An avatar is a computer representation of a user, either in 2D or in 3D. It is also used as a technical term for computer graphics representations of various kinds of any sort of online virtual bodies.
See also: Flash CS4 inverse kinematics tutorial (a technical tutorial that deals with Flash avatars) and 3D interactive environment
The SO/IEC FCD 19774 — Humanoid animation (H-Anim) standard “specifies the structure and manipulation of H-Anim figures. H-Anim figures are articulated 3D representations that depict animated characters. While H-Anim figures are intended to represent human-like characters, they are a general concept that is not limited to the same number of limbs, heads, and other pieces body parts that are typical of human beings.”
The Animation Framework extension (AFX – pronounced "effects") provides an integrated toolbox for building attractive and powerful synthetic MPEG-4 environments. It is quite difficult to find an easy and complete overview. It includes a face and body (FBA) specification, Skeleton, Muscle & Skin (SMS) content, a Bone-Based Animation (BBA) paradigm. You can buy this ISO/IEC 14496-16:2006 standard for CHF 244 (about 160 Euros).
This is a fairly complex standard. E.g. the BAP body subset includes about 183 rotation points and 110 user defined parameters. The face has 84 standardized points and 66 low level animation parameters for the Face. I also find it frankly not acceptable that it seems to be really difficult to find free and useful overview information somewhere on this...
Virtual Human Markup Language (VHML) is an XML-based language “designed to accommodate the various aspects of Human-Computer Interaction with regards to Facial Animation, Body Animation, Dialogue Manager interaction, Text to Speech production, Emotional Representation plus Hyper and Multi Media information” (retrieved 15:59, 28 November 2008 (UTC)). It includes several subsystems.
The MetaFace Framework “aims to provide developers with an extensible anthropomorphic framework to achieve intuitive user interaction for websites. This will enable developers to construct websites based on anthropomorphic metaphors, such as, lecturers, shopkeepers, bank tellers, and tour guides. The MetaFace framework involves three main areas of research: Embodied Conversational Agents (ECAs), framework architectures, and computer interface metaphors” This project ended in 2004.
The i3p group is working on a SMIL-Agent standard for SMIL