NEWS

What makes a vehicle autonomous? It’s not a straightforward question. If a car can drive itself in most situations, but a person needs to stay at the steering wheel in case there’s an emergency, is that an autonomous car? To tackle this problem, SAE International, a global developer of engineering standards, created the ‘levels of driving automation’. This is a six-point scale to classify the autonomy level of vehicles, running from 0 to 5. We’re going to take an in-depth look at the six levels of vehicle autonomy, and at where we are now. Vehicle autonomy level 0 The lowest level of vehicle autonomy is level 0. You might expect a level-0 vehicle to have no autonomous features at all, but in fact it may have minor features to support or warn the driver. For example, automatic emergency braking when the vehicle detects that it’s about to collide with something, or proximity sensors that warn the driver of obstacles when parking The human driver is driving at all times. The vehicle may be able to offer warnings, or to take quick action in emergency situations, but it can’t perform the day-to-day driving procedures of acceleration, steering and ordinary braking by itself. Without the driver, the vehicle wouldn’t be able to move at all. Vehicle autonomy level 1 At level 1, a vehicle has features that assist the driver with either steering or speed control, i.e. acceleration and braking. For example, a car might have adaptive cruise control. This allows it to detect any vehicles in front of it on the road, and to change its speed if necessary to stay at a safe distance. This would be a level 1 car, unless it has other autonomous features that move it up to level 2. Again, the autonomous features can assist, but a human is driving the vehicle at all times. Adaptive cruise control can make long drives easier, but it can’t drive a car on its own. Vehicle autonomy level 2 At level 2, a vehicle can assist the driver with both steering and speed control. For example, take the hypothetical level 1 vehicle that has adaptive cruise control. If you add an autosteer feature that keeps the car centred in the lane, it’s now a level 2 vehicle. The vehicle still can’t be said to be capable of driving itself at level 2, although it may be able to travel some distance without intervention. There are situations the vehicle won’t be able to respond to appropriately, such as approaching a traffic light. It’s the human driver’s responsibility to supervise the vehicle’s automated features at all times. As the driver, you need to keep your hands on the steering wheel, stay aware of your surroundings and be prepared to take action at any moment. The dividing line between driver support and automated driving The SAE levels of autonomy can be divided into two sets of three: 0 to 2, and 3 to 5. At the levels we’ve looked at so far, the automated features of the vehicle are driver support features. Up to and including level 2, you, the human driver, are responsible for the vehicle at all times. You need to stay alert and aware constantly, with your hands on the steering wheel and your eyes on the road. The vehicle can assist you, but it can’t drive itself; that’s your job. From level 3 upwards, you’re not the one driving. You’re there to assist the automated systems, rather than the other way around. At levels 0, 1 and 2, the vehicle is considered to be providing support to the driver, rather than driving itself. At levels 3, 4 and 5, the vehicle is capable of automated driving. Vehicle autonomy level 3 At the lower levels, a human is driving or at least supervising at all times. In some scenarios, though, a level 3 vehicle can handle itself. In addition to steering, braking and accelerating, it can monitor its environment and react to particular situations. For example, it can recognise that a nearby car is moving more slowly, make the decision to pass it and perform the passing manoeuvre itself. This is the first level at which, as the human in the car, you can take your eyes off the road. However, a level 3 vehicle can only drive itself in specific, limited situations. Outside those situations, the vehicle will ask you to take over, so you’ll need to be available to drive at all times. SAE gives the example of traffic jam chauffeur technology, which can handle the vehicle in traffic jams below a certain speed. During the traffic jam, you can engage the technology, freeing yourself up to focus on other things. When you’re clear of the traffic jam, though, you’ll be required to take over driving again. Vehicle autonomy level 4 At vehicle autonomy level 4, the vehicle can drive itself without any human input beyond being given a destination. It can find its own way while steering, accelerating, braking and responding to other traffic as necessary. It won’t be able to do this everywhere – it’ll probably be helpless if you drop it in the middle of a field, for example – but it should be able to drive itself in areas that are thoroughly mapped and monitored, once