Have you ever asked yourself how birds play their extraordinary ability to fly, at speed, through environments with several obstacles (e.g. in a forest) ?

While flying, birds have to see obstacles and control their movements to avoid collisions, pursuing, at the same time, their route planning towards a target. Seen from traditional control theory, this is an hard task, even from a computation perspective. Actually, this is an interesting and challenging problem for engineers to develop robotic motion control laws.

Well, Nature solved it in a very simple way, by adopting a different perspective.

How birds can fly at speed through environments with several obstacles ?

In fact, Researchers at Boston University have revealed the secrets behind this extraordinary capability of birds. In this paper they argue that flying animals makes use of a relatively simple “algorithm”: the basic idea is to think of the field of view of the bird, not as a set of discrete objects at different distances, but as a moving array of points: the rate of movement depends on factors such as the size and distance of objects as well as the speed of flight. With this assumption, optics of eyesight simplifies the control task: the rate of change of the object’s image size on the eyeball retina determines the time to impact. This is like to say that no “complicated” knowledge of objects’ sizes, distances or even of speed is required (or has to be processed). That’s something that can be done with direct feedback from the optical system in a highly efficient way.

This is a fascinating result which could allow the development of relatively simple algorithms and control laws to navigate an environments with obstacles by simply using image data feedbacks.

I see interesting opportunities for developing autonomous systems.

Today most of the time you fly it is likely that a computer is actually flying the plane, and it actually feels pretty good. Landings are smoother than the ones where a human pilot is at the yoke…

Marvin, the car driven by a computer…

Technology is progressing rapidly and in a few more years, let’s say next decade, we may be sitting in the back seat of our car, letting a computer do the driving.

At least, this is what Peter Stone,  a computer science professor at the University of Texas at Austin claims. And to prove it, he shows the result of his research on a car, Marvin, driven by a computer that can manage intersections and negotiate with other cars.

According to Peter, traffic will be smoother and quicker if cars are driven by computers, and more important, it will be safer. A computer does not get distracted and is not in a “hurry”.

To demonstrate the effectiveness of his algorithm he has produced a simulation, shown in the video clip below. Well looking at it I am reminded of the approach to intersection of motorbikes in Hanoi and Saigon, and that doesn’t give me a good feeling.

Looking at the clip I feel very much uneasy at the idea of sitting in the back seat and let the computer take over the wheel. No relaxation at all. But this is today. May be, just may be, in the future I would go along with it.

Ever felt green with envy for those top brass moving around in nice car and reading a newspaper since the chaffeur is taking the burden of driving around?

Well, technology is promising to deliver us a personal chaffeur embedded in the car! Take a look at this video:

And also read the articles on driverless cars and on autonomous systems.

Clearly technology has made amazing progress. Today you can feel if the plane you are on is under the control of the auto-pilot or if a human pilot is flying it. And you can tell because the flight is so much smoother when the computers are in control…

Add to this the fact that in most flying accidents pilot’s mistakes are involved. Clearly, this does not prove anything. It does not provide any figure on the instances where no accident occurred because the pilot skill saved the day, and I suspect there may be many of these instances.

If everything is going according to plan I have no doubt that computers fare much better than us human being. We tend to become distracted, to lose focus and to act unpredictably. Not so with a well designed computer.  But is something goes not according to plans then the computer, that needs to have a plan to follow, is lost. And this is where we are still superior machine. Our brain embeds so many different plans and can select one out of many other in a blink of an eye, to cover with unexpected situations. Of course this is one way to see what is happening inside our skull. A different vision is to imagine that our brain is cable to create a new plan to fit the new circumstances…

These two visions have a correspondent in the Artificial Intelligence domain: for many years scientists have tried to find out ways to “invent” new plans within a computer with results that have been interesting but remained short of meeting the goal. More recently, the abundance of resources, data, processing… and the use of statistical approaches mining big data have provided a new way to extract (create?) information and take decision. In a way, it is much more like what is going on in our brain, according to most recent study. Rather than an automatic machine our brain is an autonomous system able to finely tune its responses by mining a multitude of data. The behavior is an emerging property of the way we accumulate and interact with data.

This is why I feel that the furthering of study on autonomous systems is key to  more “intelligent” future.

