Robotics and real-time control

PRESENTATIONS : SOME NOTES

People who want real information rather than my usual rambling prose might like to select one of these items :

• The list of topics;
• The list of chosen topics;
• Who's giving which seminar when.

The presentations are meant to be a part of the course; I try to choose the list of topics to be interesting and to enlarge on the material covered in my lectures. Those giving the presentations might like to bear that in mind; those listening might like to take notes, or something.

The presentations used to be just seminars. The seminar proper is still the most obvious component so far as the world at large is concerned, but it isn't really well suited to assessment. You are therefore now also required to provide an essay on the topic of your seminar, which can be assessed and exhibited to our external assessor if required. I use the term presentation to denote both seminar and essay.

For the 773 presentation, you will probably begin with a published paper or other document, supplied by me, probably copied from a fairly recent journal. This is your source document. Unless otherwise arranged, your job is to find out about the subject of the source document and link it to the material covered in the course so that we can all understand it. If you know anything about the subject before you start, you will be unusually fortunate; it is more likely that you'll have to find out about it.

Your presentation should stress the computing component of the topic; that's what the course is supposed to be about, after all. This information isn't always given in the source document, so you might have to hunt for it. That's why I allot marks for "sources" in the presentation assessment form. It is unusual not to need to look for further information of some sort, and I expect you to do it. Even if you can't find out much about the specific software or hardware used in the source document, you can identify the problems and review appropriate methods of solution.

• The object of the essay is primarily to inform me about some topic on which you are supposed to have some special knowledge. It should be a clear account of the topic, explaining what it is and how it fits into the sort of material we discuss in the lecture course. I expect a fairly critical approach.

• The object of the seminar is primarily to inform the rest of us about some topic on which you are supposed to have some special knowledge. It is not to show us how much cleverer you are than we, nor to beat the world record for packing jargon into a telephone box, nor to mystify and bemuse us.

• Choose your topic.
• Acquire knowledge about the topic;
• Present it to an audience.

• So that it can be taken into account in the assessment ( a minor reason );
• So that we know to regard your presentation as authoritative ( a major reason ).

On choosing a topic.

The first step is to inspect the list of seminar topics. If one of these appeals, get the paper ( a copy thereof should be in a folder in the 773 laboratory ), and read it. If you'd like to use it, tell me about it by electronic mail . DO NOT remove the paper from the folder until I say you can - someone else might be interested in that topic, in which case an arbitration procedure will be needed. If you'd rather choose a topic of your own, talk to me about it first.

The second step is to decide what to do with the topic. The aim is to compose a presentation which will be informative and comprehensible; it should be based on, but shouldn't be restricted to, the source document ( if there is one ). For example, you might choose to present the material of the source document and to compare it with similar work from other sources, giving a critical evaluation; or you might choose to describe the background to the material of the source paper, and to assess it on that basis; or ....

The choice is up to you, but typically it depends on the accessibility of the material to someone meeting it for the first time in, say, a seminar. If it's a well-known topic, or something easily understandable ( robots, for example ), you might not need much background; for a more abstruse topic ( theory, electronics ) more might be appropriate.

You might get some idea of the appropriate level by considering your own reaction to the topic. What did you think when you read the article ? If you couldn't understand any of it at all, choose a different topic. If you could see the point of the article, but none of the details, concentrate on background - try to explain enough to show why the article was worth writing. ( In a well written paper, that should be covered in the introduction. ) If it was all fairly clear, concentrate on the material of the article itself - describe it, and show how it relates to other work.

On acquiring knowledge.

Knowledge can be acquired from people, organisations, computer systems, and books. It can also be acquired by introspection ( not very reliable, and should be limited to a few "it seems to me that ..." comments ), analysis of available material ( good when you can do it, but must be done well ), and original research ( welcome, but not expected, as it takes time. )

You can get in touch with people and organisations by visiting, telephoning, writing, electronic mail, world-wide web, etc.. Please check with me before pestering people outside the department, especially if they are outside the University; we might not have much of a reputation, but let's make sure we preserve what we've got. In any case, I need to know what's happening in case people ring me up about you. Do not trespass on people's goodwill; they owe you nothing, and you can't expect them to spare you a lot of time and/or give you lots of material. The department can manage reasonable postage or copying expenses. WWW is good because it bypasses most of these awkward considerations, but see below. Include a copy of any significant items with the material you hand in.