it’s received all the information it needs to handle the area. At all the previous levels, you needed a qualified human driver in the vehicle, either to do the actual driving or to take over when necessary. At level 4, you can climb into the vehicle and let it transport you even if you don’t have a driving licence yourself. In fact, you don’t necessarily need a person in the car at all. When a human driver isn’t needed, entirely new possibilities open up. A level 4 vehicle might serve as a driverless taxi, for example, trained to carry people around a particular area. It can travel from the destination to the next customer without human involvement, and the people it picks up don’t need to be capable of driving. Vehicle autonomy level 5 We’ve now moved fully into the theoretical. At level 5, a vehicle is perfectly autonomous. It doesn’t need a human driver; it doesn’t need to be familiar with an area; it can find its own way through any terrain, in any conditions. A level 5 vehicle should be able to handle any situation that a human driver could. At first glance, levels 4 and 5 can look very similar. In both cases, the vehicle can travel entirely by itself; the difference is that a level 4 vehicle can only do this in limited conditions. In fact, there’s a large gulf between level 4 and level 5. At level 4, the autonomous systems have all the information that they need to operate in a particular area. At level 5, the vehicle needs to be able to work out how to respond to a situation by itself, without guidance. This makes level 5 vehicle autonomy a more distant goal, but it’s one that we’re working towards. Where we are now There are exciting developments taking place in the field of vehicle autonomy, some of them being researched at our own SatCom Lab. Level 3 technology exists already, and level 4 technology is being developed. However, autonomous vehicles can only be driven on public roads if the law allows, and the technology must be tested thoroughly and shown to be safe before regulations can be changed. Because of this, you won’t usually see anything over level 2 on today’s UK roads. Level 2 might not seem that impressive, particularly as a vehicle’s only technically considered capable of automated driving at level 3. However, when you consider it, even level 2 is a remarkable achievement. It’s the first point at which a vehicle can travel any distance without the involvement of the human driver beyond monitoring the situation. At level 1, you’re always responsible for at least one crucial, constant aspect of driving: speed control or steering. At level 2, you may only be able to activate the automated features in particular situations – travelling down a long road, for example – but, while they’re active, you can just sit with your hands on the wheel and stay alert while your vehicle carries you along. We have an exciting future ahead of us, but it’s worth taking a moment to admire how far we’ve come already. Now, thanks to the work of countless people, true automated driving is just around the corner, and we’re playing our part to get us there. Darwin Innovation Group is a UK-based company that provides services related to autonomous vehicles and communications. If you’re interested in working with us, take a look at our careers page. If you’d like to know how we can help your organisation make use of autonomous vehicles, contact us. You can also follow us on LinkedIn or Twitter.

Space is vast and beautiful, but it’s also largely barren. If we want to send astronauts into space for long periods, we’re faced with a lot of questions. How do we keep people fed in space, given that it costs thousands of pounds to send even a kilogram of supplies into low Earth orbit? How do we supply them with enough vitamin C to prevent scurvy? How do we look after their psychological health? A lot of problems could be solved if we could grow plants in space. Plants can convert carbon dioxide into oxygen. They’re a source of nutrients, especially vitamin C, and can be stored for long periods in the form of seeds. Many of them can be eaten raw, which is convenient in an environment where cooking may not be straightforward. They could even be used for fuel. Plants are also pleasant to be around, which is valuable in itself. In lockdown, a lot of us have discovered the value of looking after houseplants, seeing trees outside our windows or going on strolls through nature. Being around plants feels refreshing, which is important when you’re in a confined, static environment. Growing plants in space isn’t easy. They can’t naturally access the light and nutrients they need, and the lack of gravity causes its own problems. We’re going to take a look at what’s been achieved so far, and at the research that will make space plants easier to grow in the future. Has anyone managed to grow plants in space? The history of plants in space is only a couple of decades shorter than the history of humans in space. As recorded by Guinness World Records, the