Amazon Robotics — My experience @ the AWS Summit
Recently I was fortunate to attend the AWS Summit in Sydney.
Amazon is a great organisation and the summit had good food, gifts, and most importantly, it gave us new ideas and thoughts on technology trends. One thing that caught my eye is Amazon Robotics. It’s a warehouse management system utilising robots, path finding, and perhaps a little bit of AI as well.
Basically, it’s an auto shelf. Pretty easy to understand, instead of letting you walk to the shelf to get the packages, the shelf will come to you and deliver.
The carrier bot and the shelf are separated, a very smart design and classic decoupling. This makes the shelves’ cost low to build and the robot flexible to re-use and manage. When you look at the floor, there are many points, lines and path/maps for the robot. The robot would likely have some sensors on the bottom to locate these and find the path.
The management system could save and remember all shelves number, locations and control all robots to move the shelves efficiently.
https://www.amazonrobotics.com/#/vision#inter-content
The point of this post is not explaining how this system works but how easy people with common tech knowledge can understand this system, however what about the cost and risk?
Why do I raise this point? People are usually focused on innovation and new tech, like AI, machine learning, but tend to ignore the cost and risk of the innovation.
Amazon Robotics is coming from a very specialised case that only Amazon has when it reaches certain warehouse level/scale. For small and medium warehouses, I doubt the investing and implementation costs compare to the existing costs. Tech cost will become relatively cheap and stable after matured, before that, a company might need to think about how much they want to spend to try or adapt a new technology.
Amazon comes up with a solution to solve their own problem, and when they want to sell this solution to others, they need to make it generic. There are many special cases that require certain sets or types of technology, but when it comes to marketing, all these different ideas will find a common term which is the most famous, adopted or eye catching name. What’s really behind AI or machine learning, for different solution providers, they might actually say very different things.
There are some companies that are actually doing this and doing it very well. Microsoft used to be a so called big follower that copied all the cool (but not very profitable stuff) to make it not so cool, but a commercial success. Even today, the cloud service is lead by Amazon, followed by Microsoft. We are excited about the success stories of innovation, but how about the failures? What’s the price of wrong innovation? And how about a smart or mediocre follower? What’s their position?
To be a smart follower usually costs less. The true innovation comes from the recognition of a company’s core problem to be solved which almost always is a special case. There’s gap to apply a generalised technology, not to mention if that tech term is more about marketing than the real value. There’s cost and difficulties to connect the new tools to a company’s long existing issue. In other words, the ability to solve a special problem makes a company’s true value or makes a company a specialist.
So, the take away here is:
- Don’t simply try to use a new trend tech without recognising the core problem you want to solve.
- The new tech has a cost in the early stages and a high risk to fail.
- How “smart” a follower is depends on their ability of recognising the core problem and picking up the cheap, matured tools to solve it.
Words by Mason Hu
