COMP3004/COMP4105
Designing Intelligent Agents
Coursework Spring 2025
Overview
The coursework for this module is based around (1) designing intelligent autonomous agents and an environment with which they interact, (2) setting those agents a task, (3) asking one or more questions about that task, and (4) evaluating it using experimental methods. You will then present the results from this in a report, video or podcast, which will also explain the context for the work. Students doing COMP4105 will in addition do a short presentation explaining how their coursework relates to contemporary research and technology.
Details
An autonomous intelligent agent is a program that operates in a particular environment, perceives aspects of that environment, and then uses its intelligence to choose actions that change that environment, to carry out some task. Typically, these actions are a mixture of responses (either immediate or deliberated) to its perception and memory, and proactive actions such as exploration.
Your task for this coursework is to design an agent-based system containing the following four aspects:
An Environment. This is the (virtual) place where the agents will operate. It could be one of:
· A simulation of a physical environment in which mobile robotic agents move. This could be the simulation used in the classes earlier in the semester (perhaps extended), a robot environment such as The Player Project (http://playerstage.sourceforge.net), or a project in Unity or a similar game environment if you are familiar with one from elsewhere.
· A language environment, for example where humans are interacting using written or spoken language with one or more intelligent agents,
· The Bristol Stock Exchange system (https://github.com/davecliff/BristolStockExchange) or a similar simulation of some aspect of the economy or society
· A game environment such as Ms. PacMan (https://gym.openai.com/envs/MsPacman-v0/), the Open Racing Car Simulator (http://torcs.sourceforge.net), RoboCup (https://www.robocup.org/leagues/23) or similar (see e.g. http://www.gvgai.net)
· One of the more complex task environments from the OpenAI Gym (https://gym.openai.com)
There is no need to develop the environment yourself—the focus of the project will be on the agents in the environment (robots, trading agents, game-playing agents, autonomous drivers, language agents, etc.) – but it is likely that you will set up the details of the environment to address your specific question. You are allowed to use the code from the classes, but please try to make it clear broadly which parts of the code are taken from the class examples, and which is your own work (we appreciate that this is sometimes complicated to do at a line-by-line level, but you should indicate this in broad terms).
Autonomous Agents. You should introduce one or more autonomous intelligent agents into the environment, which use some kind of AI to solve a task.
· Examples of AI could be an AI planning system such as Goal Oriented Action Planning (http://alumni.media.mit.edu/~jorkin/goap.html), a search algorithm such as A* search, a genetic or swarm search, a reinforcement learning algorithm, fuzzy logic, or a hard-coded reactive or state-machine AI.
· The task will be one relevant to the environment: e.g. a robot vacuum cleaner clearing up dirt, a poet trying to write verse about the day’s news, a trader trying to optimise its returns, a game player trying to get a high score in a game, etc.
Within reason, you can use any language to do this. If you are planning to use anything other than Python, Java, C/C++, MATLAB/Octave, R, JavaScript, and mainstream web technologies such as HTML/CSS/JS, then please mention this in your topic approval.
A Question. You should ask a specific question (or a set of related questions) about your system. For example:
· How do different approaches (a genetic algorithm, an A* search algorithm, a hard-coded heuristic) compare in terms of task performance?
· How does the performance of the system change as we vary the number of agents in it?
· If the system is trained on one version of the environment, does that learning transfer over to a new version of the environment
· How do different kinds of communication/coordination between agents effect the efficiency of those agents on the task?
· How much improvement does storing some information (e.g. a map of the environment) make compared to carrying out the task in a purely reactive way?
· How do different kinds of sensing/perception systems affect the capacity of the agent to carry out its task?
· How sensitive is the agent to error/noise?
A Set of Experiments. You should answer your question by carrying out a set of experiments. Remember the structure that we talked about in one of the lectures:
· implement code that carries out a run of the agent’s behaviour and measures performance
· then, run that code multiple times to get a measure of average performance
· then, repeat that process for the different conditions in your question, and use descriptive statistics, charts/visualisation, and/or inferential statistics (e.g. significance tests) to test your question
Then, you should discuss the question using these experimental results as your evidence.
If this evaluation involves asking people to interact with your code, then please read the notes on research ethics that are in the Coursework section of the module Moodle page.
Constraints. Your project is expected to demonstrate around 90 hours of effort. So, you should not just run a basic reinforcement learning or convolutional neural networks library/tutorial on a simple environment from AI Gym or similar, though you can use these as part of your work.
You are not allowed to develop a basic chatbot i.e. a single system that interacts with a human using intent matching, identity management, transaction processing, question answering etc. If you want to do something on language agents, you should do something that goes beyond this (e.g. a creative language system, a system that involves interactions between multiple language agents, or a system that involves a language-based interface to another system).
