Monday, 13 June 2011

The Key to Automation

Every day seems to see new reports about incredible robots being developed. Robots that carry stuff for soldiers, robots that perform surgery, and robots that play football. It's all very exciting, but what we really need are robots to make our lives easier. To give us more time to do what we want. To do the mundane jobs and free us up to take on more creative work that robots can't do.

But how do we go about this? How do we bring robots and automation into our everyday lives, unless robot intelligence is significantly improved? Well, while this intelligence is improving at a rapidly accelerating pace, there is a huge potential for solving problems using simple, task repeating, programmable robotics.

The key is to standardize everything. We have to "put it on rails".

As an example, let's look at making an automated dish washing system.

We have dishwashers, but we still need to load and unload them. We need to develop 2 things. Firstly, we need a robot that can safely sort dishes no matter how randomly they're piled up, and insert them into their relevant areas of the dishwasher. Secondly, and this wouldn't be so difficult once the first stage was completed, is a mechanism for unloading the dishes and putting them away.

The important thing to note is that the kitchen would need to be designed with this in mind. In many solutions, we may need to consider our existing infrastructure. This needn't be as complex as it's made out to be, and would always pay off once a working automated system was integrated.

So you would need containers either side of the dishwasher, one for humans to pile up the dirty dishes, and one where the clean dishes would be stored permanently. This would make it easier than if the storage cupboard was, say, on the other side of the room to the dishwasher.

The next step, we could consider 2 options. The first is to standardize all our plates, cups, dishes, pots, pans, and cutlery, so that the loading robot would "recognize" them all. Alternatively, we may be able to develop robots that can recognize new items and program themselves "on the fly" to deal with them effectively. This would obviously be a little harder. To achieve the first step, all the dishes we use in a house could be embedded with RFID. The robot could also have powerful sensors similar to face recognition software many cameras currently have.

Within the dishwasher, every item would have a specified place. The robot would simply grab each item and move it to its "cleaning slot". Any odd items could occupy a separate platform within the dishwasher.

You get the idea. While it still has some creases to iron out, the point is that there is plenty that can be achieved with a minimal level of robot intelligence, if we standardize our practices and the environments that automation functions in. The same theory could be applied to transportation. It's baffling why trains are still driven by human beings, when a computer could control them perfectly.

If you think that giving computers so much responsibility is dangerous, it's because you've been conditioned to seeing machines that have been challenged by real life scenarios. The point is, these machines have been limited by the programming of humans. They have been put in situations where they have not been designed to cope with all possible scenarios. They were expected to work like humans, yet they were limited in the number of possible actions. Automation system designers need to limit the scenario to a set routine, as well as limiting the influence of external factors.

To put it another way, they need to simplify what a robot has to do, and design its environment to confine it, protecting it from the need to make decisions. We can do this by standardizing its interactions. This will allow us to bring automation into our lives in more ways than we ever thought possible, even at current technology levels.

Post a Comment