A new autonomous car called RobotCar is designed by the Mobile Robotics Group (MRG) of Department of Engineering Science at Oxford University in the UK. What distinguishes this autonomous car from others is its price in comparison to its technology.
Currently, the team has 22 researchers and is in collaboration with Nissan by using their technical support and hardware of the vehicle. RobotCar is a modification of Nissan LEAF.
The primary interest of the research team is to create an infrastructure-free navigation through information engineering and autonomous perception. The project aims to ensure a safe drive by connecting with other drivers and utilizing the roads dexterously while leaving some free time for the drivers. RobotCar is able to monitor the road, pedestrians and other cars. In case of an obstacle ahead, the car slows down and stops till the road is clear again.
The car has 3D sensing, thanks to its lasers and offers to take over the drive only after the stored 3D impression memory matches with the current environment. Thus, the car first learns its environment then it is able to take over the drive.
The iPad, the Low Level Controller (LLC) and the Main Vehicle Computer (MVC) are the three computers that the car has. The iPad is the user interface. Everything about the drive can be controlled by the Main Vehicle Computer. For an autonomous driving, all the three computers must be ready. If a problem appears during an autonomous driving, the driver is advised to take over the wheel. Unless the driver takes over, the car stops on its own. The driver can take over the control at the moment he touches the brake.
The vehicle runs on electricity, therefore there is not any carbon emission so it is also an environmentally friendly vehicle.
Instead of GPS, the navigation system of the car is planned to be like the robots used in logistics which are cheaper and based on reflective beacons and guide wires. In order to achieve the technology for the robots to be precise in future, machine learning techniques and mathematical models are used to interpret data coming from sensors. What is interesting about this car is that it has an awareness of static and dynamic environmental characteristics. Static characteristics mean information on the location and type of traffic signs/ markings/lights (and so on) that change hardly. Nevertheless, the data updates itself during a drive, as well. On the other hand, dynamic obstacle detection means detecting any kind of potential obstacles with a standard laser scanner. 85 degree of field view and 13 times a second detection up to 50 meters ahead is available.
It is difficult to distinguish it from other cars just by looking outside since lasers and cameras are hidden carefully. Moreover, the main vehicle computer takes place in the boot.
Although the system costs nearly $ 7750 at the moment, reduction to $ 150 in the future is anticipated by the researchers.
For further information you can visit:
Video Credit: Mobile Robotics Group Oxford