Showing posts with label evolution. Show all posts
Showing posts with label evolution. Show all posts

Monday, December 28, 2020

She had chosen well

For this story, written as the second exercise in my Writing Short Stories course back in June, I attempted a story without dialogue. I love dialogue so expected to find this difficult, which it was. In the story I try to imagine what it might have been like to experience an extinction event, in an effort to capture a sense of being in the liminal state from a limited first-person (or rather animal) perspective.


     She had chosen well.  

    The burrow she shared with her litter was lodged within the vaulted foundations of a mighty tree. The tree had taken root in rocky soil long before her time, its vascular organs splitting the rock enough to allow her to excavate tunnels and chambers three seasons ago. 

    It had been a good spring. Her pups had almost weaned and were growing fat on insects and berries. Even the reckling was looking healthy. He was a survivor, escaping the quick-feathered hunters with sharp eyes and sharper teeth that had taken two of her litter a few moons ago.

    In her world there was much to fear. Death came in many ways: quick from the sharp-teeth or sky-claws; slow from starvation or thirst (the nearest spring was a perilous journey - although she had learned from her mother how to harvest the prickly watery green leaves which grew close to the burrow). But this hillside had one advantage; it was too high and steep for the long-necked ground-shakers that crashed and bellowed through the valley below from time to time.

    The moons passed and, as the nights started to lengthen, she began to harvest the nuts, green leaves and tubers, storing these in dry clean chambers close to the comfortable living nest.  Something – perhaps the unusual bounty of the season – made her collect more this summer.

    It was a warm dusk. After a good night’s forage she and her pups had spent the day sleeping full-bellied in the cool of the burrow. Her pups were now almost full grown and the biggest and boldest were restless to leave. Two, a brother and sister, moved to the burrow entrance with a purpose that she knew from her own time so, with a touch of their noses, mother and eldest made their farewells.  

    Then, just a few moments after she had returned to the nest chamber, the ground shook. But this was not the rhythmic shaking of the long-necks in the valley.  Nor was it the noisy anger of the fire mountain that turned their nights red from time to time. This was different: a silent deep tremor that felt as if it was coming from the belly of the earth. The tremor grew to a crescendo. Terrified the small family nest-huddled as the tree roots groaned while soil and stones rained upon them. Then it was still.

    They waited. She lifted her head and sensed around. The nest air was full of dust. She felt the silence then realised that the breeze-scent of outside was gone. She knew something was wrong, ran to the entrance tunnel and found it blocked with stones and earth. Fear rising she started to dig. She was a good digger with powerful front claws. She dug and dug until she started to feel weak, then – rest-pausing – she heard a scraping sound. A few moments later the soil and stones ahead broke apart and there was her eldest daughter. With joy and relief they touched noses, but she sensed a sadness that told her that her eldest son was gone. 

    Together mother and daughter cleared the spoil from the entrance tunnel, then – followed by the rest of the pups – they emerged, cautiously, into the night. There was no moon. Instead the sky clouds were lit high with lurid reds, greens and purples, yet – she noticed – the fire mountain was silent. The night was quiet at first although some familiar sounds slowly returned: the bellows of the long-necks in the valley below and skyward the distant cries of the sky-claws. The family fed and foraged and still fearful returned to the nest before dawn.

    After sleeping most of the day the nest family was awakened by a long roar of thunder that seemed to roll in from afar and rush over them before receding into the distance. She had heard thunder before but never like this. As it passed it hit their tree – although not with the long shake of the sleep-day before – but with a great cracking crash that was the last thing they heard for awhile. She felt an ear-pain she had never before experienced, and so – it seemed – had her pups. Dazed, deafened and frightened they did not venture out of the burrow that night.

    Restless and hungry the family stirred again before dusk the following day. She was relieved that the ear-pain had gone and her sound sense restored. Cautiously they emerged from the burrow entrance to find that their small exit platform was now a tangle of branch and leaf. Luckily it was not dense, and they quickly made a path through to the open hillside.  What they saw by the dull grey light of dusk was a world changed. No tree was left standing, including their home tree – indeed it was that tree that now provided their exit canopy.

    They sensed something moving nearby, then saw one of the sky-claws fallen onto a prickle leaf bush; it was broken winged and near death, but still able to fix them with its sharp eye. They had never before seen one of these creatures close up and – even in its death throes – their terror of its kind was undimmed, so they quickly retreated into the exit canopy and nervously fed on insects and home tree nuts.

    The next two nights, alerted by the bad tempered chirruping of sharp-teeth feeding on the sky-claw, they did not stray outside the home thicket. She noticed that the nights were cold: too cold for this early in the autumn. A few nights later the sky-claw was joined in death by the sharp-teeth, and the nest family were able to feast on the insects drawn to the carrion. But their forages were short as it was too cold to stay out for more than a few mouthfuls before returning to the warm of the nest. A few nights later even the carrion insects were gone, as the corpses had frozen. 

    With a deep sense of unease the nest family settled for their long winter sleep.


© Alan Winfield 2020


Previous stories:

The Gift (2016)

Word Perfect (2020)

Monday, April 20, 2020

Autonomous Robot Evolution: an update

It's been over a year since my last progress report from the Autonomous Robot Evolution (ARE) project, so an update on the ARE Robot Fabricator (RoboFab) is long overdue. There have been several significant advances. First is integration of each of the elements of RoboFab. Second is the design and implementation of an assembly fixture, and third significantly improved wiring. Here is a CAD drawing of the integrated RoboFab.

