|
![]() | 2018/9 | |
![]() | Department of Computer Science | |
![]() | 6 [equivalent to 12 CATS credits] | |
![]() | 120 | |
![]() | Honours (FHEQ level 6) | |
![]() |
| |
![]() | CW 60%, EX 40% | |
![]() |
| |
![]() |
| |
![]() | Before taking this module you are advised to take CM20252 | |
![]() | Aims: 1. The students will develop practical expertise of artificial intelligence to control real-time autonomous systems, including autonomous robots, scientific simulations, and virtual-reality characters. 2. The students will develop skills in constructing the three types of intelligent system covered 3. To provide students with an introduction to intelligence in nature, and an ability to evluate the commonalities and differences between natural and artificial intelligent systems. 4. To develop reserach and reading skills appropriate for short conference papers, in order to take advantage of cutting-edge research. Learning Outcomes: 1. Students should be able to analyse and evaluate available options for mechanical real-world perception, and to consequently recommend appropriate technologies for informing robotic control. 2. Students should be able to analyse and evaluate a number of mechanisms for sequencing actions, and to implement appropriate mechanisms of action selection on a variety of platforms. 3. Students should be able to analyse and consequently form predictions of the consequences of simple actions being performed by a large number of agents. 4. Students should be able to test predictions of emergent group behaviour through social simulation. 5. Students should be able to describe the state of the art in acquiring and generating primitive actions for virtual reality, and to choose appropriate technologies for particular animation tasks. 6. Students should be able to demonstrate sufficient research and information retrieval skills to enable them to analyse, evaluate and compare both current and classic intelligent control algorithms from journal and conference literature. Skills: 1. Reading and assimilating technical papers. 2. Self-learning: study skills appropriate for technology professionals. 3. IT: programming skills useful for addressing contemporary commercial and scientific applications. Content: * Introduction: why intelligent control is (computationally) hard, outline / review of historic strategies (proof / search based, reactive / dynamic planning, machine learning, hybrids of these). Course structure, introduction to labs. Sensing: sonar, IR, laser range finding, vision, touch. strengths, weaknesses, and approaches to use each. * Action: mechanisms for sequencing, goal arbitration, problem spaces and contexts. Where do action primitives come from, how does morphology do work for you. Redundancy & degrees of freedom. * Perception and Learning: sensor fusion, memory, and learning. The beginnings of cognition. [lab 1 due] * Introduction to agent-based modelling; the impact of concurrency and society; simulations in policy and science; models, simplicity and explanation. * Natural intelligence: Evolution and cognitive control, variation in cognitive strategies found in nature, individual variation in nature; perception and action selection in nature. * Writing for science and engineering: special concerns for conferences, The use & nature of evidence. experiment, proof or argument? Picking conferences, knowing a literature. [lab 2 due] * Sensing & Action primitives II: Animation and Virtual Reality. Motion capture, segment smoothing. Motion planning and basic AI for games. * Complex planning systems, achieving multiple goals, agents with emotions and personality. Likeability, believeability and engagement. * Ethics and philosophy of AI, can we build consciousness? What should our users believe about our agents? | Before taking this module you must take CM20214 OR take CM20221 |
![]() |
CM30229 is Optional on the following programmes:Department of Computer Science
|
Notes:
|