Spring 2001

Lecturer: Rodney Brooks

Time: 9:30am - 12:30pm, Fridays

Room: 2-190

Traditional artificial intelligence has been centered around the idea of representation of the world, while the approach taken in this course is centered around action in the world. Natural intelligent systems are all biological and how these work can not always be understood in purely computational terms. A biological system must be understood in terms of its environment, it ecological niche, and its evolutionary history. At a more proximal level, this involves consideration of what sensors it can use, how its actuators operate, and how its perceptions and the dynamics of its movements are affected by its physical body.

Over the last decade a new school of Artificial Intelligence has developed, encompassing a number of threads including behavior-based intelligence, adaptive behavior, Artificial Life, situated intelligence, and robots for unstructured environments. Almost all of these rely on the physicality of the systems to some degree. This course aims to synthesize these new approaches to Artificial Intelligence around that common theme of embodiment.

This course examines what it takes to build intelligent systems that have physical embodiment. After a grounding in core artificial life concepts, we look at specific problems this presents, historical solutions, and contemporary research into the area of autonomous embodied systems. Specific topics include dynamical modeling of agent/environment interaction, neural modeling of perception and action systems, fundamental issues in vision and robotics, evolutionary modeling techniques, behavior-based approaches, and pre-cognitive and cognitive architectures. All these topics are examined from the point of view of having a physically embodied system in the world; this both presents unique problems and sources of simplification relative to unembodied intelligence.

Prerequisites: 6.001, 6.034, 18.03, 18.06.

This is a graduate class and will be treated as such. The problem sets will often be very open ended and require you to use your initiative to gather resources and organize yourself. There will not be any hand holding from the TAs on this. It will be like being a research student in a research group--you will need to figure things out yourself, or at least figure out how to find the necessary information on the web or in libraries.

There will be significant programming required in this class. If you are scared of writing a few hundred lines of C code from scratch, or being given a few hundred or thousand lines of C code and being asked to modify it then this class is not for you. Basic competence in Lisp is also assumed.

Course Staff:

  Name Office Email Phone Hours
Lecturer: Rodney Brooks NE43-940 brooks@ai.mit.edu 3-5223  
Course Secretary: Sally Persing NE43-942 sally@ai.mit.edu 2-2651  
Teaching Assistants: Bryan Adams NE43-938 bpadams@ai.mit.edu 3-7593 M:4-5pm; R:4-5pm
  Jessica Banks NE43-937 jessical@ai.mit.edu 3-7471 T:10-11am; W:10-11am

Class email list: 6836@ai.mit.edu

Staff email list: 6836-teach@ai.mit.edu

Announcements and News

Course Syllabus

Handouts

Final Project Presentation Schedule