The ARE RoboFab has four major subsystems: up to three 3D printer(s), an organ bank, an assembly fixture and a centrally positioned robot arm (multi-axis manipulator). The purpose of each of these subsystems is outlined as follows:
  • The 3D printers are used to print the evolved robot’s skeleton, which might be a single part, or several. With more than one 3D printer we can speed up the process by 3D printing skeletons for several different evolved robots in parallel, or – for robots with multi-part skeletons – each part can be printed in parallel.
  • The organ bank contains a set of pre-fabricated organs, organised so that the robot arm can pick organs ready for placing within the part-built robot. For more on the organs see previous blog post(s).
  • The assembly fixture is designed to hold (and if necessary rotate) the robot’s core skeleton while organs are placed and wired up.
  • The robot arm is the engine of RoboFab. Fitted with special gripper the robot arm is responsible for assembling the complete robot.
And here is the Bristol RoboFab (there is a second identical RoboFab in York):


Note that the assembly fixture is mounted upside down at the top front of the RoboFab. This has the advantage that there is a reasonable volume of clear space for assembly of the robot under the fixture, which is reachable by the robot arm.

The fabrication and assembly sequence has six stages:
  1. RoboFab receives the required coordinates of the organs and one or more mesh file(s) of the shape of the skeleton.
  2. The skeleton is 3D printed.
  3. The robot arm fetches the core ‘brain’ organ from the organ bank and clips it into the skeleton on the print bed. This is a strong locking clip.
  4. The robot arm then lifts the core organ and skeleton assemblage off the print bed, and attaches it to the assembly fixture. The core organ has metal disks on its underside which are used to secure the assemblage to the fixture with electromagnets.
  5. The robot arm then picks and places the required organs from the organ bank, clipping them into place on the skeleton.
  6. Finally the robot arm wires each organ to the core organ, to complete the robot.



Here is a complete robot, fabricated, assembled and wired by the RoboFab. This evolved robot has a total of three organs: the core ‘brain’ organ, and two wheel organs.
Note especially the wires connecting the wheel organs to the core organ. My colleague Matt has come up with an ingenious design in which a coiled cable is contained within the organ. After the organs have been attached to the skeleton (stage 5), the robot arm in turn grabs each organ's jack plug and pulls the cable to plug into the core organ (stage 6). This design minimises the previously encountered problem of the robot gripper getting tangled in dangling loose wires during stage 6.

And here is a video clip of the complete process:



Credits

The work described here has been led by my brilliant colleague Matt Hale, very ably supported by York colleagues Edgar Buchanan and Mike Angus. The only credit I can take is that I came up with some of the ideas and co-wrote the bid that secured the EPSRC funding for the project.

References

For a much more detailed account of the RoboFab see this paper, which was presented at ALife 2019 last summer in Newcastle: The ARE Robot Fabricator: How to (Re)produce Robots that Can Evolve in the Real World.

Related blog posts

First automated robot assembly (February 2019)
Autonomous Robot Evolution: from cradle to grave (July 2018)
Autonomous Robot Evolution: first challenges (Oct 2018)

Wednesday, July 31, 2019

On the simulation (and energy costs) of human intelligence, the singularity and simulationism

For many researchers the Holy Grail of robotics and AI is the creation of artificial persons: artefacts with equivalent general competencies as humans. Such artefacts would literally be simulations of humans. Some researchers are motivated by the utility of AGI; others have an almost religious faith in the transhumanist promise of the technological singularity. Others, like myself, are driven only by scientific curiosity. Simulations of intelligence provide us with working models of (elements of) natural intelligence. As Richard Feynman famously said ‘What I cannot create, I do not understand’. Used in this way simulations are like microscopes for the study of intelligence; they are scientific instruments.

Like all scientific instruments simulation needs to be used with great care; simulations need to be calibrated, validated and – most importantly – their limitations understood. Without that understanding any claims to new insights into the nature of intelligence – or for the quality and fidelity of an artificial intelligence as a model of some aspect of natural intelligence – should be regarded with suspicion.

In this essay I have critically reflected on some of the predictions for human-equivalent AI (AGI); the paths to AGI (and especially via artificial evolution); the technological singularity, and the idea that we are ourselves simulations in a simulated universe (simulationism). The quest for human-equivalent AI clearly faces many challenges. One (perhaps stating the obvious) is that it is a very hard problem. Another, as I have argued in this essay, is that the energy costs are likely to limit progress.

However, I believe that the task is made even more difficult for two further reasons. The first is – as hinted above – that we have failed to recognize simulations of intelligence (which all AIs and robots are) as scientific instruments, which need to be designed, operated and results interpreted, with no less care than we would a particle collider or the Hubble telescope.

The second, and more general observation, is that we lack a general (mathematical) theory of intelligence. This lack of theory means that a significant proportion of AI research is not hypothesis  driven, but incrementalist and ad-hoc. Of course such an approach can and is leading to interesting  and (commercially) valuable advances in narrow AI. But without strong theoretical foundations, the grand challenge of human-equivalent AI seems rather like trying to build particle accelerators to understand the nature of matter, without the Standard Model of particle physics.

The text above is the concluding discussion of my essay On the simulation (and energy costs) of human intelligence, the singularity and simulationism, which appears in an edited collection of essays in a book called From Astrophysics to Unconventional Computation. Published in April 2019, the book marks the 60th birthday of astrophysicist, computer scientist and all round genius, Susan Stepney.

Note: regular visitors to the blog will recognise themes covered in several previous blog posts, brought together in I hope a coherent and interesting way.

Thursday, February 21, 2019

First automated robot assembly

This month saw the first important milestone toward Autonomous Robot Evolution: the Bristol and York team demonstrated automated assembly of a complete working robot, from evolved and 3D printed parts. In essence we demonstrated one robot assembling another.

