Assessment Schedule
Introduction
This schedule describes the three assignments for SEE326 Artificial Intelligence for Autonomous Systems, which constitute the full assessment for this unit. Each entry contains a description of the assignment task, , the proportion of the unit mark that the assignment mark contributes and a guide as to how you will be assessed for each task. If, after reading this document, you are unsure as to what is required of you for any particular assessment task, you should speak to your tutor, lecturer or the unit chair as soon as possible.
How you should use this document
This document provides an outline of the information relevant to your assessment in this unit. It does not contain all of the information necessary to complete all of the assignments. Additional information may be found on CloudDeakin under the relevant assignment entry. Make sure you read and understand all documentation before submitting an assignment solution for assessment, including the relevant sections of the Unit Guide. If you have any difficulties in understanding what is expected of you, make contact with your lecturer as soon as possible to discuss the matter.
Resources
The lecture materials will cover the general approaches to these problems and will investigate some of the solution methods. You must conduct your own research into solutions you feel are appropriate and reasonable given the assessment tasks. There are many sources of information within the library and online and you may seek guidance from your lecturer as to appropriate materials for particular problems. You are also encouraged to discuss these problems with your fellow students and to discuss possible solution methodologies. Remember the rules regarding appropriate referencing of information and plagiarism.
As to resources for writing reports and programming, you will find many appropriate books within the University library. Again, your own research is required to obtain this information.
Assessment
Your assignments contribute to your final mark for the unit in the following manner. Each assignment solution submitted for assessment is given a mark out of 100, which is then multiplied by the assignment weighting. These weighted marks are then summed to determine the overall numeric mark awarded for this unit. For example, if you scored 63/100 for assignment #1, 76/100 for assignment #2 and 35/100 for assignment #3, then your final unit mark would be 61 (63*0.2 + 76*0.5 + 35*0.3).
Note: You MUST pass assignment #2 to pass this unit. Failure in assignment #2 will result in a maximum unit mark of 44.
Assignment 1 - Write a Project Proposal
Introduction
This assignment requires you to produce a written project proposal pertaining to the problems presented in assignment 2. Your proposal should be viewed as a document that sets out and guides the development process of your solutions to these problems, which would be used by a project manager to guide a project team. In this case, you are both the manager and the team.
Learning Objectives
This assignment has 3 principal learning objectives. Students will:
• Learn a structured approach to project work through the authoring of relevant project documentation;
• Develop critical analysis and design skills relevant to real time, event driven autonomous systems; and
• Gain an understanding of the project life cycle
Report Structure
You must submit your proposal in electronic form (MS Word, PDF or Postscript; not hand written) in 12pt font using one of the following typefaces: Calibri, Arial, Palatino, Courier, Times New Roman, Helvetica or Computer Modern Roman (if you use TeX/LaTeX). Maximum report length is 10 pages (not including the title page). Deviations from these restrictions will result in a mark penalty. Your report must conform to the following structure
1. Completed Student Declaration
2. Title Page (Title, Student Name, Student ID, Date)
3. Introduction (no more than one page)
4. Problem Analysis
5. Different AI methods available
6. Proposed Solution
7. Software Design Proposal
8. Workflow Chart
9. Bibliography/References
Report Content
Your proposal must address the following key issues:
• What is the problem?
• What different AI methods are available?
• What is the proposed solution?
• What must be done to implement this solution?
• How will this solution be tested?
• What are the expected outcomes?
In addressing each of these questions, you should not simple state "The problem is to build an autonomous pilot", or "We will run the software many times to test it". You should consider the problem in terms of requirements and constraints, present solutions that meet the requirements and satisfy constraints, discuss how the solutions will be achieved and how you expect them to perform.
Your introduction must contain the following information: a brief outline of the problem being addressed in this work, an outline of the solution methodology, a summary of the expected benefits and a summary of the resources needed to create the solution (e.g., development time). This page is essentially a summary of what is to come and should convey succinctly what the rest of the document discusses.
The Problem Analysis section should contain your analysis of the problems that need to be solved in order to successfully achieve the performance standards (or goals) dictated in assignment #2. This analysis should contain as much detail as necessary to convey your understanding of these problems and permit a reader to understand all of the key features and constraints that the solution must address. For example, in analysing the problem of collision avoidance, you would need to consider factors such as the speed of the vehicle and hence time to collision and how this couples with the steering rate to determine avoidance capabilities. You may present the problems faced in each task in separate subsections, or you may choose to present them together.
The Proposed Solution section should present first an outline of your general approach to solving the problem of autonomy. Do you intend to create a behaviour- based system, a purely reactive agent, a deliberative agent or a hybrid. Then, for each problem presented in the previous section, describe your solution. You should include any descriptions of AI algorithms or techniques necessary to convey an understanding of how the problem is to be solved. For example, for the aforementioned collision avoidance problem, you might choose to use a steering behaviour. In which case, you should present an algorithm description of this method, perhaps with some mathematical support and discuss how this behaviour will be activated. Finally, for each solution you propose, you must discuss the expected limitations of the solution (for example, situations under which it might not work, or factors that determine its performance and the expectation of encountering these unfavourable situations).
Your Software Proposal should be a description of the structure of your agent implementation. For example, a description of major data structures and their relationships, together with an algorithmic description of the logic flow of your agent.
You might choose to submit this as a UML diagram, a flowchart or some other descriptive form that conveys how your solutions relate to a software implementation.
Your Workflow Chart should be a timetable for completing your project, including a list of design, implementation and evaluation tasks to be completed before submission of your assignment solution. Indicate the expected start and end dates for these tasks (or the actual date, if it has passed). Indicate on your chart the dates by which your development milestones will be reached (these milestones are up to you to choose, but might be things like ‘successful test of velocity control', or ‘successful demonstration of solution for performance criteria #3', etc.). You may choose to use a formal method such as a Gant chart, or submit a spreadsheet or table of your own design (provided that it clearly indicates the required information).