NASA is currently exploring autonomous and autonomic systems concepts directed towards enhancing future space-mission self-management and survivability in harsh operational environments. There are several publications about this research avenue and the related applications: for instance, NASA ANTS missions are targeted by this research.

Recently, I’ve read this paper describing the ASSL (Autonomic System Specification Language) as a framework for formally specifying and generating autonomic systems.

ASSL seems allowing modelling autonomic properties of system of systems through the specification of self-managing policies and service-level objectives.

Is this area of research so far away from designing autonomous and autonomic system of systems in future networks? I see striking similarities specifically when looking at the network edges, where we’ll see in the future swarm of autonomous and autonomic nodes supporting any sort of services by using local processing and storage resources.

I was reading Technology Review the other day and I stumble on an article reporting that a team of researchers managed to measure the processing speed of a human brain in responding to visual stimula, http://www.technologyreview.com/blog/arxiv/24030/

I was intrigued at two levels. The first goes back to my youth times when I studied physics and the various aspects of entropy. Of these, and there are many, I did not remember anyone related to measuring speed of some phenomena. This is what was used by a researcher of the University of Provence in France, Fermin Moscoso del Prado Martin -does not sound like a French name to me….

Relation between entropy (bit) and respose time in a human brain engaged in visual recognition

You may want to read the article to know more about this intriguing relation between entropy and measuring reaction time of our brain.

The second aspect that made me think was the amazingly slow speed at which our brain seems to process information, on the one hand, and the incredible results it can derive from such processing, on the other hand. We usually say we need broadband communications to match our visual sensor reception capability. How could it be that we need to transmit at several megabit/s to fool our eyes and our brain into believing it is looking at something real, when it processes visual information so slowly? Indeed, all comments from readers attached to the article were of disbelief. The unanimous outcry was that that scientist got it wrong.

Well, I do not know if he got it right or not but although the result is somewhat unbelievable it is …possible.

This has to do with the power of connectivity in a meshed network. The meaning, the perception we have of the world around us is not coming from an algorithm producing a single value, like 324 is red, 7864 is a rose,…rather it is a state of mind, or better the state of activation of a certain (huge) number of neurons. This activation may result from very few bits, and then create an avalanche leading to the activation of all that neurons’ network. It is like a bomb: it may take just one switch (two bits) to activate it and the results may be felt over a very large area.

It made me think about the similarity with business ecosystems. A tiny action on the part of a single player, generating in comparison very few bits of information, may trigger significant changes in a whole ecosystem and as a result the behaviour of several actors may change, even though there is no direct exchange of information from one to the other. It is the change of state in the business ecosystem that triggers the response.

We see this happening more and more as we move from value chains to ecosystems. The former has a sort of regulated information flow (regulated by contractual obligation from one ring in the chain to the next one and although the efficiency in the value chain may be high the time it takes to perculate to the end of the value chain may be significant), the latter is the result of a change of state and it propagates instantaneously to all actors in the ecosystem. Of course some of them will take notice and react, some others will disregard the information. In an ecosystem all actors are basically autonomous systems and react based not on communications with other actors but based on the change of context.

Take a look at this video:

 http://link.brightcove.com/services/player/bcpid1827871101?bctid=25954395001

It is about interactive building blocks invented by David Merrill, a Media Lab MIT researcher. Siftables, this is the name of the blocks, are small square tokens with a video screen on one of the faces and a sensing capabilities to recognize and talk to other Siftables nearby. Inside each token there is an accelerometer to measure movements and inclination. Based on this data and on the vicinity of other Siftables complex behaviour can emerge.

It is a good example of an ecosystem where each Siftable is a player and the type of interactions are dictated by the software embedded in each one. This software is characterising each token, one may be a container of colour, another of music, another of information. As an example you can have tokens displaying a number, one displaying an arithmetic operation and another an equal sign. By combining them and placing an empty token by the equal sign you can see the result of the operation.

The amazing thing is to look at the complex behaviour that can emerge from very simple tokens juxtapposed one another.

 This is not jus interesting from the point of view of modelling ecosystems, it is also interesting because it gives a glimpse on what it means to have interacting objects all around us and how easy it is to move seamlessly from atoms to bits.

In the future get ready to experience this in many environments, in your home and office to start with.