Our evolved robots consist of 3 elements:

* pre-designed modules which we call organs (for sensors, actuators, controllers, etc),
* an evolved and 3D printed skeleton, and
* cables (with 3.5mm jack plugs) to connect the organs and the controller.

Note that the organs are not evolved but hand designed; the rationale for this approach is outlined here.

Here are 3 basic organs:

On the left is a sensor, in the middle a controller and on the right a motor + wheel assembly.









And here are screenshots from the video showing the steps involved:











Step 1 shows the skeleton in the process of 3D printing. In step 2 the skeleton has been manually moved from the print bed onto the assembly area: note the organ and cable bank at the back of the assembly area. Step 3 shows the robot arm inserting the organs into the skeleton. Step 4 shows the robot arm connecting the cables. Step 5 shows the wheels being manually added, and in step 6 the robot is complete. Step 7 shows the assembled robot powered and running.

And here is the complete video:



Our aim is of course to automate the whole process and right now the team are working on the two problems of (1) how to remove the 3D printed skeleton from the print bed ready for transfer to the assembly area, and (2) how best to secure the skeleton in the assembly area ready for the processes outlined above.



Related blog posts:

Autonomous Robot Evolution: from cradle to grave (July 2018)
Autonomous Robot Evolution: first challenges (Oct 2018)

Friday, October 26, 2018

Autonomous Robot Evolution: first challenges

Just spent an exciting two days at the first 'all hands meeting' of our new EPSRC funded project: Autonomous Robot Evolution (ARE): from cradle to grave (read here for an introduction). There are eleven of us in total: 4 postdocs (one from each partner university), 2 PhD students, 1 technician, and the four seniors (co-investigators).


Much of the meeting was spent discussing the fundamental (and tough) questions of (1) how we design the genotype, and the mapping between genotype and phenotype, and (2) how exactly we will physically create the robots. 

Let me outline where we are going with these two questions.

1. Genotype-phenotype mapping. As I explained here most evolutionary robotics research has, to date, used a direct mapping approach, in which each parameter of a robot's genome specifies one feature of the real robot (phenotype). For the robot's controller those parameters might be the weights of the robot's artificial neural network, and for the robot's body they might each specify some physical characteristic of the body (such as the length of leg segments in the illustration here). Of course in biology the mapping is indirect; to put it very simply, the genome determines how an organism develops, rather than the organism itself. And because the expression of genes is affected by the environment in which the organism is developing, identical genotypes give rise to non-identical phenotypes (albeit very similar as with identical twins); this is called phenotypic plasticity.

Because we are looking for both biological plausibility and phenotypic plasticity in this project, we have decided on an indirect mapping from genotype to phenotype. Exactly how this will work is still to be figured out, but I feel sure the genotype will need to be split into two parts: one for the robot's controller and the other for its body, and I rather suspect the mapping will be different for those two parts.

2. How to create the robots. In ARE we will adopt the engineering approach 'in which the process is embodied but takes place in a kind of evolution factory'. Now, in theory we could evolve every part of a robot's hardware, listed below.


But in practice this would be impossible; evolving any one of these subsystems would be a research project in its own right, and we're not attempting to re-run the whole of evolution in this project. Instead we will be designing and fabricating discrete modules for sensing, signalling, actuation and control, that we call 'organs'. So what will we actually evolve? It will be:
  • the number, type and position of sensing, signalling and actuation subsystems, and
  • the 3D shape of the robot’s physical structure or chassis.
At this point you're probably thinking: hang on a minute - if you're designing the organs then what's left to evolve? It's a fair question, but in fact evolution will still have huge freedom to choose which and how many organs and where to position them in the body. And when we bear in mind that we will be co-evolving the robot's controller then the space of all possible phenotypes is vast. Of course we may need to introduce some constraints: for instance that there must be at least one controller. But in general we want as few constraints as possible so that evolution is free to explore the phenotypic space to find the best robots, bearing in mind that we will be breeding robots to be able to operate in challenging environments.

And I would argue that in specifying and designing organs we have not compromised on biological plausibility at all. Biological evolution is, after all, highly modular. Most of the organs (and systems of organs) in your body were evolved long before hominids: livers, hearts, eyes, noses, vascular systems, digestive systems, central nervous systems; all of those evolved in early vertebrates (with some repurposing along the way*). Architecturally humans have a huge amount in common with all mammals. My dog is not so different from me (and in some aspects superior: her senses of hearing and smell are much better); our key differences are in morphology and intelligence. These are the two properties that we will be exploring through co-evolution in this project.

So, in the coming few months we have some big mechanical and electronic engineering challenges in this part of the project. Here are just a few:
  • experiment with 3D printing materials and print heads,
  • specify, design and prototype the organs (including their packaging and interconnects),
  • decide on how to power the organs (i.e. a single central power organ, or a battery per organ) and figure out how to re-charge the batteries,
  • determine how to connect the organs with the controller and each other (i.e. with wires or wirelessly), and
  • work out the best way of picking and placing organs within the robot as it is 3D printed.
Challenging? For sure, but we have a wonderful team.

*See Neil Shubin's wonderful book Your Inner Fish.

Related blog posts:

Saturday, July 07, 2018

Autonomous Robot Evolution: from cradle to grave

A few weeks ago we had the kick-off meeting, in York, of our new 4 year EPSRC funded project Autonomous Robot Evolution (ARE): cradle to grave. We - Andy Tyrrell and Jon Timmis (York), Emma Hart (Edinburgh Napier), Gusti Eiben (Free University of Amsterdam) and myself - are all super excited. We've been trying to win support for this project for five years or so, and only now succeeded. This is a project that we've been thinking, and writing about, for a long time - so to have the opportunity to try out our ideas for real is wonderful.

