The big names in car manufacturing are neck and neck to have the first autonomous car. Tesla CEO, Elon Musk recently revealed that within three years, they would have a driverless car. Google said the same but by the year 2018. Nissan made the announcement that by 2020, they would be producing driverless cars. As the days progress, all dimensions about autonomous cars have shifted from futuristic dreams to something more adventurous than expected. Companies are upgrading their technology so that they become the first to roll out production of driverless cars.
Driverless cars can manage similar maneuvers as human do, but with more precision and minimal damage. Take for instance Audi’s self-driving car which managed 156 turns of the 12-mile Hill Climb course in Colorado. According to car manufacturers, this future is attainable, and should sufficiently correct on human errors causing accidents and increased road traffic. So, how do driverless cars work?
Autonomous cars in a nutshell
A driverless car is capable of making maneuvers without human assistance. To enable such control, the car should be able to sense its environment to make navigations. To this effect, it has working GPS systems, an inertial navigation system, laser rangefinders, radar, and video. The vehicle creates a 3D image of its environment by interpreting data from the GPS and the internal navigation system.Data from the sensors is refined to remove noise, then welded with other data sources to have a real-time image. The control system is the overall recipient of the data, which is then used to make the correct navigation.After the vehicle decides the best route to take, the information is broken into commands that are fed into the actuators which control steering, braking, and throttle.
Mapping and localization
Even before any navigations are made, the vehicle has to capture its surroundings by using laser rangefinders and camera. The laser rangefinder will use beams to calculate the time it takes to reach objects and back. The camera is resourceful when developing a color scheme for the image. When the information is coupled to that from the rangefinders, an accurate representation, both in depth and length can be deduced. However, it can be difficult to have an image greater than 100m radius since laser beams diverge as they traverse through space. Thus, it is essential the vehicle be able to use other sensors for path planning.The vehicle uses the GPS, inertial navigation system and sensors to precisely position itself on the map with relation to other objects. For better accuracy, the vehicle would use previous data of the same location to have fewer errors. As the vehicle is in motion, more information is obtained and added to its internal map.
The internal map includes all the information of the environment: both static and in motion, within proximity depending on how good the data is retrieved to match the predetermined shapes and motion of the descriptors. The vehicle will use an intelligent system to categorize and predict the probable pathway of the shapes produced by sensors. This will ease movement as it approaches busy sections of a road. Thus, previous, current and predicted future positions of obstacles are added to the internal map from whence the vehicle plans its path.
The objective of path planning is to use information derived from the internal map to safely navigate while staying out of the way of obstacles. Car developers will input long range algorithms while the car develops its own short range algorithm utilized in instances such as changing lanes, turning right or left, or driving forward.The vehicle then analyses the road and chooses the best and safest alternative. This creates a command that adjusts throttle, braking, and steering. In the end, obstacle detection, mapping, localization and path planning is caused to be a cycle until the destination is reached.
Into the future
Although a lot of advances have been made by car manufacturers towards self-driving cars, some areas still render them unsafe for public use. For instance, the GPS cannot be very reliable, and the imaging system cannot match the impeccable human interpretation.
Some of the disadvantages of self-driving vehicles are: Driverless cars would likely be expensive for most ordinary people. Truck drivers and taxi drivers will lose their jobs. Any malfunction by the computer, even just a minor glitch, could cause fatal accidents than anything that human error might bring about. Hackers getting into the vehicle’s software system and controlling or affecting its operation is a major security concern. The cars would rely on the collection of location and user information, creating major privacy concerns. There are major concerns on the operation of the autonomous systems in adverse weather conditions such as heavy rains. Reading human road signs would be a major hurdle for a robot. Also, human signals such as hand signals would be challenging for the computer to discern. There is also a possibility of cars being loaded with explosives by terrorists. There is a challenge on how police would interact with the driverless vehicles, for instance in the case of an accident or crime.
Nonetheless, all these obstacles can be overcome. The quality of sensors is being upgraded as technology advances to have fully capacitated cars.
Do you think self-driving cars are safe?
What if the system is hacked?