Computer Vision Plays A Vital Role In Amazon Delivery Drones
Posted By Anirudh Bhardwaj | 20-Jan-2017
The world’s biggest e-commerce giant, Amazon took the whole world by storm by launching its Prime Air services for faster delivery of products via high speed drones. As we all know, Prime Air has been a dream project of Amazon and the company had been working meticulously on this over the last couple of years. Finally on December 7, 2016, Amazon created history by successfully delivering its first order via Prime Air Delivery Drone in Cambridge, UK. However, it didn’t go too far away. The drone flew within a 5 miles radius, dropped the deliverables at the destination and then flew back. In this short aerial journey, it delivered an Amazon Fire TV box along with a back of popcorns. The entire journey was completed in 25 minutes and the delivery was made within the 13 minutes of the order being placed. Although it wasn’t a long journey and it didn’t go too far, it is no doubt a major achievement not only for Amazon but for the entire world.
Amazon is still dedicated toward making its Prime Air services better by leveraging the cutting-edge technologies and advanced IoT tools. Technically speaking, an Amazon Prime Air drone is an elegant combination of IoT, Computer Vision and Machine Learning. In this blog, you’ll know how Amazon built this beautiful masterpiece and how it works. So let’s get started!
Also read 4 Stupefying Inventions In Computer Vision.
Some Major Challenges
Amazon Prime Air is a magnificent example of how far we’ve come in the field of IoT and aeronautics. Although the concept of UAVs isn’t new and it’s quite common to spot UAVs flying and recording video footage in music concerts and soccer matches. But using them for product delivery is quite a challenging task. The flight largely depends on weather conditions and the obstacles in the path which is perhaps the biggest of all challenges. The Amazon Prime’s first drone delivery was executed in a remote village with no obstacles at all in its path. But assume the same situation in a metropolitan city where there are hundreds of buildings and skyscrapers. So in order to deliver a product, the drone should be adroit enough to change its path every time an obstacle is detected. Imagine that in a city like Manhattan. You can’t even imagine how many obstacles will be there which intensifies the possibility of collisions.
Computer Vision Is The Key
So that’s where Computer Vision comes to the rescue. As a matter of fact, Computer Vision acts as the key technology behind delivery drones and Amazon has been using them in its Prime Air services from the very beginning. It uses Computer Vision to differentiate between different objects and detecting and avoid small obstacles in path. For instance, let’s assume you have a small swimming pool in the backyard and right next to it you have a courtyard with blue colored flooring which is where the drone should drop the deliverables. Given that situation, there are equal chances that the product will be dropped in the swimming pool instead of the courtyard. But problems like these can be avoided using the advanced Computer Vision technology. This can be done by showing it all sorts of pictures of the objects and programming it recognize places, detect objects and change its path too if needed.
However, relying completely on Computer Vision and Machine Learning won’t suffice. There has to be a wise use of Neural Networks as well. Owing to the high scalability of the neural network, hundreds or even thousands of computers can operate in parallel and these computers can be used to give real-time updates to the drone during its aerial journey to ensure a safe delivery.