In ARE we aim to investigate the artificial evolution of robots for unknown or extreme environments. In a radical new approach we will co-evolve robot bodies and brains in real-time and real-space. Using techniques from 3D printing new robot designs will literally be printed, before being trained in a nursery, then fitness tested in a target environment (a mock nuclear plant). The genomes of the fittest robots will then be combined to create the next generation of ‘child' robots, so that – over successive generations – we will breed new robot designs in a process that mirrors the way farmers have artificially selected new varieties of plants and animals for thousands of years. Because evolving real robots is slow and resource hungry we will run a parallel process of simulated evolution in a virtual environment, in which the real world environment is used to calibrate the virtual world, and reduce the reality gap*. A hybrid real-virtual process under the control of an ecosystem manager will allow real and virtual robots to mate, and the child robots to be printed and tested in either the virtual or real environments.

The project will be divided into five work packages, each led by a different partner: WP1 Evolution (York), WP2 Physical Environment (UWE), WP3 Virtual Environment (York), WP4 Ecosystem Manager (Napier) and WP5 Integration and Demonstration (UWE).

Here in the Bristol Robotics Lab we will focus on work packages 2 and 5. The goal of WP 2 is the development of a purpose designed 3D printing system – which we call a birth clinic – capable of printing small mobile robots, according to a specification determined by a genome designed in WP1. The birth clinic will need to pick and place a number of pre designed and fabricated electronics, sensing and actuation modules (the robot’s ‘organs’) into the printing work area which will be over printed with hot plastic to form the complete robot. The goal of WP5 will be to integrate all components, including the real world birth clinic, nursery, and mock nuclear environment with the virtual environment (WP3) and the ecosystem manager (WP4) into a working demonstrator and undertake evaluation and analysis.

Here is an impression of what the birth clinic might look like















One of the most interesting aspects of the project is that we have no idea what the robots we breed will look like. The evolutionary process could come up with almost any body shape and structure (morphology). The same process will also determine which and how many organs (sensors, actuators, etc) are selected, and their positions and orientation within the body. Our evolved robot bodies could be very surprising indeed.

And who knows - maybe we can take a step towards Walterian Creatures?


*Anyone who uses simulation as a tool to develop robots is well aware that robots which appear to work perfectly well in a simulated virtual world often don't work very well at all when the same design is tested in the real robot. This problem is especially acute when we are artificially evolving those robots. The reason for these problems is that the model of the real world and the robot(s) in it inside our simulation is an approximation. The Reality Gap refers to the less-than-perfect fidelity of the simulation; a better (higher fidelity) simulator would reduce the reality gap.

Related materials

Article in de Volkskrant (in Dutch) De robotevolutie kan beginnen. Hoe? Moeder Natuur vervangen door virtuele kraamkamer (The robot evolution can begin. How? Replacing Mother Nature with virtual nursery), May 2018.

Eiben and Smith (2015) From evolutionary computing to the evolution of things, Nature.

Winfield and Timmis (2015) Evolvable Robot Hardware, in Evolvable Hardware, Springer.

Eiben et al. (2013) The Triangle of Life, European Conference on Artificial Life (ECAL 2013).

Monday, December 22, 2014

Robot Bodies and how to Evolve them

Evolutionary robotics has been around for about 20 years: it's about 15 years since Stefano Nolfi and Dario Floreano published their seminal book on the subject. Yet, surprisingly the number of real, physical robots whose bodies have been evolved can be counted on the fingers of one hand. The vast majority of ER research papers are concerned with the evolution of robot brains - the robot's control system. Or, when robot bodies are evolved often the robot is never physically realised. This seems to me very odd, given that robots are real physical artefacts whose body shape - morphology - is deeply linked to their role and function.

The question of how to evolve real robot bodies and why we don't appear to have made much progress in the last 15 years was the subject of my keynote at the IEEE International Conference on Evolvable Systems (ICES 2014) in Orlando, a week ago. Here are my slides:



The talk was in three parts.

In part one I outlined the basic approach to evolving robots using the genetic algorithm, referring to figure 18: The four-stage process of Evolutionary Robotics, from chapter 5 of my book:

I then reviewed the state-of-the-art in evolving real robot bodies, starting with the landmark Golem project of Hod Lipson and Jordan Pollack, referencing both Henrik Lund and Josh Bongard's work on evolving Lego robots, then concluding with the excellent RoboGen project of Josh Auerbach, Dario Floreano and colleagues at EPFL. Although conceptually RoboGen has not moved far from Golem, it makes the co-evolution of robot hardware and controllers accessible for the first time, through the use of 3D-printable body parts which are compatible with servo-motors, and a very nice open-source toolset which integrates all stages of the simulated evolutionary process.

RoboGen, Golem and, as far as I'm aware, all work on evolving real physical robot bodies to date has used the simulate-then-transfer-to-real approach, in which the whole evolutionary process - including fitness testing - takes place in simulation and only the final 'fittest' robot is physically constructed. Andrew Nelson and colleagues in their excellent review paper point out the important distinction between simulate-then-transfer-to-real, and embodied evolution in which the whole process takes place in the real world - in real-time and real-space.