first species of plant to flower in space was Arabidopsis, a type of rockcress, grown aboard the Soviet space station Salyut 7 in 1982. Since 2014, NASA has managed to grow assorted plants in the Vegetable Production System (Veggie), a miniature garden on the International Space Station (ISS). The growth experiments have included lettuce, mustard, algae, zinnias and kale. You can read about NASA’s efforts here. You can’t just plant a seed in soil and trust it to stay in place without Earth’s gravity. The Veggie seeds are glued with a natural gum onto ‘plant pillows’: small bags, secured in place, which contain soil, controlled-release fertiliser and water that’s injected through a valve. After all, without gravity to draw it downwards, you can’t just pour water in space. When the seeds are glued in place, they’re carefully positioned to encourage the roots to grow ‘down’ into the pillow, as the seeds can struggle to grow in the right direction without the guidance of gravity. A bank of LEDs above the plants gives off light designed to help the plants flourish, and this also aids in guiding the plants ‘upwards’. These experiments allow NASA to study how plants grow without Earth’s gravity, which is obviously important for future space plant efforts. They’re also making a difference right now; the astronauts aboard the station have the pleasure of looking after the plants and, when they’ve grown an edible crop, of eating the results. The ISS also houses the Advanced Plant Habitat, an automated growth chamber with cameras and sensors. The sensors constantly feed back to an Earth-based team who can control the system themselves, meaning the crew on the space station don’t have much involvement. In other words, remote connectivity has made it possible for people on Earth to grow plants in space. Other organisations have also experimented with plant growth on the ISS. The European Space Agency (ESA) has an interesting article on an experiment conducted in the Columbus ISS laboratory, studying the effects of weightlessness on thale cress seedlings. What is the AstroPlant project? Even though there have been successful efforts to grow plants in space, there’s a lot of room for improvement. To get better at growing plants in space, we need more data. That’s where the AstroPlant project comes in. At the moment, as a species, we don’t have the resources to perform a lot of space-based plant-growing experiments. NASA’s Veggie experiments are incredible, but the Veggie gardens on the ISS are very small, and there are only two of them. Between 2014 and 2019, the Veggie units produced just over 20 crops. That’s still an impressive number of crops to grow in space, but it means that the Veggie experiments can only give us information slowly. The AstroPlant project, in association with ESA, allows anyone to grow plants in systems similar to the ones used in space and submit their findings through an app. You can grow the plant at home, but, like the ground-based NASA team operating the Advanced Plant Habitat, you can remotely connect with the system’s CO2 sensor, lights, camera and fans. With the help of the public, ESA can gather data through this project and get an idea of what plants might be good candidates to grow in space. Although he doesn’t work directly with growing plants in space, Carlos Carbajal has experience of developing systems that allow plants to grow in confined indoor spaces, without natural sources of light, water and nutrients. Each plant needs different conditions, he explains, and the AstroPlant project can help find the perfect conditions for different plants much more quickly than would be possible otherwise. For example, you might want to test the effects of different combinations of water, light, nutrients and carbon dioxide on wheat. ‘Each test will require at least a couple of weeks to see the results, and thus it could take years to study just one plant,’ Carlos says. ‘AstroPlant is trying to spread this load and collect data from many participants around the world.’ With this collected information, the project can draw conclusions about the best conditions for each plant. ‘We are far from understanding space-like conditions, including no gravity, no sunlight, high radiation, isolation, and many other yet unknown factors,’ Carlos warns. ‘However, the insights produced by AstroPlants will help astrobiologists to develop more accurate predictions of what could happen to the plants in space, and what experiments they should carry out in the future.’ Here at Darwin, we’re getting involved in the AstroPlant project ourselves. You can follow the experiment on our AstroPlant page. Darwin Innovation Group is a UK-based company that provides services related to autonomous vehicles and communications. If you’re interested in working with us, take a look at our careers page. If you’d like to know how we can help your organisation make use of autonomous vehicles, contact us. You can also follow us on LinkedIn or Twitter.

