you're going to see in most autonomous vehicles, if not all of them. where we really distinguish ourselves is in the software. so, our research focuses

on what i call social intelligence. we're trying to build machines that understand people, and understand human social behaviour, and can predict what other people are going to do, and then act appropriately. it's a skill that humans have. we navigate driving effortlessly, even though we can't read other people's minds, and it's a skill that computers are going to have to have, if they're ever going to drive cars in the world, with the rest of us. and then there's the challenge of making the technology work, in difficult conditions. inspired by an earlier project to help drone see in the dark, the team at the university's autonomous robots lab has combined lidar, radar and cameras that use advance image recognition. not only does this dramatically improve what the car can comprehend, it's also pretty cheap. once that technology is safe and ready, the plan is to deploy it on electric busses, like this one. but until then, the team plans to use the autonomous tech to gather

large amounts of data in preparation for a self—driving future. this bus made by a california—based company is already out on reno's roads, but right now, with a more traditional type of driver. it is not autonomous yet. the idea is to, at some point, focus on that project. however, right now, we are focusing on data collection for what we call the living lab, and data collection thatis going to be used for the mobility programme. that is going to be used for the mobility programme. for the foreseeable future, these buses will gather data for the living lab programme in reno, a city that perhaps knows more about what's going on on its streets than almost any other city in the world. that was dave, and now to something that we've been hearing a lot about recently — augmented reality. now, it works by overlaying graphics on top of the real world, and while ar games like pokemon go have enjoyed global success,

the most hyped bit of ar kit, magic leap, is still waiting to be released. ar remains a technology that promises more than it delivers. but by combining ar with al, researchers in florida are hoping to create new ways to train people to perform complex tasks. marc cieslak took their ar kit for a test drive, or should that be a test flight? the university of central florida has a long—established relationship with the simulation industry. helping create simulated experiences for everything from driving... ..to supermarket shopping. the simulation lab here's latest project is a bit more highflying than high street, though. as long as we've had pcs, we've had flight simulators. but if you're really serious about learning how to fly, then you need an aircraft

and a human pilot, to teach you what to do. well, this lab is about to be transformed into an aircraft cockpit, with the help of this augmented—reality headset. and when i put it on, it will also provide me with my very own virtual captain. called project cap, it's a collaboration with aerospace giants boeing. necessary information. sure, i have doubled my flow. i'm ready for my take—off brief. give it your best try. the ar cockpit companion is designed as an additional training aid for pilots. the portability of the hollow lens meaning they can brush up on skills or practice in almost any environment. it does feel as if i can reach out and touch the controls, and i'm very much tempted to. and i do that, and, of course, there's nothing there. thin air. ok, cap, you seem to have a better idea of what to do in this aircraft. so call for taxi. roger, the sequence will be

two followed by one. you are clear to start number two. ready. at the moment, cap responds to a very small number of voice commands or questions. beacon. on. mic check. mic check, one, two, three. but simple practise can still be useful for training pilots. augmented reality gives us a chance to bridge the gap between the things that have always been trapped in the digital world and the real world around us. how can we start to merge those things together in effective ways? how can we create holograms, right before you, for things that would be less safe, if you were to do them in the real world, or that you might need additional information besides what you can build around you in the real world? closed and locked. looks behind him to check that. v1 120. vr 11:0. it's a very convincing illusion that there's a pilot in here with me. any questions? do we have a specific altitude restriction for this sim? a year in development,

cap himself is actually modelled on a real pilot. we had an opportunity to take some of our friends who are pilots, in this case one in particular, and see if he would actually subject himself to a full—body scan, to then be able to use him as our avatar. so that's who we have, an actual pilots, who knows the mannerisms and gestures, that we can put into that virtual pilot's seat. but is this another instance of technology putting people out of theirjobs? no, not at all. it's to provide student pilots with the opportunity to practise interpersonal skills, before they actually get to a flight training centre, with real pilots. and we can provide them with a greater breadth of experiences, through introducing different variables, such as different cultural types or personality styles, that they can practice with. i do wonder about other applications for this sort of kit. somebody that might be able to teach you how to drive a car, for instance, or teach you how

to operate various bits of equipment and machinery. some of the work that we had done before doing the work with boeing was in things like medical simulation. being able to have a holographic overlay, so that you could see the x—rays laid on top, exactly placed, or the ct scans or mris. those are things that we think hold great promise, not onlyjust because they'll help with visualisation, but they might also lead to better quality of care, or lifesaving, because you have better access to data right when you need it. so one day, beyond the cockpit, cap‘s digital descendants might help teach us how to do all kinds of things. sure, no problem. and from boeing to boozing. i'm on my way to the tipsy robots, a bar where mixology has been given a hi—tech makeover. here, the drinks are shaken and served by these two chaps.

i can even invent my own cocktail, by choosing from some of the 120—odd spirits hanging from the ceiling, or i assume all of the 120 odd spirits, if i want. can i do that? no, i can't do that, apparently. these droids can mix over 100 cocktails an hour between the two of them. that sounded impressive, until i discovered some human bartenders can do ten times that. and that is it for click in the us for this week. don't forget, you can follow us on twitter, where you can see loads of extra backstage videos and photos. although trust me, you don't want to see what happens after i have one or two of these. cheers, see you soon. fruity. hello, once again.

it's not unusual to have a real mishmash of weather across the british isles at this time of year, and we certainly have one through the rest of the day. across the northern parts, mainly scotland and the north—east of england, it's still that wintry look to proceedings. sunshine, yes, sunshine, no further south. northern ireland, a greater part of england and wales, you're stuck close to a weather front, so a real contrast in fortunes. still some winteriness about the showers across the north of scotland, elsewhere dry, fine and sunny. look at the temperatures at lunchtime, one, two, three degrees or so. similar sort of prospect in the north—east of england. come back towards northern ireland, right through the greater part of england and wales, its on the cold side, leaden skies, dank and drizzly affair. may be a bit of winteriness about proceedings here across the high ground of wales and into the peaks as well. maybe across towards east anglia as well in the more intense downpours of rain. much of this light and patchy for the most part.

it remains windy across the far south—west. much milder here, as you will see in just a second. and there is a chance of some sunshine here but elsewhere for the greater part of england and wales, it is a three, four, five afternoon and leaden skies, and as i say, drizzly and chill to go with it all. further north, some sunshine, one, two, three degrees only. far too late in the day we will get rid of that rain down towards the near content, keeping the supply of showers going across the north. and into that cold reservoir of air we push another frontal system from the atlantic. it stays pretty mild across this south—western quarter, but we have a real issue looming for sunday morning. snow and ice, because as the rain falls into that cold air, i'm sure you know what's coming next, we are going to convert some of that rain into snowfall. i highlighted scotland and northern england. it isn't just there, but as we get that moisture up and over the peaks and the pennines getting into the high ground of scotland, and then not necessarilyjust on the high ground

but to quite low levels in the east we could find two or three centimetres of snow lying for a time through the afternoon until we roll a little bit more in the way of warmth. it is only a a little bit but it is enough to flick that equation back from snow to rain. then we will begin to see the first signs of the mild air beginning to win out, which it does as we start the new week. but it will be a wet and windy start for a time across the southern counties of both england and wales. good afternoon. the us national government has officially shut down, a year to the day after donald trump took office as president. senators had failed to agree a stop—gap budget — amid a bitter dispute president trump's republicans and the opposition democrats over immigration and border security. the impasse will affect hundreds of thousands of federal workers. but essential government services will carry on. the recriminations have already begun. david willis reports from washington.