In part two of the talk I outlined two approaches to embodied evolution. The first I call an engineering approach, in which the process is completely embodied but takes place in a kind of evolution factory; this approach needs a significant automated infrastructure: instead of an manufactory we need an evofactory. The second approach I characterise as an artificial life approach. Here there is no infrastructure. Instead 'smart matter' somehow mates then replicates offspring over multiple generations in a process much more analogous to biological evolution. This was one of the ambitious aims of the Symbrion project which, sadly, met with only limited success. Trying to make mechanical robots behave like evolving smart matter is really tough.

Part three concluded by outlining a number of significant challenges to evolving real robot bodies. First I reflect on the huge challenge of evolving complexity. To date we've only evolved very simple robots with very simple behaviours, or co-evolved simple brain/body combinations. I'm convinced that evolving robots of greater (and useful) complexity requires a new approach. We will, I think, need to understand how to co-evolve robots and their ecosystems*. Second I touch upon a related challenge: genotype-phenotype mapping. Here I refer to Pfeifer and Bongard's scalable complexity principle - the powerful idea that we shouldn't evolve robots directly, but instead the developmental process that will lead to the robot, i.e. artificial evo-devo. Finally I raise the often overlooked challenge of the energy cost of artificial evolution.

But the biggest challenge remains essentially what it was 20 years ago: to fully realise the artificial evolution of real robots.


Some of the work of this talk is set out in forthcoming paper: AFT Winfield and J Timmis, Evolvable Robot Hardware, in Evolvable Hardware, eds M Trefzer  and A Tyrrell, Springer, in press.

*I touch upon this in the final para of my paper on the energy cost of evolution here.

Saturday, July 19, 2014

Estimating the energy cost of evolution

Want to create human-equivalent AI? Well, broadly speaking, there are 3 approaches open to you: design it, reverse-engineer it or evolve it. The third of these - artificial evolution - is attractive because it sidesteps the troublesome problem of having to understand how human intelligence works. It's a black box approach: create the initial conditions then let the blind watchmaker of artificial evolution do the heavy lifting. This approach has some traction. For instance David Chalmers, in his philosophical analysis of the technological singularity, writes "if we produce an AI by artificial evolution, it is likely that soon after we will be able to improve the evolutionary algorithm and extend the evolutionary process, leading to AI+". And since we can already produce simple AI by artificial evolution, then all that's needed is to 'improve the evolutionary algorithm'. Hmm. If only it were that straightforward.

About six months ago I asked myself (and anyone else who would listen): ok, but even if we had the right algorithm, what would be the energy cost of artificially evolving human-equivalent AI? My hunch was that the energy cost would be colossal; so great perhaps as to rule out the evolutionary approach altogether. That thinking, and some research, resulted in me submitting a paper to ALIFE 14. Here is the abstract:
This short discussion paper sets out to explore the question: what is the energy cost of evolving complex artificial life? The paper takes an unconventional approach by first estimating the energy cost of natural evolution and, in particular, the species Homo Sapiens Sapiens. The paper argues that such an estimate has value because it forces us to think about the energy costs of co-evolution, and hence the energy costs of evolving complexity. Furthermore, an analysis of the real energy costs of evolving virtual creatures in a virtual environment, leads the paper to suggest an artificial life equivalent of Kleiber's law - relating neural and synaptic complexity (instead of mass) to computational energy cost (instead of real energy consumption). An underlying motivation for this paper is to counter the view that artificial evolution will facilitate the technological singularity, by arguing that the energy costs are likely to be prohibitively high. The paper concludes by arguing that the huge energy cost is not the only problem. In addition we will require a new approach to artificial evolution in which we construct complex scaffolds of co-evolving artificial creatures and ecosystems.
The full proceedings of ALIFE 14 have now been published online, and my paper Estimating the Energy Cost of (Artificial) Evolution can be downloaded here.

And here's a very short (30 second) video introduction on YouTube:


My conclusion? Well I reckon that the computational energy cost of simulating and fitness testing something with an artificial neural and synaptic complexity equivalent to humans could be around 10^14 KJ, or 0.1 EJ. But evolution requires many generations and many individuals per generation, and - as I argue in the paper - many co-evolving artificial species. Also taking account of the fact that many evolutionary runs will fail (to produce smart AI), the whole process would almost certainly need to be re-run from scratch many times over. If multiplying those population sizes, generations, species and re-runs gives us (very optimistically) a factor of 1,000,000 - then the total energy cost would be 100,000 EJ. In 2010 total human energy use was about 539 EJ. So, artificially evolving human-equivalent AI would need the whole human energy generation output for about 200 years.


The full paper reference:

Winfield AFT, Estimating the Energy Cost of (Artificial) Evolution, pp 872-875 in Proceedings of the Fourteenth International Conference on the Synthesis and Simulation of Living Systems, Eds. H Sayama, J Rieffel, S Risi, R Doursat and H Lipson, MIT Press, 2014.

Related blog posts:

Friday, September 20, 2013

The Triangle of Life: Evolving Robots in Real-time and Real-space

At the excellent European Conference on Artificial Life (ECAL) a couple of weeks ago we presented a paper called The Triangle of Life: Evolving Robots in Real-time and Real-space (this links to the paper in the online proceedings).

As the presenting co-author I gave a one-slide one-minute pitch for the work, and here is that slide.



The paper proposes a new conceptual framework for evolving robots, that we call the Triangle of Life. Let me outline what this means. But first a quick intro to evolutionary robotics. In my very short introduction to Robotics I wrote:
One of the most fascinating developments in robotics research in the last 20 years is evolutionary robotics. Evolutionary robotics is a new way of designing robots. It uses an automated process based on Darwinian artificial selection to create new robot designs. Selective breeding, as practised in human agriculture to create new improved varieties of crops, or farm animals, is (at least for now) impossible for real robots. Instead, evolutionary robotics makes use of an abstract version of artificial selection in which most of the process occurs within a computer. This abstract process is called a genetic algorithm. In evolutionary robotics we represent the robot that we want to evolve, with an artificial genome. Rather like DNA, our artificial genome contains a sequence of symbols but, unlike DNA, each symbol represents (or ‘codes for’) some part of the robot. In evolutionary robotics we rarely evolve every single part of a robot.