The major source of information from computer systems is now the world-wide web. You probably know at least as much about that as I do. For example, you probably know that the great majority of the information available is unreliable; checked sources are always better than unconstrained www pages. Technical reports are usually reasonably respectable. If you do refer to material from the www in your reports, identify its source, and give the URL in your list of references.

You will find several books in the library. You will also find journals. Computer and Control Abstracts is fairly hard work, but gives you a route into the recent literature. Don't forget the Engineering Library; they have a lot of material on control and related topics. There is the interloan service for books which we don't have, and it will also get photocopies of single journal articles. It's good in principle, but it can sometimes be very slow, and now we're stuck in semesters instead of years it's likely to be too slow. It might also be worth looking up plausible topics in the library's subject index. If you find one book which seems relevant, look also at the books near to it on the shelves; the subject index is soso, but the books are fairly well classified into topics.

Any or all of these methods might be appropriate for your 773 presentation. Unless otherwise agreed, your job is to explain the material of the source document in comprehensible terms, enlarging on it as necessary by reference to other sources. Please bear in mind that you are expected to find material for yourself. It is not sufficient simply to rephrase the contents of the source document. The document will usually cite some references, which might or might not be worth following. It should certainly give some hints as to possible keywords or phrases with which you can start work on the Abstracts journal.

On presenting your knowledge.

First decide what you want to say; then say it. ( "Say" includes "write", whenever appropriate. )

The constraints on what you say are how much you know, how much time you have, how much your audience or readership can absorb, how you organise the material, and what I think you ought to say. ( "What I think you ought to say" is not a dictatorial infringement on your democratic rights, but an attempt to keep the material relevant to robotics and real-time control. There's a SORT OF CHECKLIST containing a list of headings under which you might be able to make useful comments on topics. It's primarily for the assignments, but use it as a guide in the presentations too. ) In both essay and seminar, I shall expect something like :

• some explanation of how the topic fits into more basic things ( where "more basic" means, roughly, covered in the lecture course or textbook );
• evidence that you've referred to other sources - references to books, journals, etc.;
• an emphasis on the computing aspects of the topic, rather than on the engineering details.

The essay need not - and probably should not - be identical with the notes for the seminar, though there is obviously likely to be a close connection. ( We've tried handing in real seminar notes, and found that they aren't reliable indicators of your expertise. ) It should be laid out reasonably carefully, and must tell me about all the important points you want to make, how they fit together, the references you've used, and anything else that I should know. Remember too that your essay must be self-contained, as you won't be there to answer questions when it's assessed.

While it's impossible to set precise guidelines, it seems to need at least eight to ten pages of text ( not including diagrams ) just to get in all the requirements. Gaps in arguments ( bad reasoning, cases you've skipped ), biased discussions ( advantages without disadvantages, personal preferences ), and other sorts of uneven treatment are very obvious.

Because of that, it's probably a better strategy, other things being equal, to concentrate on clarifying and explaining one issue thoroughly than to aim at a full account of the source paper you begin with. Of course, other things aren't equal, and different topics invite different sorts of treatment, but the accent on thoroughness is likely to remain.

Mere description isn't enough. I want some evidence of significant intellectual activity on your part. Probe, compare, dissect, discuss, analyse, criticise. Etc. That's why it's very good to find related but different work - and also why it's important not to get too abstruse. If you can find examples of two ( three, N ) ways of doing things, compare them, say which is better ( best ) under what circumstances, and otherwise exercise your mind on them. If you can find only one way, perhaps you can think up another and show why it's worse. ( Or, more exciting, show why it's better. ) Look for flaws, special cases, assumptions which might not always be satisfied, and other weaknesses. Of course, all this has to be done honestly, with proper arguments and good evidence.

You might find it appropriate to dissect, discuss, analyse, and criticise the presentation of the source paper. That doesn't fit into the seminar very well, but a section in the essay might be appropriate. Some papers are written exceedingly badly, and it's fair to comment on that if it makes your task harder. For that matter, some papers are wrong, and if you can demonstrate that it is certainly appropriate to comment. Your essay shouldn't review the source paper, so don't overdo it, but if you think it's worth mentioning that's not unreasonable.