You are at liberty to use any readily available libraries, frameworks, APIs, services, datasets etc. as you want—but your work must demonstrate substantial work in using, combining and building on what you use. You are not permitted to simply re-submit coursework from another module, but you could potentially build on code that you have written previously and build on it in a new way. If your work extends or re-uses coursework from other modules, it must make a substantial and distinctive contribution beyond what was in those courseworks, and you must clearly identify what the new contribution is.
Examples
Here are a few examples of things that you could do. You don’t have to do one of these—indeed, we would prefer you to come up with your own idea—but, these would all be acceptable project ideas if you want to do them:
· To take the “robot vacuum cleaner” from the early classes, and experiment with different numbers of robots, and different coordination strategies (e.g. robots try to stay a fixed distance from each other, compared to sharing a map that they build up)
· Explore how multiple language agents, each of which has a particular set of knowledge or perspective can, discuss/argue with each other to explore a topic.
· Take a number of different trading strategies and run them in the Bristol Stock Exchange system with varying amounts of noise/uncertainty, to see how robust each strategy is.
· Develop a game-playing agent or a control agent for an NPC or opponent in a game, and explore different strategies for playing (or learning to play) a game.
· Take the “avoid the cats” problem from the class, and compare a number of strategies for the problem: warning the cats vs. moving out of the way, and learning when to act based on a simple statistical approach vs. a decision-tree approach.
· Consider the problem of planning a robot’s movement around a mapped environment (e.g. the map generated from WiFi triangulation introduced in one of the classes). Contrast A* search and genetic algorithms on this problem, and compare them both against random wandering.
· Explore a swarm simulation such as the Boids flocking algorithm explored in one of the classes, and map out how the overall behaviour of the birds relates to the parameters in the algorithm.
Topic Approval
You should submit a short description of your project idea (a couple of paragraphs, 100-200 words) on the Moodle page by 15:00 on 26th March 2024. We will then give you feedback on whether the project is an acceptable one, and how it might be modified or improved if it is not acceptable. If you submit before this date, we will endeavour to give you early feedback.
Submission
By 3pm on 13th May 2025 you should submit the following via Moodle. This deadline may be extended if you have a support plan or extenuating circumstances. Late submissions will incur a penalty of 5% per working day, in line with the standard University late policy.
COMP3004
Your submission should contain:
A report, between 2500-4000 words, or a 20-minute video, or a 20-minute audio podcast, where you describe:
· The core ideas of your project; clearly state the question that you are trying to answer
· A review of relevant ideas, technologies and research papers
· How you designed the environment and agents in order to address that question
· Technologies used, and challenges that you met in doing the implementation
· How you set up and ran your experiments
· The results from your experiments
· A discussion of the question in light of the experimental results
· A conclusion, where you summarise the work, reflect on its successes and limitations, and briefly mention some ideas for how you would take the work forward if you had more time
The target audience of this is students in your year on your degree—so, there is no need to explain basic computer science ideas, but you should not assume a deep knowledge of your particular topic.
A copy of your code, either as an upload or a link to a repository
Anything else that you think would be helpful for the markers, e.g. sample outputs from your system, a link to a brief video demonstrating it working, reports from user studies/interviews etc.
For students on COMP3004, the mark for this portfolio of work will count as 100% of the module mark—there is no exam.
COMP4105
In addition to the portfolio of work described for COMP3004, students on COMP4105 should also do a 10-minute presentation about your work (dates/times will be arranged; this is likely to be between 14th-16th May 2024). This should give an overview of your project and explain how it is informed by research ideas from AI and intelligent agents.
For students on COMP4105, the portfolio of work (including the report/video/podcast) will count for 90% of the marks on the module and the presentation for 10%—there is no exam.
How the Work will be Marked
Marking will take into account:
· The intrinsic complexity of the overall project
· background research and how you have used it to contextualise your work
· the choice of task environment and how you have used it/adapted it for your specific project
· the effective use of artificial intelligence and agent-based systems ideas from the course and your wider studies in designing your autonomous agents
· how clear your question(s) are, how well the experiments have been designed to answer them, and your level of rigour in planning and analysing the experiments
· how well the report answers the question by using the evidence from the experiments
· the overall clarity and structure of the report, appropriate use of scientific and technical English, and the quality of charts, diagrams, pseudocode where relevant
· the quality of reflection on the successes and limitations of the work
· (for students doing a presentation) the structure of the presentation, the clarity of explanations, and good use of slides or other visual aids
Marking Scheme—Main Project
Each of the following descriptors gives a broad idea of the achievement expected for a mark in that range. Clearly, individual projects may fall short in some areas and show excellence in others. The marking should also be adjusted to reflect the intrinsic difficulty of the project.