A robot consists of a physical body with an embedded control system - normally a microprocessor running control software. Without that control software the robot just wouldn't do anything - it would be the robot equivalent of a physical body without a mind. In biological evolution bodies and minds co-evolved (although the dynamics of that co-evolutionary process are complex and interesting). But in 20 years or so of evolutionary robotics the vast majority of work has focussed only on evolving the robot's controller. In other words we take a pre-designed robot body, then use the genetic algorithm to discover a good controller for that particular body. There has been little work on body-brain co-evolution, and even less work on evolving real robot bodies. In fact, we can count the number of projects that have evolved new physical robot bodies on the fingers of one hand*. Here is one of those very rare projects: the remarkable Golem project of Hod Lipson and Jordan Pollack.

This is surprising. When we think of biological evolution and the origin of species, our first thoughts are of the evolution and diversity of body shapes and structures. In the same way, the thing about a robot that immediately captures our attention is its physical body. And bodies are not just vessels for minds. As Rolf Pfeifer and Josh Bongard explain in their terrific book How the Body Shapes the Way We Think, minds depend crucially on bodies. The old dogma of Artificial Intelligence, that we can simply design an artificial brain without any regard to its embodiment, is wrong. True artificial intelligence will only be achieved by co-evolving physical bodies with their artificial minds.

In this paper we are arguing for a radical new approach in which the whole process of co-evolving robot bodies and their controllers takes place in real space and real time. And, as the title makes clear, we are also advocating a open-ended cycle of artificial life, in which every part of the robots' artificial life cycle takes place in real space and real time, from artificial conception, through to artificial birth, artificial infancy and development, then artificial maturity and mating. Of course these words are metaphors: the artificial processes are at best a crude analogue. But let me stress that no-one has demonstrated this. The examples that we give in the paper, from the EU Symbrion project, are just fragments of the process - not joined up in reality. And the Symbrion example is very constrained because of the modular robotics approach which means that the building blocks of these 'multi-cellular' robot organisms - the 'cells' - are themselves quite chunky robots; we have only 3 cell types and only a handful of cells for evolution to work with. Evolving robots in real space and real time is ferociously hard but, as the paper concludes: Our proposed artificial life system could be used to investigate novel evolutionary processes, not so much to model biological evolution – life as it is, but instead to study life as it could be.

Full reference:

Eiben AE, Bredeche N, Hoogendoorn M, Stradner J, Timmis J, Tyrrell A, and Winfield A (2013), The Triangle of Life: Evolving Robots in Real-time and Real-space, pp 1056-1063 in Advances in Artificial Life, ECAL 2013, proc. Twelfth European Conference on the Synthesis and Simulation of Living Systems, eds. Liò P, Miglino O, Nicosia G, Nolfi S and Pavone M, MIT Press.


*was surprised to discover this when searching the literature for a new book chapter I'm co-authoring with Jon Timmis on Evolvable Robot Hardware.

Related blog posts:
New experiments in embodied evolutionary swarm robotics
New video of 20 evolving e-pucks

Friday, April 29, 2011

Ill robots might get a temperature too

Just spent 4 days at the beautiful Schloss Dagstuhl in SW Germany attending a seminar on Artificial Immune Systems. The Dagstuhl is a remarkable concept – a place dedicated to residential retreats on advanced topics in computer science. Everything you need is there to discuss, think and learn. And learn is what I just did – to the extent that by lunchtime today when the seminar closed I felt like the small boy who asks to be excused from class because “miss, my brain is full”.

Knowing more or less nothing about artificial immune systems it was, for me like sitting in class, except that my teachers are world experts in the subject. A real privilege. So, what are artificial immune systems? They are essentially computer systems inspired by and modelled on biological immune systems. AISs are, I learned, both engineering systems for detecting and perhaps repairing and recovering from faults in artificial systems (in effect system maintenance), and scientific systems for modelling and/or visualising natural immune systems.

I learned that real immune systems are not just one system but several complex and inter-related systems, the biology of which is not fully understood. Thus, interestingly, AISs are modelled on (and models of) our best understanding so far of real immune systems. This of course means that biologists almost certainly have something to gain from engaging with the AIS community. (There are interesting parallels here with my experience of biologists working with roboticsts in Swarm Intelligence.)

The first thing I learned was about the lines of defence to external attack on bodies. The first is physical: the skin. If something gets past this then bodies apply a brute force approach by, for instance raising the temperature. If that doesn’t work then more complex mechanisms in the innate immune system kick-in: white blood cells that attempt to ‘eat’ the invaders. But more sophisticated pathogens require a response from the last line of defence: the adaptive immune system. Here the immune system ‘learns’ how to neutralise a new pathogen with a process called clonal selection. I was astonished to learn that clonal selection actually ‘evolves’ a response. Amazing – embodied evolution going on super-fast inside your body within the adaptive immune system, taking just a couple of days to complete. Now as a roboticist I’m very interested in embodied evolution – and by coincidence I attended a workhop on that very subject just a month ago. But I’d always assumed that embodied evolution was biologically implausible – an engineering trick if you like.  But no – there it is going on inside adaptive immune systems. (As an aside, it appears that we don’t understand the processes that prompted the evolution of adaptive immune systems some 400 million years ago – in jawed vertebrates).