In the seminar, plan to speak for no more than 20 minutes; that leaves time for questions, which are an important part of the exercise ( and also for any extraneous matters to do with the course, which have nothing whatever to do with the exercise ). The members of your audience are all highly intelligent ( just like you ), and can probably absorb quite a lot, provided that it's linked to things they know already - but you might be surprised at how little you can actually say in 20 minutes. It is a very short time. If the subject is at all unfamiliar, it might be best to give no more than an introduction; if you really want to present a more detailed account, leave it to the essay, but I don't expect ( and don't really want ) the essay to go beyond the seminar in content.

Don't just copy the source paper. Don't even paraphrase it. The paper was written for experts in the field, which the 773 group is usually not. ( And neither am I. ) Perhaps you can pick up something which shows why the work is important ( use the references ), and how it fits into other work, or some principle which is needed to understand the paper; then present that in the 773 context. I do expect you to read more widely than the source paper.

Most people find it easier to absorb new material if they have an example to think about. It's often possible to present the bones of something new as a comprehensible case study, then bring in the other material you want while discussing the example. It works because people can fill in a lot of gaps in a practical example, so you don't have to cover every detail; it's much harder to do that just given an abstract discussion. You can sometimes invent a satisfactorily plausible example even if the source document doesn't give one.

There are several ways of conveying information : for the essay, text, pictures, graphs, tables; for the seminar, talking, writing on a board, overhead projection, multimedia presentations, handouts, working models, and any others which might take your fancy ( so long as they're cheap ). Each has its own purpose; use the one that does the job you want - bearing in mind that you probably don't want to spend several hundred hours preparing it. Avoid spectacle for the sake of spectacle; overhead projection from transparency or Datashow can be effective, but if it only presents your lecture notes it's a waste of time ( not to mention transparencies ) and a distraction. It is absolutely not a good idea to rely on presenting a few headings by overhead projector and hoping you can find something to say about them at the time. ( It has been tried. Don't. ) Perhaps the essay would be a good source of handouts for the seminar. Notice that these seminars aren't quite the same thing as the seminars about projects and theses - those are isolated events, and it doesn't much matter if everyone forgets them five minutes after they've finished, but the 773 variety are part of the course, and those in the audience are supposed to remember them, so handouts might be particularly appropriate. Bear in mind too that ( unless your topic is so boring as to discourage us completely, or so complete as to cover the subject in its entirety ) we might want to find out more about it; tell us how - cite sources, with adequate references. I hope that any relevant topic which you don't mention in the seminar will be picked up in the questions, so be ready for anything.

Ways and means.

Here are some collected notes on how to go about using various presentation media. I've just included things that have happened to come my way; if you want to try anything else, ask.

Transparencies for overhead projection : You can get a few ( how many depends on the current budget, but for a 20-minute seminar you shouldn't need more than five or six at the most ) from the office; identify yourself as a 773 student. You can either write on them, or copy onto them using an ordinary photocopier. Ask the office people ( or me ) for help if you need it - do not mess up our photocopiers !

Seminar room video display : ( from Peter Shields ) If the presentation will display on a VGA monitor then it will work in our seminar room. The data projector is usable with all PCs and Macs that have standard VGA or Mac video sockets on them. This covers most desktops and most of the modern laptops.

Powerpoint or other PC presentations : ( from Peter Shields ) If the person wants us to provide the PC and the powerpoint software we would have to use a staff machine as the only computer in the seminar room is a PowerMac 7200. I want to put a PC in there when we have some money. ( from me ) - so if you haven't got a conveniently portable PC, give us some warning.

On preparing your knowledge for presentation.

If you're accustomed to giving information on technical subjects, you will have worked out a way to do it which suits you. For those who haven't, here are some hints.

• First, collect information. Begin with the source paper. Read it, and try to make sense of it. Even if you're only proposing to give an introduction to the topic, it helps to have an idea of the specific application described in the paper. Note down points which you find puzzling; these are things you should try to look up somewhere else.

• Read books, papers, reviews, anything you can find which is reasonably relevant to the topic. Bear in mind that you will typically need only a rather vague idea of the application area, but you should find out more about the robotics, control, and computing parts of the work. If you can't find anything, you could always try asking me, but do have a look round first. The order isn't very important yet; but you should try to acquire a feel for the topic, and some idea of how it's structured. As you go, make notes on what you find. You might find it helpful to keep the notes on separate pieces of paper, labelling them with the topics which they represent; later you can sort them into order, when you find what the order is.

