Adaptive Systems Spring Term. UG Essay
 
 

Please, refer to the course handout for details on assessment including handing-in dates.

General:

The essay (2,000 words) corresponds to 50 % of the final mark. You must write on some relevant topic of your choice drawn from those discussed during the seminars. You can extend the range of topics, but it will be evident to the examiners if you're not incorporating teaching discussed in lectures and seminars. You are encouraged to get feedback on the suitability of your essay topic and programming exercise from Xabier Barandiaran (xabier.academic@barandiaran.net) before week 7 -- ie by Fri 19th Feb.
 
 

What's expected:

Structure and narrative flow are key to a good essay. They cannot be achieved without careful thought and planning. There are no fixed rules on what constitutes a good essay, but a series of loosely coupled regurgitated opinions and quotes will not achieve a good mark. You should justify your arguments using the literature on the topic. Comparing and contrasting opinions and theories and coming to your own conclusions is a tried and trusted technique. The highest marks will be given to coherent well argued essays that have a clear and distinctive voice, this cannot be achieved by just summarising other peoples' views. The classic beginning middle and end structure is never a bad idea. The essay would start by introducing the topic and outlining the central aims of the piece. The main arguments are then carefully developed in the bulk of the essay before drawing the strands together in a punchy and insightful conclusion.
 
 

Suggested topics: (see also Questions page)

You should pick a topic from those discussed during the seminars.

What is an adaptive system? Discuss the different meanings of `adaptation' as seen during the course. How do they relate? Are they consistent uses of the term?

 
What is the relationship between the different adaptive processes observed in natural organisms (suborganismic, organism, ecological, social, evolutionary)? Are there any useful lessons to be derived for the design of autonomous adaptive agents?

Does the cybernetics of 50 years ago have anything to offer current research in AI?

How important has technology been in the development of theoretical ideas in cognitive science? Can technological development always be considered advantageous from the point of view of the history of ideas?

Discuss the advantages and limitations of Ashby's ideas on the origins of adaptive behaviour? What are the meaning of concepts such as essential variables and breaks and how do they relate to real organisms and possibly to robot design.

Is the framework of autopoiesis of any practical use for understanding living systems and adaptation? For designing adaptive systems? Try to find a concrete example where it would make a difference.

What are the possible meaning of `autonomy' when we talk about autonomous robots? Discuss implications of different meanings on the desirability and feasibility of designing truly autonomous robots.

Discuss the differences in perspectives between micro- and macro-evolution. Are there different meanings of adaptations that can be applied at different levels?

"Natural selection always leads to adaptive design". Discuss.

Discuss (some of) the major transitions in evolutions. What sort of problems do they pose, how can they be studied and what useful information can we obtain from them for designing artificial adaptive systems?

How can adaptation to visual distortions be explained? What does it tell us about the adaptive systems at work? Could an Ashbyan framework be used to understand this phenomenon?

What is a habit? How is it related to adaptive behaviour?

Discuss the viability of evolutionary robotics. Compare with other design methodologies. Suggest improvements, solutions to problems, etc.

Discuss the respective advantages and problems of studying/designing adaptation using simulation vs. using physical instantiations? What different roles can be played by simulations in evolutionary design of autonomous robots?

Is the dynamical systems perspective on autonomous agents complementary to traditional AI or in opposition to it?

Discuss the ideas of embodiment and situatedness together with the principle of designing whole agents. Where do these ideas come from? What are the modern manifestations? Where are they going or should the go? Support your arguments with critical uses of examples.