With the help of David Owens, our chief drone pilot, we recently talked about drone laws in the UK. David offered so much interesting information that it was hard to fit it into one article, so, having examined the laws, we’re going to look more at the operation and applications of drones today. What level of autonomy is currently possible with drones? At the moment, we can fly autonomous missions with drones, as long the drone remains within visual line of sight (VLOS) at all times, with software like PIX4D or Mission Planner. However, the remote pilot must be ready to take control in case of an emergency. The remote pilot must also follow the ‘see and avoid’ principle, actively watching for other air traffic at all times. Many modern drones have a level of autonomy. For example, the operator tells the drone where to go, and the drone manages how to get there, manoeuvring around buildings and trees and taking care of keeping itself in the air. The drone can also return home and land itself if its battery is running low. This is all fine as long as VLOS is maintained throughout. However, the operator needs to stay focused and monitor the drone at all times, so they can react if something unexpected happens. A drone’s autonomous systems can lower the stress on the pilot, but we’re not yet at the point where drones can be left to operate themselves entirely without supervision. Ideally, given a destination, future drones would be able to get there safely without the need for constant conscious monitoring, or the need to maintain VLOS. Going by SAE International’s levels of driving automation (which run from 0 to 5, where 5 is ‘completely automated’), it’s currently accepted that most commercial systems are at level 2: the drones can operate themselves, but they need to be supervised by a human at all times. We’ll talk more about the levels of automation in a future news post. The plan for the drones we’re building for Darwin is to have all the autonomous systems duplicated on board, so there’s a backup to take over if the main system somehow fails. We’re also carrying this principle of redundancy over to the physical aspects of the drone; the drones can operate on four motors, but they’ll have an additional four motors as a backup. If there’s an issue with one of the motors, we want to make sure the drone will still be able to return home or land in a controlled way. Autonomous systems can be used to bolster safety as well as to aid in flight. Our drones will have presence detection on all sides, so the drone can detect danger and stop itself if it somehow gets too close to an obstacle. As the UK’s detailed drone laws and restrictions demonstrate, it’s important to consider safety at every stage of drone operation. What are the potential uses of drones? The obvious commercial possibility of drones is using them for product delivery, and that’s an area we’re interested in at Darwin. Of course, some products are too heavy or bulky to be carried by drone, but many items are feasible candidates for drone transportation. In 2016, for example, a spokesperson for Amazon said that 90% of Amazon sales are products weighing 2.3kg or less, as mentioned in a Guardian article about drone delivery. In order to run a delivery service using drones, it’s useful to have certain permissions from the Civil Aviation Authority (CAA). For example: An Operational Safety Case permission to fly beyond visual line of sight (BVLOS), i.e. more than 500 metres from the drone operator. Permission to fly within 50 metres of buildings. Without this permission, you need to keep your drones at least 50 metres away from any building that isn’t under the control of the remote pilot, which creates obvious obstacles for drone delivery. Permission to carry hazardous substances. You’d need this permission if you plan to transport, for example, lithium-ion batteries, which are often included in mobile phones or laptop computers. To get any one of these permissions, you’ll need to write a detailed Operational Safety Case, providing evidence of the safety measures and risk management systems you have in place, which would need to be approved by the CAA. More crucially, drones can be used to save lives. Although they can’t transport people, they’re not restricted by regular traffic in the way that, for example, an ambulance would be, and some can travel at 50mph or more. This means they can very quickly deliver urgently needed medical equipment, or supplies that can’t stay outside a particular temperature for too long: blood or vaccines, for example. As blood and vaccines would be classed as hazardous substances, you’d again need specific permission from the CAA to do this. A drone equipped with a defibrillator could fly to the scene of a cardiac arrest, with instructions included, so anyone at the scene could use it in an attempt to bring the patient’s heart back to a normal rhythm. The BBC has a short video from 2017 on the research being done in this area. Drones can also be used to help rescue lost people, and they can do this in a couple of different ways. If someone is lost on a mountain, for example, a drone with cameras or thermal sensors would be able to survey the area more efficiently than ground vehicles or people on foot, and could be deployed more quickly than a helicopter. In addition to this, drones can aid lost people by providing connectivity. If someone gets lost in an area without mobile phone coverage, they can’t call for help. However, a correctly equipped drone could temporarily carry phone service into the area – essentially a complete network in a box – and allow them to make an emergency call. The stranded person can explain where they are, ask for help and request anything they need urgently while they’re waiting for rescue (food, water, medicine or blankets, for example), which could then, again, be transported to them using a drone. There are many other uses of drones. For example, they can be used for filming or aerial photography. They can be used to monitor or map areas that would be difficult to get a full picture of from ground level, or areas that humans can’t enter without taking safety precautions, such as construction sites, mines or sewers. They can gather visual information from cameras, but they can also give a fuller picture of an environment by using, for example, temperature or humidity sensors. Drone technology is an area with huge potential, and we’re excited to be working with drones at the Darwin Drone Lab. Darwin Innovation Group is a UK-based company that provides services related to autonomous vehicles and communications. If you’re interested in working with us, take a look at our careers page. If you’d like to know how we can help your organisation make use of autonomous vehicles, contact us. You can also follow us on LinkedIn or Twitter.

