This post is a continuation of Google driverless car series that we are posting for quiet a while.
After going through today’s post, you will know the various parameters that Google driverless car use in order to differentiate between a bike rider, cyclist, pedestrian and other vehicles.
Detecting Vehicles, reading road signs in autonomous mode
The driverless car use its object detection system to identify, track and predict the movements of pedestrians, bicycles, other vehicles, or objects like road signs on the roadway.
The car computer makes strategies like applying brake, taking turn, giving pass to rear vehicle and the like based on the location and shape of the objects.
This information is also used to change the vehicle’s current path and speed based on the presence of detected road signs. For example, the vehicle will be capable of automatically slowing down from 50 mph to 35 mph if it come across a road sign indicating the speed limit is 35 mph.
Apart from that, the autonomous car computer simultaneously uses the data stored in its memory to predict the location of road sign, traffic signals, intersections and controls the vehicle accordingly.
So you can say that in future, your car will automatically control its speed so that it always gets Green in the traffic lights.
Predicting Expected Movement of Vehicles on Road
Google Robotic Chauffeur can also predict expected movement of a detected object.
First of all, the car determines its type. For example, whether the detected object is a traffic cone, person, car, truck or bicycle, and later use this information to predict object’s future behavior.
The type of object is determined by considering various characteristics like size, shape, speed or the amount of heat coming out from detected object.
For example, to differentiate between a car and a bicycle, the computer use following parameters
- A bicycle is larger than a breadbox and smaller than a car,
- the speed of the bicycles do not tend to go faster than 40 miles per hour or slower than 0.1 miles per hour,
- the heat coming out from bicyclist is less than coming out from engine of a car.
In addition, the objects are classified based on their specific attributes; such as information contained on a license plate, bumper sticker, or logos that appear on the vehicle. For example, after detecting Kenworth’s logo the computer identifies the vehicle as tractor trailer and the logo of Harley-Davidson signifies that it’s a bike .
After detecting and determining the type of object, the computer searches its database and extract the behavior pattern followed by a particular type of vehicle on the road. The car computer already stores behavior data of different vehicles and objects on the road uses it to make control strategies.
For a way of example, a bike is likely to react differently than a tractor-trailer in a number of ways. Specifically, a bike is more likely to make erratic movements when compared with a tractor-trailer.
Like a driver, the computer of driverless car also becomes intelligent by spending more time on road and by collecting more data.
The vehicle computer uses its object detection sensors and other sensors to detect the behavior of moving objects like vehicle, pedestrians and bicycles on road.
It keeps a check on their speed and directions with respect to each other. Apart from that, it also analyses that how frequently a vehicle is changing its heading and based on all these parameter controls a vehicle.
For example, in figure-1 , on the right hand side, a car(blue in color) and a truck are moving in same lane while the driverless is moving in different lane.
From the movement of blue colored car, the computer will estimate that it will overtake truck within next few seconds (figure-2).
The autonomous car also wants to overtake the slow moving truck. However, it has determined that further acceleration can make it to collide with the blue car.
Hence, it will maintain its current speed and will let blue car to overtake the slow moving truck first.