Your bibliography must conform to the Harvard style for referencing (author and year of publication in text, alphabetical author listing in bibliography). Ask your lecturer if you are unsure as to the information required in a bibliographic entry.
Assessment
Your proposal will be assessed both with regards to its contents and how this is presented. The online marking rubric should be used to indicate the relevant importance of these factors.
Assignment 2 - Develop an Autonomous Spaceship Pilot
Introduction
This assignment requires that you design and implement in software an autonomous pilot capable of controlling a simulated space vehicle in a constrained domain, subject to a required performance specification.
You will be provided with a skeleton client application within which you will write your own code that implements your autonomous pilot. You will be provided with the interface specification for communicating with the ship to be controlled, as well as a description of the simulation environment and the tasks your ship is required to fulfil for assessment.
The assignment is broken into 2 sections; the first part is due about week 9. This (Part
A) is so that you can get some feedback on how you are performing. Part A is worth 5%. Remember that if you can't submit part A in time not only will you lose 5% but it will be harder to finish as you will have less time and more to do. The assignment specification document breaks the project into tasks. Tasks 1,2 and 3 are included in part A and all tasks are in part B. (Part A tasks are included in part B). When assignment part A is due only those tasks listed as part A are required to be handed in. When assignment part B is due all tasks are to be handed in and all tasks will be evaluated.
Learning Objectives
This assignment has five principal learning objectives, such that students will:
1. attain a practical understanding of a variety of artificial intelligence techniques relevant to creating autonomous systems, with a focus on the benefits and limitations of these methods;
2. gain practical experience in the design of real time, event driven autonomous systems
3. further develop skills in software programming for engineering
4. further develop skills in engineering project management
5. develop a practical software application as part of their portfolio to show prospective employers
Submission Requirements
The provided client application is Unity with all scripts written in C# and you will be provided with all source, header and resource files needed to compile the provided skeleton application. To this you will add your own source code in a single file. You must implement this solution in C#, although C coding will work too. Unity comes with its own code writing package call MonoDevelop. It is however, recommend to use the MS Visual C# compilers.
You must submit your solution source code in electronic form as a text file. Submissions are to be made using the appropriate drop box in CloudDeakin. Contact your lecturer prior to the submission deadline if you are unsure of how to use this facility. All submissions must be received before the deadline, in accordance with the University submission policies (as given in the Unit Guide).
Your assessor will compile your software source code and assess it using the assessment guide given below. Failure to submit code that compiles will result in the following:
1. You will be notified by email that your code does not compile and you will be provided with a compiler error output.
2. If the deadline for submission has passed, you will have a maximum of 72 hours from the time the notification email is sent, or the submission deadline (whichever is the latter) to resubmit your application for assessment. After that time your application will not be assessed.
3. If the assignment deadline has passed then until you resubmit, for each 24 hours (or part thereof) after the latter of the notification or the assignment deadline, you will receive a penalty to your assessment mark equivalent to a late submission of 1 day (as described in the Unit Guide). That is, resubmission within 24 hours will incur a 5 mark penalty; resubmission between 24-48 hours, a 10 mark penalty; resubmission between 48-72 hours, a 15 mark penalty.
If you are having trouble debugging your software, it is in your best interests to contact your lecturer for assistance well before the submission deadline and to also seek the assistance of your fellow students. You have a whole semester to complete this task, so don't wait until the last minute!
Assessment
Your application will be assessed for its performance on the specified tasks within the simulation domain. Marks are awarded for completing the requirements of each task and the mark allotment is given in the task specification documentation. Details of this domain, your ship's capabilities, the control and interface and the task descriptions are provided in the software specification documents (published on CloudDeakin).
Assignment 3 - Write a Project Final Report
Introduction
This assignment pertains to the project work conducted in assignment 2 and forms the basis of the communication of the outcomes of that work. Furthermore, the final report provides a means of communicating a critical self-evaluation of the student's work.
Learning Objectives
In this assessment task, students will:
• Learn a structured approach to project work through the authoring of relevant project documentation;
• Learn the relationship between project work and project reporting;
• Gain practical skills for the communication of the results of research and development work;
• Learn about the benefits and necessities of critically evaluating the results of project work to improve future performance.
You must prepare a written final report of not more than 15 pages (not including title page or source code) on your project that provides an analysis of what was done, how it performed, what its shortcomings were and how you could improve upon the system. You must abide by the restrictions given for assignment #1 with regards to font typeface and size. Your report must include a completed student declaration and title page and must address the following:
• A description of your final agent implementation, including:
o a description of the AI methods used to solve the prescribed tasks
o where these methods differed from the project proposal, a justification for the change
o a description of the problems faced in applying the chosen methods
o a description of how you resolved these problems including, where necessary, reference to other people that assisted you, or other systems that you gained inspiration from
• An outline of the merits of your system for solving the given tasks
• A critical evaluation of the shortcomings of your system
o What went wrong
o Why did it go wrong
o Performance criteria not met
• A discussion of changes that could be made to improve your system
• With reference to the major tasks and milestones documented in your project proposal indicate which of these were met, which slipped (finished late) and which were missed entirely. Give a brief explanation as to why
• Your report must include your source code as an appendix
Assessment
As in assignment #1, your report will be graded on both its presentation (structure and style) and contents (the quality of information contained within). You will also be graded on your implementation (as detailed in your code). The online rubric will be used to apportion marks and this should give you an indication of the relative importance of these factors.
Attachment:- Assignment.rar