As part of our work at Harwell, we’re building a drone corridor with the help of the European Space Agency. What is a drone corridor, though, and what are the laws and restrictions involved in creating one? In this post, we’re going to dig into some of the legal and safety issues of commercial drone operation. To put together this information, we spoke to David Owens, our chief drone pilot, who’s been flying unmanned aircraft for three decades. Thank you for offering your advice, David! What is a commercial drone corridor? A commercial drone corridor is a flight path that’s been designated for the safe and legal operation of drones. It’s not visible, but it can be pictured as a corridor in the air. The corridor has a ceiling, sides and a floor: it needs to be a certain level above the ground, and there are limits on how high and how far sideways the drones can go. There are some aspects to drone operation that can’t be predicted. For example, you can’t guarantee that birds won’t fly into your drone corridor. But creating a space specifically for commercial drone operation, avoiding structures, crowded areas and other air traffic, means the drones will have as few risks as possible to be mitigated against. Drones and more specifically multicopters have limited battery life, so ensuring they have a clear flight path can make a real difference to how far they can fly. A few seconds here and there are valuable to a drone with a 30-minute flight time. Because of this, drone corridors help to extend the range of drones. Of course, drone corridors also benefit other air traffic. If drones travel through a designated corridor, rather than moving freely, they’re unlikely to disrupt anyone else. Do you need a licence to operate drones? If you’re a hobbyist, you must have two IDs in place before flying most drones or model aircraft (weighing between 250g and 25kg) outdoors in the UK; you must pass a theory test to get a flyer ID, and the person who owns that drone or model aircraft must register for an operator ID too. Most people get both a flyer ID and an operator ID at the same time. If you hope to use drones for commercial purposes, you previously needed a Permission for Commercial Operation (PfCO) from the Civil Aviation Authority (CAA), which regulates unmanned aircraft in the UK. From 31 December 2020, the PfCO has been replaced by a new qualification called the General Visual Line of Sight Certificate, or GVC. To get the GVC, you’ll need to pass a short training course, an exam and a practical flight test. You’ll also need to prepare an operations manual that explains how you’re planning to manage your commercial operation and the safety measures you’re taking. If all of this is approved, you’ll get a ‘permission’ from the CAA that lets you remotely operate small unmanned aircraft for commercial purposes. You’ll also need a valid insurance certificate when you’re operating a commercial drone. You’ll need an additional Operational Safety Case (OSC) permission from the CAA if you’re planning to remotely pilot drones beyond your visual line of sight. What is visual line of sight? It’s a critical rule of drone operation that you need to be able to see the drone you’re flying. In other words, you need to keep the drone within your visual line of sight (VLOS). Even if you can still see it, you can’t fly the drone more than 500 metres from you, or more than 400 feet (120 metres) above the ground. (By convention, pilots use metres for distance and feet for height, so that they can’t be confused easily.) Safety is taken very seriously in drone operation, and operating within VLOS is a simple way to reduce the risk of harm. If you can see the drone, you can also see anything the drone might collide with, and you can take quick action if there’s danger. If you can show you’ve considered and managed the risks, you may be able to get permission from the CAA to fly beyond VLOS (BVLOS). Even with permission, you can only do this within a defined drone corridor, which is part of the reason drone corridors are so useful for commercial drone operation or research. Drone corridors legally have to be designed to reduce any risk of collision; we’ll go into more detail on this below. The planned Darwin Drone Lab corridor at Harwell is about 900 metres long, so, if the drone operator stands at one end, the drone will hit the VLOS limit of 500 metres before it reaches the other end of the corridor. By providing evidence of our safe and legal operations, including the systems in place to minimise risks, we hope to demonstrate to the CAA that we deserve permission to fly BVLOS within 6 to 12 months of VLOS operation. At the moment, it’s still possible to fly drones the full length of the corridor by having a remote pilot in the middle of the corridor with observers stationed at both ends, providing additional safety information to the pilot. This means the drone can fly uninterrupted from 500 metres in one direction to 500 metres in the other without compromising safety or breaking any laws. What are the other laws of operating drones or creating a drone corridor? Even once you have your PfCO/GVC, there are limits on how much your drone can carry, how fast it can fly and where it’s allowed to travel. If you fly over someone else’s land, for example, you