A new simulation system allows driverless vehicles to be trained in an environment with infinite steering possibilities. While autonomous vehicles typically rely on datasets from real-world human drivers, simulated testing opens up opportunities for cars to encounter and navigate through a variety of worst-case scenarios before they’re even released out on the streets.

“It’s tough to collect data in these edge cases that humans don’t experience on the road,” said researcher Alexander Amini. “In our simulation, however, control systems can experience those situations, learn for themselves to recover from them and remain robust when deployed onto vehicles in the real world.”

Real-world environments are ultimately more complex. Realistically, there are numerous factors to consider, making it impossible to incorporate all variables in a simulator. This creates a discrepancy between the results in a simulator and when the vehicle is finally deployed in real-world streets.

For this purpose, researchers from MIT created the Virtual Image Synthesis and Transformation for Autonomy (VISTA), a photorealistic simulator capable of rendering detailed virtual roads. The biggest difference with this simulation engine is that it uses datasets from drivers to “synthesize” trajectories consistently, particularly in terms of road appearance, as well as the distance and motion of objects in a given space. The vehicle’s control system must learn on its own how to reach its destination safely. This includes the ability to recover and maintain control when swerving between lanes or encountering a near-crash scenario.

Using an end-to-end framework, the team also integrated a reinforcement learning approach, providing feedback signals whenever the control system commits an error. This means it will only receive a feedback signal if, for example, the car crashes. When that happens, a new simulated location is immediately generated. From this point, the system has to once again attempt to successfully use different steering angles to avoid crashing again.

The control system was deployed on an autonomous vehicle after driving 10,000km in the simulated environment. This is the first controller system using an end-to-end reinforcement technique that has ever been successfully deployed onto an actual autonomous vehicle.

In multiple tests using VISTA with a driverless car, the control system was able to successfully and safely navigate across streets it was deployed in for the first time. The vehicle was able to recover back into a safe driving trajectory within seconds whenever the simulator replicated near-crash situations.

According to the researchers, they aim to further simulate more types of road conditions from a single driving trajectory, including night and day and sunny and rainy weather. Additionally, they expressed that they will be studying how to simulate more complex interactions with other vehicles on the road.

The complete study was published in the IEEE Robotics and Automation Letters journal. More details will be presented at the upcoming ICRA conference in May.

For similar stories, check out how augmented reality is being used to improve autonomous vehicle testing here.

The post New Autonomous Driving Simulator Uses End-to-End Reinforcement Learning appeared first on Artificial Intelligence.

Craving a bite out of a freshly grilled ballpark frank? Two robots named Jaco and Baxter can serve one up. Boston University engineers have made a jump in using machine learning to teach robots to perform complex tasks, a framework that could be applied to a host of tasks, like identifying cancerous spots on mammograms or better understanding spoken commands to play music. But first, as a proof of concept—they’ve learned how to prepare the perfect hot dog.

Researchers still don’t fully understand exactly how machine-learning algorithms—well, learn. That blind spot makes it difficult to apply the technique to complex, high-risk tasks such as autonomous driving, where safety is a concern. In a step forward published in Science Robotics, Calin Belta, a BU College of Engineering professor, and researchers in his lab taught two robots to cook, assemble, and serve hot dogs together. Their method combines techniques from machine learning and formal methods, an area of computer science that is typically used to guarantee safety, most notably used in avionics or cybersecurity software. These disparate techniques are difficult to combine mathematically and to put together into a language a robot will understand.

Belta, a professor of mechanical, systems, and electrical and computing engineering, and his team employed a branch of machine learning known as reinforcement learning. When a computer completes a task correctly, it receives a reward that guides its learning process. Although the steps of the task are outlined in a “prior knowledge” algorithm, how exactly to perform those steps isn’t. When the robot gets better at performing a step, its reward increases, creating a feedback mechanism that pushes the robot to learning the best way to, for example, place a hot dog on a bun.

Integrating prior knowledge with reinforcement learning and formal methods is what makes this technique novel. By combining these three techniques, the team can cut down the amount of possibilities the robots have to run through to learn how to cook, assemble, and serve a hot dog safely. Belta sees this work as a proof-of-concept demonstration of their general framework, and he hopes that moving forward it can be applied to other complex tasks, such as autonomous driving.

The post Meet Jaco and Baxter, Machine Learning Robots Who Cook Perfect Hot Dogs appeared first on Artificial Intelligence.