Of course while listening to this fascinating stuff I was all the while wondering what this might mean for robotics. For instance what hazards would require the equivalent of an innate immune response in robots, and which would need an adaptive response. And what exactly is the robot equivalent of an ‘infection’. Would a robot, for instance, get a temperature if it was fighting an infection. Quite possibly yes – the additional computation needed for the robot to figure out how to counter the hazard might indeed need more energy – so the robot would have to slow down its motors to direct its battery power instead to its computer. Sounds familiar doesn’t it: slowing down and getting a temperature!

Swarm robots with faults is something I’ve been worrying about for awhile and, based on the work I blogged about here, at the Dagstuhl I presented my hunch that – while swarm of 100 robots might work ok – swarms of 100,000 robots definitely wouldn’t without something very much like an immune system. That led to some very interesting discussions about the feasibility of co-evolving swarm function and swarm immunity. And, given that we think we’re beginning to understand how to embed and embody evolution across a swarm of robots, this is all beginning to look surprisingly feasible.

Monday, January 24, 2011

New experiments in embodied evolutionary swarm robotics

My PhD student Paul has started a new series of experiments in embodied evolution in the swarm robotics lab. Here's a picture showing his experiment with 3 Linux e-puck robots in a small circular arena together with an infra-red beacon (at about 2 o'clock).

IMG_8016

The task the robots are engaged in collective foraging for food. Actually there's nothing much to see here because the food items are virtual (i.e. invisible) blobs in the arena that the robots have to 'find', then 'pick up' and 'transport' to the nest (again virtually). The nest region is marked by the infra-red beacon - so the robots 'deposit' the food items in the pool of IR light in the arena just in front of the beacon. The reason we don't bother making physical food items and grippers, etc, is that this would entail engineering work that's really not important here. You see, here we are not so interested in collective foraging - it's just a test problem for investigating the thing we're really interested in, which is embodied evolution.

The point of the experiment is this: at the start the robots don't know how to forage for food; during the experiment they must collectively 'evolve' the ability to forage. Paul is here researching the process of collective evolution. Before explaining what's going on 'under the hood' of these robots, let me give some background. Evolutionary robotics has been around for nearly 20 years. The idea is that instead of hand-designing the robot's control system we use an artificial process inspired by Darwinian evolution, called a genetic algorithm. It's really a way of automating the design. Evolutionary algorithms have been shown to be a very efficient way of searching the so-called design space and, in theory, will come up with (literally evolve) better solutions than we can invent by hand. Much more recent is the study of evolutionary swarm robotics (which is why there's no Wikipedia entry yet), which tackles the harder problem of evolving the controllers for individual robots in a swarm such that, collectively, the swarm will self-organise to solve the overall task.

Still with me? Good. Now let me explain what's going on in the robots of Paul's experiment. Each robot has inside it a simulation of itself and it's environment (food, other robots and the nest). That simulation is not run once, but many times over within a genetic algorithm inside the robot. Thus each robot is running a simulation of the process of evolution, of itself, in itself. When that process completes (about once a minute), the best performing evolved controller is transferred into the real robot's controller. Since the embodied evolutionary process runs through several (simulated) generations of robot controller, then the final winner of each evolutionary competition is, in effect, a great great... grandchild of the robot controller at the start of each cycle. While the real robots are driving around in the arena foraging (virtual) food and returning it to the nest, simulated evolution is running - in parallel - as a background process. Every minute or so the real robot's controllers are updated with the latest generation of (hopefully improved) evolved controllers so what we observe is the robots gradually getting better and better at collective foraging. If you think this sounds complicated – it is. The software architecture that Paul has built to accomplish this is ferociously complex and all the more remarkable because it fits within a robot about the size of a salt shaker. But in essence it is like this: what’s going on inside the robots is a bit like you imagining lots of different ways of riding a bike over and over, inside your head, while actually riding a bike.

Putting a simulation inside a robot is something roboticists refer to as ‘robots with internal models’ and if we are to build real-world robots that are more autonomous, more adaptable – in short smarter, this kind of complexity is something we will have to master.

If you’ve made it this far, you might well ask the question, “what if the simulation inside the robot is an inaccurate representation of the real world – won’t that mean the evolved controller will be rubbish?” You would be right. One of the problems that has dogged evolutionary robotics is known as the 'reality gap'. It is the gap between the real world and the simulated world, which means that a controller evolved (and therefore optimised) in simulation typically doesn't work very well - or sometimes not at all - when transferred to the real robot and run in the real world. Paul is addressing this hard problem by also evolving the embedded simulators at the same time as evolving the robot's controllers; a process called co-evolution. This is where having a swarm of robots is a real advantage: just as we have a population of simulated controllers evolving inside each robot, we have a population of simulators - one per robot - evolving collectively across the swarm.



Related blog posts:
Environment-driven distributed evolutionary adaptation
Walterian creatures

Friday, October 15, 2010

New video of 20 evolving e-pucks

In June I blogged about Nicolas Bredeche and Jean-Marc Montanier working with us in the lab to transfer their environment-driven distributed evolutionary adaptation algorithms to real robots, using our Linux extended e-pucks. Nicolas and Jean-Marc made another visit in August to extend the experiments to a larger swarm size, of 20 robots; they made a YouTube movie and here it is:



In the narrative on YouTube Nicolas writes
This video shows a fully autonomous artificial evolution within a population of ~20 completely autonomous real (e-puck) robots. Each robot is driven by its "genome" and genomes are spread whenever robots are close enough (range: 25cm). The most "efficient" genomes end up being those that successfully drive robots to meet with each other while avoiding getting stuck in a corner.

