Google had been the frontier of Innovation since the past couple of decades. From Search engines to Operating systems, Internet balloons to Head Mounted displays, Google had been innovating in almost every facet of technology that made people’s lives easier.
One such innovation in its kitty is Google Self-Driving Car, an intelligent, self-driven vehicle that can take you places without the need for any human input. According to its monthly report, Google had test driven its fleet of driverless cars in autonomous mode a total of 1,842,496 miles (~2,965,209 KMs) as of July 2016.
Though truly fascinating, one question that ponders is what makes these smart cars truly smart? Which technologies made autonomous driving not a mere possibility but a huge success?
To answer these deep questions, we excavated all the patents related to Google’s self-driving cars and found some pretty exciting state of art technologies that makes these chauffer-less cars a success.
Want to know which technologies affect the certain function of a car? If yes, join me on this chauffer-less ride to figure out the why’s and how’s.
Fasten your seatbelts, here we go…
Object Detection and Identification
Unless you’re a famous Bollywood actor in a corrupted country, the chances of getting away after hitting pedestrians are pretty grim. That’s one reason why it is important for a driverless car to accurately detect objects in order to avoid collisions.
But, is detecting the objects at a distance just enough? No. When you’re a car on your own, it is important to not just detect objects but also identify them such that situation-specific actions could be taken.
Google’s self-driving cars uses a combination of RADAR, LIDAR, SONAR and Stereo Vision cameras that helps identify and detect objects both at long and short ranges.
But how does a particular object gets identified? There are thousand different things on the road including but not limited to- buses, cyclists, pedestrians, Special purpose vehicles, speed limit boards and even construction zone objects.
Could the smart car detect all this and maneuver accordingly?
Yes. According to the patents filed by Google, it is evident that Google has it all sorted.
According to the patent titled Bus detection for autonomous Vehicles, the processors of the smart car uses image recognition technology to identify if the object detected by the camera matches the size and color of a school bus. Once the parameters are met, the control system enters into a cautious driving mode such that collisions could be avoided.
Buses are huge. How about detecting the hand signals of a cyclist?
According to the patent titled Cyclist hand signal detection by Autonomous Vehicle, the car uses its sensors to calculate the height of an object and the algorithm then determines whether the height falls within the range of a pre-existing database of heights of cyclists that were identified previously. Once matched, the computers calculate the distance between the top of the head and pavements at various ranges to find its position on the road.
The sensor detects the distance between cyclists’s hand and head while camera would look at the angle at which the cyclist’s elbow is bending to figure out the hand movements of the cyclists. Further, it uses image recognition to identify the hand signals of passerby cyclists and manoeuvres accordingly.
Pretty cool, ain’t it?
But this is not it. Google even filed a patent to detect special-purpose vehicles and modifying the state of vehicle during its presence. This might mean the cars would give way to ambulances, police cars and emergency vehicles when on the same roads helping these special purpose vehicles continue on their path without blocking it.
Detecting and identifying objects is a big plus for these driverless cars but wouldn’t it be better if other cars in the fleet could use the knowledge gained by one car.
It would be great. That’s exactly what the developers of self-driving car thought while implementing a communication protocol among driverless car.
According to this patent application titled Reporting road event data and sharing with other vehicles, autonomous cars in the fleet receive information reports from the fleet of other AVs which is partly validated based on sensor data of the vehicle. The validated information is combined into a driving information map which is periodically filtered to remove outdated information reports, portions of which are communicated to other vehicles in the fleet.
On similar lines, data shared include construction zone data which is updated on the driving information roadmap so that other vehicles could take an alternate route avoiding delays.
The main function of a car is to reach a destination from source safely. When it comes to Google driverless cars, the inventors got it all covered, as the cars uses its sensory system to detect aggressive drivers, tailgaters and dynamic and static obstacles to with instructions to manoeuvre accordingly.
Further, the car uses map data to identify curves, lane boundaries and selecting movements accordingly. The self-driving car makes sure that the passengers inside are safely transported avoiding any collisions in the smoothest way possible, abiding by speed limits and traffic rules while avoiding blind spots of other vehicles.
Also, in case a pedestrian bumps into a vehicle despite every other measure taken, Google has patented a sticky layer on top of its car titled “Adhesive vehicle front end for mitigation of secondary pedestrian impact”, so that if a pedestrian bumps into a car, he would end up sticking on it, thereby avoiding any major injuries.
Making the Transition
There are times when a self-driving car is not capable enough to operate in autonomous mode and manual intervention is needed. It might be either due to lack of inadequate map data or an unknown location whose data is not available within the system.
In such cases, it is necessary to make the transition to manual mode but with care. To make sure that the driver operating the vehicle is fully aware of the location, he is made to complete a few tests before the vehicle is transitioned to manual mode.
How to find that a driver is not sleeping? By detecting the grip on Steering. The self-driving cars have steering enabled with touch sensors that detects the grip and only upon meting the threshold, the steering controls are activated.
Smart car, isn’t it?
When you’re a car that can drive on your own, it’s no big deal to maintain yourself. The self-driving car is hassle free as it has a passive wiper system that automatically cleans its sensors at periodic intervals.
Also, the car has a mechanism that tracks wear and tear which are scheduled for maintenance at specific intervals. Further, oil changes and repairs are taken care of by the fallback mechanism which makes the driverless car one hassle free solution in a life full of problems to solve.
Cool, ain’t it?
Feature Image Credit: TechCrunch