Technology: Cloud-to-car Mapping System for Driverless Vehicles
Our world is extremely dynamic. Everything is evolving so quickly. Especially things closely associated with information technology. We are already used to cars that tell us how and where we should drive, so it’s no wonder that the next step of evolution is the driverless car. There is no doubt that autonomous driving vehicles will soon be a commercial reality.
Wikipedia defines an autonomous car (also known as a driverless car, self-driving car, robotic car) as a “vehicle that is capable of sensing its environment and navigating without human input. Autonomous cars use a variety of techniques to detect their surroundings, such as radar, ladar (laser lights), GPS, the odometer and computerised vision. Truly autonomous cars do not require human intervention other than setting the destination and starting the system. The automatic system can drive and make its own decisions.
The automotive industry has become a very crowded place as automakers lost their position as the key players in the market. IT giants are moving in on this industry, promoting automation and domination of AI on the roads. The future highways and city streets are being looked at through the eyes of their intended future inhabitants: driverless cars and trucks.
The best known driverless car projects already on the road
Tesla Autopilot, Google Waymo, and Self-Driving Uber have become real buzz words in this domain. Though these cars are still not approved for public-access roads, they could become part of our day-to-day routines very soon.
According to Tesla (https://www.tesla.com/autopilot?redirect=no), all Tesla vehicles have the hardware needed to be fully self-driving vehicles at a safety level substantially greater than that of a human driver.
The hardware includes eight surround cameras that provide 360 degrees of visibility around the car and twelve updated ultrasonic sensors that detect both hard and soft objects. All this is in addition to the forward-facing radar with enhanced processing capabilities. The manufacturer states that its self-driving car will match speed to traffic conditions, keep within a lane, automatically change lanes without requiring driver input, transit from one freeway to another, exit the freeway when your destination is near, self-park when near a parking spot and be summoned to and from your garage.
In May 2014, Google presented its new prototype of driverless cars without a steering wheel, gas pedal, or brake pedal. The car was fully autonomous. By March 2016, Google’s fleet of driverless cars had driven in autonomous mode a total of 1,500,000 miles (2,400,000 km). The Google self-driving car project is new called Waymo.
According to Waymo, their vehicles have sensors and software that are designed to detect pedestrians, cyclists, vehicles, road work and more from a distance of up to two football fields in all directions. Waymo says that its self-driving cars can adjust driving to unexpected changes on the road and drive defensively trying to stay out of blind spots and steering away from big trucks.
Uber has launched a driverless car service with the goal to substitute common cars with automated vehicles for shared use. For now, Uber’s self-driving fleet consists of specially modified Volvo XC90 sport-utility vehicles as the company has no intention to manufacture cars of its own. Volvo cars are fitted with dozens of sensors that use cameras, lasers, radar, and GPS receivers. Though the cars have the potential to be autonomous, Self-Driving Ubers have a safety driver in the front seat because it is required by law and the cars need human intervention in many complex conditions, such as bad weather.
The shift to full automation requires reliable autonomous car mapping
Apart from other challenges to be solved in the course of autonomous car evolution such as liability issues, implementation of a legal framework and establishing government regulations for self-driving cars, there is another substantial prerequisite for safe and problem-free autonomous driving. To unlock the potential of driverless cars, reliable mapping is necessary. Because if a car is driverless, software becomes the core part of the car.
Maps for driverless cars must be highly detailed (HD), containing every critical road feature, including slope and curvature, lane marking types and roadside objects. Moreover, it is necessary to build the mapping system for autonomous cars that will become a part of driverless car software, ensure the vehicle is able to locate itself within it, and allow for the rich near-real-time addition of contextual awareness to the traffic situation around the vehicle.
It is obvious that simply GPS-based maps are not accurate enough for next generation cars that won’t have drivers at all. When we deal with a regular map, the car’s position can be pinpointed to a meter or so. HD maps ensure positioning to within 10 centimeters. The requirement for precise locating and positioning involves real-time, or near-real-time, updating of the mapping environment. Introducing Cloud-to-Car Mapping Systems.
Cloud-to-car mapping systems, allow manufacturers, suppliers, and users of driverless cars to get the most up-to-date information about the road network for locating, positioning and maneuvering the vehicle automatically. This fresh information is provided via “over-the-air” software updates. Live, constantly updated HD maps on the cloud allow highly or fully automated vehicles to choose the optimal driving strategies based on roadway profile, lane curvature and terrain, congestion, detours, and other factors influencing driving safety.
High-definition and highly-detailed mapping providing accurate, border-to-border model of the road is key to safe, autonomous driving. To build detailed environment models, mapping experts have to create AI for cloud-to-car mapping as well as process huge arrays of information and ensure real-time data communication between the cloud and the cars.
Dozens of companies have been creating multi-layer, highly detailed maps of roads and highways and there is still a lot of work still ahead. As a rule, an HD-map is created with the help of multiple cameras and sensors installed on cars collecting data. The collected data is then analyzed in real-time with powerful on-board computers capable of processing input from multiple sensors. Afterward, the data is compressed and transmitted to the cloud. The cloud server aggregates and reconciles the transmitted streams of data, and produces the cloud accessible state-of-the-art HD map necessary to create mapping system for driverless vehicles.
Any HD map has several layers, containing both static and dynamic information, which form the vehicle environment map. The dynamic information is no less important. Of course, information on lanes and the route is critical, but to make informed decisions on behalf of the driver, automated cars must also know lane usage rules, road signs, and plenty of other things that can change from time to time. In his interview for carandrive.com, John Ristevski, vice president of reality capture and processing at Here (a mapping company owned by BMW, Daimler, and the Volkswagen Group) said, “a dynamic layer in Here’s HD maps aggregates live data from fixed sensors, government agencies, and other cars on the road to adapt to temporary lane closures, accidents, and traffic jams, along with permanent changes to traffic patterns”.
What does it take to connect an autonomous car and mapping system?
As we can see, some companies develop mapping systems for self-driving cars, and some of them, including NVIDIA and TomTom, develop their own toolkits for autonomous driving capabilities to combine with HD maps.
To enable driverless cars to drive safe and efficiently, it is not enough to just develop a cloud-to-car mapping system and connect a car to it. Software developers have to create the infrastructure software for this technology and integrate the navigation systems into a common information field. Autonomous vehicles armed with cloud-accessible high-definition maps and fitted with deep learning algorhithm software will be able to obtain information not only from satellites, but also from other cars and even city infrastructure.
So, it may not be long before we see the smartest cars pick us up at front door and take wherever we need without any driver input. Those cars will choose the route, set the speed, leapfrog slow-moving traffic, and communicate with other cars so they negotiate interchanges, choose the correct lane for an exit, and obey all the traffic rules.
Archer Software is a recognized provider of embedded software solutions for connected cars. Our portfolio includes various software products in the automotive domain. We have a great deal of experience developing mobile applications for Android, iOS, Windows Phone and Blackberry. Projects include geolocation and geopositioning services, as well as safety and monitoring apps.
For more information about how we can develop connected car software solutions that are right for you, contact our experts at email@example.com.