There is no human-defined pressure on robot behavior. There is no human-defined objective to perform.

The environment alone puts pressure upon which genomes will survive (ie. the better the spread, the higher the survival rate). Then again, the ability for a genome to encode an efficient behavioral strategy first results from pure chance, then from environmental pressure.

In this video, you can observe how going towards the sun naturally emerges as a good strategy to meet/mate with other (it is used as a convenient "compass") and how changing the sun location affect robots behavior.

Note: the 'sun' is the static e-puck with a white band around it.

Sunday, April 15, 2007

Walterian Creatures

In Daniel Dennett's remarkable book "Darwin's Dangerous Idea" he describes the Tower of Generate-and-Test; a brilliant conceptual model for the evolution of intelligence that has become known as Dennett's Tower. I propose here another storey to the Tower, for what I want to call Walterian Creatures, after the pioneering neurophysiologist W. Grey Walter, inventor of the world's first electro-mechanical autonomous mobile robot.

In a nutshell Dennett's tower is set of conceptual creatures each one of which is successively more capable of reacting to (and hence surviving in) the world through having more sophisticated strategies for 'generating and testing' hypotheses about how to react. Read chapter 13 of Darwin's Dangerous Idea for the full account, but there are some reasonable précis to be found on the web; here's one fullsome description. But for now here's my very brief outline of the storeys of Dennett's tower, starting on the ground floor:
  • Darwinian creatures have only natural selection as the generate and test mechanism, so mutation and selection is the only way that Darwinian creatures can adapt - individuals cannot.
  • Skinnerian creatures can learn but only by literally generating and testing all different possible actions then reinforcing the successful behaviour (which is ok providing you don't get eaten while testing a bad course of action).
  • Popperian creatures have the additional ability to internalise the possible actions so that some (the bad ones) are discarded before they are tried out for real.
  • Gregorian creatures are tool makers including - importantly - mind tools like language, which means that individuals no longer have to generate and test all possible hypotheses since others have done so already and can pass on that knowledge.
  • Scientific creatures. Here Dennett proposes that a particular way of rigorously, collectively and publically testing hypotheses - namely the scientific method - is sufficiently powerful and distinct to merit a further floor of the tower. (I'm not sure that I agree, however that isn't important to the point I'm trying to make in this blog.)

Like the Tower of Hanoi each successive storey is smaller (a sub-set) of the storey below, thus all Skinnerian creatures are Darwinian, but only a sub-set of Darwinian creatures are Skinnerian and so on.

Gregorian creatures (after Richard Gregory) are tool makers, of both physical tools (like scissors) and mind-tools (like language and mathematics), and Dennett suggests that these tools are 'intelligence amplifiers'. Certainly they give Gregorian creatures a significant advantage over merely Popperian creatures, because they have the benefit of the shared experience of others, expressed either through using the tools they have made or refined or, more directly, through their knowledge or instructions as spoken or written. Arguably the most powerful intelligence amplifier so far created by one particular species of Gregorian-Scientific creature: man, is the computer, for with it we are able to simulate almost any reality we can imagine. Simulation is potent stuff, gedanken thought experiments are no longer doomed to remain flights of fancy and mathematical models need no longer remain dry abstractions. And one of the most remarkable kinds of computer simulation is of intelligence itself: Artificial Intelligence.

What if the tools made by Gregorian creatures take on a life of their own and become, in a sense, independent of the tool-makers? Embodied AI (= Artificial Life) has this potential. Walterian creatures are, I propose, smart tools that have learned to think, grown up and left the toolbox. Think of future intelligent robots (far more capable than the crude prototypes we can currently build) that might co-exist with humans in an extraordinary and fulfilling symbiosis.

The defining characteristic of Walterian creatures is that they are artificial. They've not only left the toolbox but crawled out of the gene pool. No longer bound by the common biochemistry of Earth's biota, yet sharing both the inheritance and evolutionary (albeit artificial) processes of their Darwinian ancestors. So what does this mean for Walterian creatures? Well, all of Walterian’s ancestors share the fact that however simple or sophisticated their strategies for hypothesising about possible actions those actions have to be undertaken by the self-same physical creatures that do the hypothesising. Ok, Gregorian- Scientific creatures can augment themselves with magnificent tools for compensating for their own sensory or physical limitations, like electron microscopes, submarines or manned spacecraft, or remotely operated robot space probes that act as sense extenders, but one thing Gregorian individuals cannot do is evolve themselves as part of the generate and test process. Consider this scenario. A future intelligent autonomous robot is exploring a planet about which very little is known. As part of its generate and test strategy this Walterian can in simulation fast-forward artificial genetic algorithms to evolve its own physical capabilities and then re-build parts of itself on-the-fly to best deal with the situation it has encountered. It could, for instance, artificially evolve and re-engineer itself the means to make best use of whatever energy sources are to hand. (It would be like you or I falling into a river and being able to artificially evolve and grow gills in less time than it takes to drown.)

Walterian creatures are, like Gregorians, able to share tools, knowledge and experience. They will be fully interconnected, so that any individual - subject only to the physical delays of the networking technology - can instantly seek information or resources from the shared Walterian artificial culture. However, unlike Gregorians, these individuals are capable of Lamarckian learning. Need a skill fast? If you’re a Walterian creature then, providing at least one other individual has already learned the skill and is either online or has previously uploaded that skill, then you simply download it. Walterian creatures would surely be profoundly different - and perhaps unimaginable by our merely Gregorian - kinds of minds.