The Innovator’s DNAInnovation | July 11, 2014
The Innovator’s DNA Being an innovator requires a mindset to...
From next-gen safety features, to remote-sensing technology and state-of-the-art navigation working on the Mahindra Rise Prize – Driverless Car Challenge will be an ambitious technological adventure. Here’s a tip before you begin the challenge: Think of everything you can do while driving a car—and automate all of that!
If you’re reading this, we know you are confident about two things – driving cars and building them. You’re probably aware ofLIDAR, the US $70,000 laser radar system atop theGoogle Self-Driving carthat is used to capture 3D images of the environment. If you’re aware of that, perhaps you also know about theinfrared camerafitted into the 2014 Mercedes S-Class. However, Google is not the only company in the world working on robotic cars; almost every large car manufacturer – including BMW, Volvo and Audi, to name a few – has now assembled an A-team of researchers to automate several aspects of the driving process.
In this post, we aim to enable you think to through the Driverless Car Challenge in a methodical fashion. In an earlier post, I said: “If you can solve problems x, y and z, then you have a breakthrough product.” For the Driverless Car Challenge, the x, y and z translates into braking, steering and accelerating – all three done automatically without driver intervention.
Take a look at the various technologies you need to master to build a winning prototype for the Driverless Car Challenge.
The braking problem
Imagine sitting in the passenger seat of a driverless car somewhere in India. The car needs to brake in several situations – when the traffic light is red, when it’s too close to the car ahead, when another vehicle or pedestrian cuts in front.
What should you do, to automate the braking process? Here’s a quick list, which you can further research upon. Papers published by The Center of Automotive Research by Stanford University will be a good place to start.
a) Sensors: That can be connected to your wheel or bumper, to detect objects close to your car. As of today, BMW is building prototypes that use ultrasonic and radar sensors to detect objects.
b) 3D imaging device: Google uses LIDAR – the spinning laser radar system to capture 3D images of the environment, to input into the overall data model, and feed into the various algorithms.
c) Cameras: Video, stereo and infrared cameras – that spot road markings, traffic signals and direction signs. Stereo cameras specifically help with capturing 3D images.
d) Braking algorithm: Finally, you need to design a braking software algorithm, to take in all this data from your sensors, scanners and cameras; crunch the data and instruct the car to brake accordingly.
The acceleration problem
Cracking the acceleration problem requires an understanding of traffic patterns, accessing accurate maps and maintaining speed limits. In addition, this technology will also require greater levels of efficieny while driving, meaning smoother starts and stops. Many of the technologies (cameras, sensors, 3D imaging) used to solve the braking problem will be reused to solve the acceleration problem as well.
However, early research done by large automobile companies suggests the crucial element in all of this, is the algorithm’s ability to take split-second decisions, based on inputs from the sensors, cameras and navigation systems.
Experts also believe that robotic cars need to be fitted with an additional computer just to keep track of the other automated systems and take control, if all else fails.
e) The acceleration (and deceleration) algorithm: The algorithm needs to understand the science t of passing other cars, switching lanes and letting other cars pass, even as it figures out the speed at which to do this.
Like in the braking situation, you need to understand what objects are around the car – front, rear, sides and blind spots – and take a call accordingly. This algorithm requires skills in data crunching and developing mathematical models to act upon this data.
The steering problem
The obvious goal of building a driverless car is to enable people to move from point A to point B, without any trouble. The technologies involved in cracking the steering problem include the ability to read maps, road directions and help passengers reach their destination.
f) GPS receiver and navigation systems: The car needs an inbuilt navigation system to understand the car’s position and the ability to navigate turn-by-turn. Google Maps and companies likeMapMyIndia have already made we’ve made reasonable progress in this area. The next step is to get your driverless car to read and understand these maps.
Some global car companies, which are now exploring partial-automation technologies, use a differential GPS receiver, which combines signals from ground-based stations and satellites to know where exactly a particular car is on the route map. Knowing this position helps instruct the robotic driver on its next move.
To conclude, we’d like to reiterate that the above list is only a roundup of technologies used by global car companies, to automate select parts of the driving process. Building a completely driverless car requires the ability to make all of this work in tandem. If you plan to contest in the Driverless Car Challenge, you need to assemble your A-team with expertise in mechatronics, software and algorithms, mathematical modelling and automobile engineering.
Lead the way to make the world’s foremost driverless car. Go beyond the box and reset the rules of travel!
If you enjoyed this article, keep up on new stories and conversations about Rise×