Everybody in the automotive industry is hustling to progress autonomous driving technologies and deploy driverless cars. This year, we’ve seen huge steps on account of rising technologies, for example, augmented reality and artificial intelligence. In light of current circumstances, Tesla CEO Elon Musk may really have the option to make a completely autonomous vehicle before the end of 2020.

Augmented reality highlights are as of now deployed by a few automobile manufacturers and choices for aftermarket augmented reality products is developing. Because of the administrative issues of completely autonomous driving and different limitations that still should be worked through before autonomous driving innovation is widely adopted, it’s a decent bet that augmented reality will be the next automotive innovation to be seen on the roadways.

There’s a marketing challenge with autonomous driving. The general population and a significant part of the press still don’t see how it functions. What’s more, if they don’t see how it functions, they’re not liable to purchase in.

Civil Maps, a startup whose mission is to crowdsource maps for self-driving vehicles, thought of an extraordinary solution. They made sense of how to give individuals access to the mind of a self-driving machine. The innovation utilizes augmented reality to show a “vehicle’s eye view” while driving. During a ride in one of the specially-equipped cars, a screen gives the rider a visual representation of sensor information as the vehicle processes it.

The designs are unrefined, comprised of essential hues and wireframes. It would appear that a 90s form of Google Street View. However, it’s rapidly apparent exactly what amount is truly going on. With a couple of hues and some floating boxes, the demo shows how the vehicle sees the state of the street, road signs, navigational milestones, and different vehicles. It can even tell if a traffic light is red, green, or yellow. It’s an unbelievably accessible look into an extremely mind-boggling innovation.

By overlaying essential data for the driver inside their line of sight, augmented reality can improve security. Rather than a driver expecting to look down at the dash or to their phone to get driving information or data, the guarantee of augmented reality would have the information accessible on a heads-up show, the windshield or projected on the road in the driver’s line of sight. Notwithstanding assisting with route and information from gauges, augmented reality can make the driver mindful of risks and other emergency notices. Through Nissan’s Invisible-to-Visible innovation, vehicle drivers later on could pick to have augmented reality travelers in the vehicle who can cooperate with them, perform co-pilot duties, and much more.

Augmented reality’s ability to incorporate a virtual domain into this present reality makes it perfect for testing driverless cars. It gives a quicker way to deal with testing as well as an increasingly affordable one.

The University of Michigan utilizes AR and other virtual technologies to make a protected space for testing self-driving vehicles in their Mcity Test Facility. It has 32 acres of land of fake street and infrastructure. From that point, genuine vehicles can connect with computer produced autos in real-time. In this reenactment, specialists have set up various conditions and situations that mirror true challenges. This empowers them to survey the security of driverless vehicles.

Scientists likewise utilized this AR simulation to test a patent-pending programming that empowers real and virtual cars to communicate. By trading data out and about, self-driving vehicles can settle on better choices. Hence, they can explore any environment securely without imperiling the lives of their travelers. At Mcity, analysts are additionally leading tests to improve self-driving experience. Their simulations assess how drivers react to unexpected stops and different difficulties when the vehicle is accountable for the driving. In doing such, they try to improve the quality of the ride.

In comparison, a large group of smaller start-up self-driving organizations have been discreetly creating and working constrained local driverless taxi benefits in a few areas. Drive.ai, for instance, started working a self-driving shuttle service in Arlington, TX and Fresco, TX, in the territories of the city involved principally of entertainment, commercial and retail foundations. Every one of the organization’s vans is furnished with self-driving tech and can be brought utilizing an application. The project is still, in fact, a pilot, with obligatory safety drivers, yet if they keep on observing achievement it’s anything but difficult to accept that Drive.ai, or one of their rivals, could present the first wide-scale commercialized service.

Other start-up efforts incorporate a deal between Kroger markets and start-up Nuro to deliver groceries utilizing a completely autonomous vehicle called the R1, or the organization Voyage offering self-driving taxi service at a Florida retirement network called the Villages.

There’s a tremendous amount of investment in research and development of autonomous driving advancements. In any case, not many products have hit the market. Starting now, everything we can do is wait. Obviously, completely autonomous vehicles will change our everyday lives. They will disrupt enterprises as much as or maybe considerably more than augmented reality as of now has.

The post AUGMENTED REALITY IN AUTONOMOUS CARS ADVANCEMENTS appeared first on Artificial Intelligence.