|
Band
|
Guidelines
|
|
90-100
|
Marks in this range are reserved for a superb all-round performance. Work done in all aspects of the work go beyond even high expectations. The student has shown a thorough understanding of the problem. All tasks, including very challenging aspects and extensive stretch goals, have been successfully completed. The project shows depth and engagement with research ideas, and everything has been completed to a high standard. The report could form. the basis of a publishable conference/workshop paper.
|
|
80-89
|
Excellent contributions to all areas of the project. Exceeded expectations in some areas by carrying out implementations and experiments of a high level of complexity or sophistication. Demonstrates knowledge and understanding of the ideas that goes beyond the material covered in the module. Clear appreciation of the project as a whole, its adequacies, limitations and possibilities for future development. The project demonstrates insight and depth beyond that usually expected in undergraduate/master’s work. Report presented very clearly and to a good professional standard.
|
|
70-79
|
Very good contributions to all areas of the project. Successful completion of the project tasks including some more challenging aspects and stretch goals. Demonstrated initiative and creative problem-solving ability. Able to undertake the work in a competent and independent manner. Able to reflect accurately on adequacy and limitations of the project’s achievements. Report presented very clearly and to a good professional standard.
|
|
60-69
|
Good appreciation of background. A good attempt at applying this to the task, with demonstrated ability to cope with difficulties. Good technical skills in several areas. Whilst most of the core aims of the project have been achieved, it might come a little short in some areas. Good reflective understanding of the project. Well organised and structured report, with perhaps a few parts unclear.
|
|
50-59
|
Postgraduate pass level. Satisfactory background reading and a competent attempt at their tasks. Reasonable technical competence demonstrated. The core parts of the agents, environment and experiments have been completed satisfactorily, but little achieved beyond that. Able to reflect satisfactorily on the project. Or, a lower-difficulty project carried out well. Decently written and structured report, but with perhaps some aspects unclear, unbalanced or underdeveloped.
|
|
40-49
|
Undergraduate pass level. Competent background reading and appreciation of the project area. Basic technological competence. Some areas of the core tasks may be incomplete, but a decent attempt has been made at them. Able to reflect in a limited way on the project. Report gives an overview of the project and is readable, but perhaps lacks detail, clarity or focus in some areas and/or is too informal or unbalanced.
|
|
30-39
|
Unsatisfactory. Some attempt has been made at the background reading but clearly only partial understanding of project topic. Incomplete attempt at the core tasks. Weak technical competence. Little ability to reflect adequately on the project. Report fails to cover some aspects of the work.
|
|
20-29
|
Inadequate background reading, but shows some limited understanding of how ideas can be linked to the task. Minimal attempt at the core tasks, showing poor understanding. A substantial amount of work is still needed to achieve the core tasks. Minimal reflection on the project. Report is disorganised, overly-short and contains areas that are very unclear and/or demonstrates misunderstandings.
|
|
10-19
|
Minimal attempt at background reading, inappropriate use of material, almost no attempt at core tasks. Very poor understanding of the problem. Minimal or no reflection on the project. Minimal report, or report shows substantial misunderstandings.
|
|
0-9
|
No or almost no significant attempt.
|
Marking Scheme—Presentation (COMP4105 only)
Each of the following descriptors gives a broad idea of the achievement expected for a mark in that range. Clearly, individual projects may fall short in some areas and show excellence in others. The marking should also be adjusted to reflect the intrinsic difficulty of the project.
|
Band
|
Guidelines
|
|
9-10
|
A professional-level presentation of exceptional clarity and very clear structure, with very high-quality slides or other visual aids that flow seamlessly with the spoken presentation. Shows a thorough and up-to-date understanding of research and technologies in the area and how they relate to the work.
|
|
7-8
|
A clearly structured presentation, which explains all aspects of the project well, and which has high-quality slides or other visual aids that are tied in strongly with the spoken presentation. Well-grounded in an understanding of research and technology in the area, including contemporary developments.
|
|
5-6
|
Postgraduate pass level. A competent presentation that has a decent structure, gives a competent description of most aspects of the project, and where the slides and other visual aids are largely clear and related to the spoken presentation. Shows some awareness of how the project connects to research and technology in the area.
|
|
3-4
|
A presentation that has some level of organisation but where the topics are not presented in a clear order or where the presentation jumps from topic-to-topic, some explanations not clear, visual aids provided but not very clear and/or not very related to the spoken presentation. Shows lack of awareness of the wider research/technology context.
|
|
1-2
|
A presentation that mentions some aspects of the project work but is largely disorganised to the point where it cannot be followed and where most explanations are unclear, and where visual aids are unclear and/or not related to the spoken presentation. Minimal attempts to connect the work to research/technology.
|
|
0
|
No significant attempt at presentation
|