need the permission of the landlord, and a drone can’t pass over buildings if it has cameras that could be used to spy through the windows. The law says that you can’t fly a commercial drone within 50 metres of people, vehicles, vessels, buildings or other structures that aren’t under your control. This means that drone corridors won’t necessarily run in a straight line. Flying in a straight line is the most efficient way to cover a large distance, but it might bring your drone into areas that are off-limits, so planning is key to safe operation. Drones have to stay at least 150 metres away from large crowds of 150 people or more, and a drone corridor can’t pass over parks or other places where people are likely to gather. It’s possible for drones to cross a footpath, but these regulations are put in place to ensure that drones don’t pass over ‘uninvolved people’, so the drone operator should make sure the footpath is clear before allowing the drone to cross it. Darwin’s drone corridor at Harwell passes mainly over open fields and trees. Whenever you’re flying a drone, whether you’re a hobbyist or a commercial drone pilot, you need to make sure you’re operating legally and you’re operating safely. Most of the drone laws in place exist to make sure drones can coexist safely with people going about their daily lives. On the occasions when accidents do occur, or when drones pass too close to other aircraft (an air proximity or ‘airprox’ event), these incidents are reported, examined and learnt from in order to make the skies even safer. UK drone law is an extensive and complicated subject, but this overview should help to give you an idea of it! We also have an article where we talk more about the practical applications of drones. Darwin Innovation Group is a UK-based company that provides services related to autonomous vehicles and communications. If you’re interested in working with us, take a look at our careers page. If you’d like to know how we can help your organisation make use of autonomous vehicles, contact us. You can also follow us on LinkedIn or Twitter.

If you’re wondering what it’s like to work for Darwin, here’s a quick interview with one of our employees to give you some insight. Zarrar Jadoon is one of our software developers, working with connected and autonomous vehicles (CAVs). He’s planning to start his own business, using some of the skills he’s refined at Darwin, but we had a chat with him before he left. Hi, Zarrar. How long have you been working for Darwin? I’ve been working with Darwin for about a year. I started working at the Harwell campus in early 2020, but from March we started working from home because of the COVID situation. How was the transition to working from home? At the start, it was a bit hard; I wasn’t used to working from home, and I needed to set up my office and everything. I was thinking, ‘Oh, no, this isn’t going to work.’ But I went into my living room and went, ‘Okay, this table will be mine,’ and I built my own small office from there. After a week or two, I felt so comfortable working from home. It was amazing to realise it was something I could actually do. What were you doing before Darwin? I started working with Darwin while I was finishing my Master’s degree in Advanced Computer Science at Oxford Brookes. I’d previously done some freelance work as well. What’s your role at Darwin? I work in the software development team, on the front end. I’ve been creating a website that collects and processes data from CAVs. It fetches real-time data from automated vehicles and presents the information clearly to the user, using graphs and charts, so they can see, for example, the speed and location of the car. It’s really interesting work. It’s amazing to work with new technology. We’ve been using React, and it’s very quick and lightweight. How have you found the experience of working with Darwin? There have been a lot of opportunities for learning, and also for working as a team. I used to work as a freelancer on individual projects, so being part of a team was a new experience for me. It took me a little time to get used to that level of communication, but slowly my team pushed me to communicate more and to become more comfortable with meetings. I think I learnt a lot from that. The team are amazing. They support me in everything. I really feel I’ve had a lot of help here, even working from home. You can just email or text anyone in the Darwin team if you need something. What's your favourite thing about the work you’ve been doing here? Learning new technology and being involved in such a big, exciting project. The work with Harwell has been an amazing experience for me, seeing how CAVs can send real-time data to the cloud and how it can be retrieved in a millisecond. I think that’s very impressive. Thank you so much to Zarrar for taking the time to answer our questions! It’s been good to have you with us, and we wish you the best in your future endeavours. Darwin Innovation Group is a UK-based company that provides services related to autonomous vehicles and communications. If you’re interested in working with us, take a look at our careers page. If you’d like to know how we can help your organisation make use of autonomous vehicles, contact us. You can also follow us on LinkedIn or Twitter.