• Once you have a clear idea of the area, decide what topics you think are relevant, and concentrate on those. Sort out the information you've already collected, collate it all, find out areas on which you need more, and weed out areas on which you have too much. ( Don't throw it away; this superfluous material is often useful for answering questions or giving examples. )

• Then write it down. It is probably sensible to aim to write down the essay as a first draft first, by about the mid-semester break, then to use the essay as the basis for the seminar. It is unlikely to be satisfactory to deliver the essay as the seminar; the two modes of presentation are different, and need rather different approaches.

• For the seminar, you should have a written version at least a fortnight before your seminar date. As you don't know when that will be, the only safe policy is to get it ready in advance of the beginning of the seminar series. If this is your first try, write down the whole of your address ( including funny jokes and spontaneous asides ) in full, and practise reading it out. It is a good idea to do that in private. You will probably have far too much material, or far too little. Adjust accordingly. Do not allow time for applause. Go through the motions of showing transparencies, giving demonstrations, or other non-speaking activities, and allow a little time for people to absorb them if that's needed. Also decide where things will be during the talk. Where will you put the unused and used transparencies ? Where will you put any special equipment ?

• Then review the seminar. Do any points need more or less emphasis ? Is the order of presentation right ? ( For example, do you use results before you introduce them ? ) The topic should develop in a reasonable way. Make any necessary amendments, and try again. Aim to get to know the material sufficiently well that you can give your seminar without over-frequent reference to your notes - but don't learn it off by heart ( unless you're an experienced actor ), or it will sound stilted. Don't expect people to remember things for a long time - it's fair enough to begin by sketching what you're going to say in very broad outline, but don't give a long list because no one will remember it.

• Once you are pretty certain what the seminar will be like, prepare any transparencies or other material. If you do this too soon, you can waste a lot. If you have equipment which must be plugged into anything, make sure you have somewhere to plug it, and the cables you need. DO NOT LEAVE ANY OF THIS UNTIL THE MORNING OF THE SEMINAR : it's often painfully obvious if you do, and it's very unimpressive.

• Finally, go back to the essay. You will probably find that the exercise of preparing the seminar has helped you to understand things more clearly; it might be appropriate to amend the essay to take account of your improved knowledge. Then prepare an improved draft, and give it to someone else to read and criticise. Repeat that step several times if necessary, and then prepare the final version.

• ( Experience suggests that I should add this bit, though I don't think I should need to. ) When writing your final version, take some care with the presentation. If you can't see the point of that as a matter of principle, then you might reflect that I have to read it and understand it, and so does the external examiner. This is not a request for seventythree different flashy typefaces and multicolour printing; all we need is a reasonably clear layout, spaces between paragraphs, page numbers, clear and uncluttered diagrams, and things like that. Good spelling, grammar, and punctuation are also welcome, though I don't give or withhold marks for them or their absence.

Always be critical about your topic. Assess, compare - particularly with other work. There is usually something, even if it isn't very closely related.

Give examples if at all possible : they're much more illuminating than unadorned description, however good. If the source document doesn't give any, look for some others or make one up by working through some problem similar to the subject of the paper, making intelligent guesses as appropriate. ( Then say you've made it up, of course. )

The essay should NOT include the jokes.

Handing in.

I don't particularly want to collect the essays until all the seminars are finished, though if you want to get rid of them I'll take them at any time. You can give them to me or put them in my post slot by the office.

Please also return the source paper, if there is one. I need it to see how your essay compares.

Marking.

After each seminar, I shall ask each course member present to assess the choice of material, organisation of material, and presentation of material. I shall also assess these qualities in the talks. I shall make known the average marks allotted by the course members, so that it's possible to maintain the same standards throughout.

The final mark for the presentation will be composed from my assessment of the seminar and essay and the course members' assessment of the seminar, in proportions roughly 1:2:1. The mark for a satisfactory treatment with no specially good features will be around 60% to 70%. More than that means that you did something good; less means that you didn't.

I don't mark the essays until they're all finished, to make sure that I can assess them all together and at least try to use similar standards throughout. When they're done, I return to you my marking sheet and comments. You can inspect a specimen marking sheet if you wish. It is the version I used when I assessed the seminars at the end of the 1995 course, and there are some comments on how the marks are calculated. It might change slightly, but the general principles are likely to stay pretty well constant.

Alan Creak,
February, 1998.

Go to the 773 main course page;;
Go to me;
Go to Computer Science.