|FOUNDATION FOR INTELLIGENT PHYSICAL AGENTS||Torino|
|Source: L. Chiariglione||fipa_history|
A guided tour to FIPA
FIPA is arousing a keen interest in many environments. Purpose of this document is to provide a detailed account of the steps that led to its establishment and current activity for the benefit of those who were not part of the process.
The idea of FIPA
The idea of FIPA was born in the second half of December 1995. At the root of FIPA are a set of facts:
In the months January to March 1996 a set of documents (Rationale and Scope) were circulated to a selected number of individuals who contributed to the refinement of the basic ideas.
The first meeting
On 18 and 19 April 1996 a group of interested people met at the Imperial College, London. The meeting was by invitation but anybody who requested to join was accepted. The group reached agreement on several points:
The agreements were worded in a set of resolutions approved by the participants. A press release was approved. This received a wide distribution.
After the London meeting Chiariglione and Mamdani produced a definition of basic agent capability list that was widely circulated.
The second meeting
The second FIPA meeting was widely announced and due notice was given along with the FIPA Opening Forum program. A considerable number of participants gathered at the IBM T. J. Watson Research Center in Yorktown, NY on 24-26 June 1996. The first two meeting days were used to hear and discuss some 35 presentations addressing a wide range of agent-base applications and agent-enabled technologies. On the third day (26 June) three basic documents were produced:
along with a list of agent technologies candidate for 1997 specification.
The rest of the third day was used to deal with administrative issues such as
All the decisions were duly recorded in a set of resolutions.
The establishment of FIPA
On 5 September 1996 five individuals representing five companies met in the office of Me. Jean-Pierre Jacquemoud in Geneva and signed the papers establishing FIPA as a not-for-profit association under Swiss Civil Law. The five representatives elected Messrs. L.Chiariglione (CSELT), R. Nicol (BT Labs) and P. Schirling (IBM) as Directors of the Association.
The yearly membership fees were set at 3,500 Swiss Francs (about 2,500 US$) for Full Members but the fee was halved for Associated Members, i.e. members with no voting rights.
Information on the formal establishment of FIPA was largely distributed with an invitation to join. This was done both by mail, electronic mail and using the FIPA Web site established immediately after the Yorktown meeting.
The third meeting
The third meeting, hosted by NHK in Tokyo on 07-11 October 1996, was the first formal meeting of the Association, but was still open to anybody interested. It was a challengingly complex meeting to manage.
The Call for contributions, widely disseminated by mail and email, had requested contributions proposing applications and the corresponding technologies to enable drafting of a Call for Proposals of agent technologies to be used for the development of FIPA '97 specification.
The dedication and competence of the experts attending the meeting enabled FIPA to produce a high-quality Call for Proposals through the following sequence of steps:
The meeting appointed K. Enami (NHK) as 4th FIPA Director and decided that, starting from the 4th meeting, attendance would be limited to FIPA members. The exception for the following meeting was for those who would submit a contribution in response to CFP1.
A set of resolutions documenting the results were approved.
The fourth meeting
As for past information regarding the 1st FIPA Call for Proposals (CFP1) was widely disseminated by email. The html version of CFP1 was also posted on the FIPA home page and received over 1,500 hits in 3 months. The meeting, hosted by CSELT in Torino on 20-24 January 1997, was duly announced to FIPA members but information was also posted on the FIPA home page, because those who made a submission in response to CFP1 were invited to attend, irrespective of their membership.
The Responses to CFP1 provided most of the basic technology needed to develop FIPA '97 specification capable of supporting the four selected applications.
First the decision was made to subdivide the FIPA '97 specification in 7 parts. The first three parts would be of "normative" type (i.e. they would specify the agent technologies), while the remaining 4 parts would be of "informative" type (i.e. they would explain how to use the technologies specified in parts 1-3 to implement the 4 selected applications.
The table below gives the title of the 7 specifications.
|No.||Title||Type of spec.|
|2||Human/agent and agent/agent communication||Normative|
|4||Personal travel assistance||Informative|
|6||Audio-visual entertainment and broadcasting||Informative|
|7||Network management and provisioning||Informative|
The meeting established 3 Technical Committees (TCs) to develop the first three parts and appointed P. O'Brien (BT Labs), D. Steiner (Siemens) and Yeun-Bae Kim (NHK) as the corresponding Chairmen. The TCs produced a baseline draft specification for part 1, 2 and 3. These were posted on the FIPA home page and in 45 days have received more than 700 hits.
The meeting also succeeded in drafting a baseline draft specification of part 4.
Lastly the meeting produced a Call for Proposal (CFP2) for the purpose of obtaining the few items of agent technologies that had not been obtained from CFP1.
The fifth meeting