PlayMate Requirements Meeting
Monday 16th Jan to Friday 20th Jan 2006
The robot lab is in the basement of our building. We'll meet in
the basement at 9am on Tuesday, then, at about 9.50 move up to
room 245 on the top level (level 2). (Ask at reception if you
come late.) Some of the time we'll split into two groups, one
using room 217, also on the top level.
Note that there may be a network access problem on the morning
of Tuesday 17th Jan as the UK academic network is havng
maintenane work done between 07:00 and 09:00 GMT.
Revisions
17 Jan 2006: Added Link to Somboon's
PDF slides on vision and action.
14 Jan 2006: Added Link to
Marek's
notes on the Katana arm.
Added information about maps and restaurants.
12 Jan 2006: Minor revisions
7 Jan 2006: Slight changes of format. Inserted lunch break on
Thursday
6 Jan 2006: Various changes made by AS following Jeremy's
suggestions.
Meeting Attendees
- BHAM: Aaron Sloman, Jeremy Wyatt, Jackie Chappell, Nick
Hawes, Somboon Hongeng, Marek Kopicki: 16th - 20th
Jan.
- FREI: Michael Brenner: 17th (11.55) - 20th
(11.40) Jan .
- TUD: Bernt Schiele: 17th (7.50) - 18th (19.20)
Jan.
- TUD: Gyuri Dorko, Mario Fritz: 17th (7.50) - 19th
(19.20) Jan .
- UOL: Danijel Skocaj: 16th (15.55) - 20th (14.00)
Jan .
- UOL: Ales Leonardis: 17th (20.25) - 20th (14.00)
Jan .
- DFKI: GJ Kruijff: 16th (11.55) - 20th (11.40)
Jan .
- DIT: John Kelleher: 18th (7.45) - 21st (10.55) Jan.
NOTE: Times are flight arrival and departure times. Expect to take up to
two hours from landing to get to the University. We assume people will
have to leave about three hours before departure time.
NB: Is anyone planning to come who is not listed here? If so please let
us know.
Objectives
The primary purpose of the meeting is to agree on:
- the main objectives
for
month 30 (Kitty)
for the PlayMate scenario, including ensuring that
they constitute a suitable milestone towards the objectives of CoSy,
- to derive from that specific short term objectives (next
three months), and a likely demonstration target for the 24 month
evaluation (i.e. demonstration to be reported by August, and shown soon
after)
- to determine who will do what
A secondary purpose is to help with clarification of Explorer
objectives and to identify opportunities for synergy between the two.
Many of the issues are discussed (briefly) in appropriate boxes
in the CoSy
requirements
matrix (still under development.
Some links into specific parts of the matrix are included below.
Messages sent to the cosy list, about targets, plans and tools are
here
Schedule
Monday 16th January
GJ and Danijel arrive. Possible informal discussions and evening
meal.
Tuesday 17th January - Room 245
0900: Tour of the robot lab, demo of anything that is
demonstrable, and free discussion.
(Bernt, Gyuri, Mario arrive)
1000: AS+BHAM people presentation and discussion of cosy
requirements
matrix, focusing mainly on requirements for 'kitty' the 30 month
target scenario and restrictions for month 24 demo (August/Sept
2006)
1045: Coffee
Brief overview presentations by CoSy partners on their views on the
scenarios and their aims for the meeting.
1100: UOL
1120: DFKI
1140: TUD
1200: Discussion then lunch
1300: Outline of scenarios and questions for the PlayMate
for month 30. (Led by JLW)
1400: (If Michael has arrived) Freiburg
1430: Action requirements
See the initial sketch in section 'D: Manipulation in Kitty' in
the general requirements for
kitty, and also
Marek's
notes on the Katana arm.
Discussion to include:
PlayMate action requirements
Explorer action requirements --
can we unify PlayMate and Explorer
action
requirements in any useful way. (e.g. planning and action
formalisms, and some aspects of planning and plan execution
mechanisms? Also some common spatial understanding???)
1600: Break
1630: Requirements for representation of the processes,
structures, objects, etc. in the environment:
Includes discussion of ontology required (what will need to be
represented).
Mainly PlayMate, but also Explorer as appropriate.
(This will feed into the discussion of vision the
next day, but is also a prerequisite for discussing
linguistic interactions.)
Note that there is already a great deal on requirements for
representation in
Deliverable 2.1
(also copied
here.)
1700: Linguistic interactions for PlayMate and
Explorer.
See
In the evening Ales arrives, and we can arrange a meal in the
city near the hotel. If Ales arrives in time we could start
discussing vision briefly, otherwise some requirements for
learning and possibly project management.
NOTE:
1930 Dinner a the
'Don Salvo' restaurant
at the Mailbox, B1 1RG,
(156 - 158 Wharfside Street,)
on Tuesday night.
Wednesday 18th January - Room 217
Wed a.m.: Discussion of vision requirements.
0900: Requirements for vision in PlayMate
1000: Discussion of possible Explorer/PlayMate
commonalities in representations for vision, architectural
requirements for vision, visual learning mechanisms
1030: break
1100:
Vision work needed to satisfy month 30 scenarios for the PlayMate
(Including some suggestions for
work to be done in Bham.
)
1200: Discussion and lunch
1300:
Discussion of collaboration between partners on vision needed for
the PlayMate.
1445 (until Bernt leaves at about 1620): Provisional
summing up of targets for month 30. Provisional division of
labour
After Bernt goes we can continue refining ideas so far, and
perhaps initiate discussion of architecture required to combine
components in Kitty.
See also
Deliverable 1.1
copied
here
We have made no dinner plans for Wednesday night. Arrangements will be made based on how the day goes and the wishes of individuals. Many restaurants (particular Indian ones) are available near the univeristy, or in the city near the hotel.
Thursday 19th January - Room 245
0900: Further detailed discussions
- Low level and intermediate level vision requirements
- Nature-nurture issues: what should be innate
- Planning requirements
- How linguistic competence will interface to the rest.
See
section F: in
kitty requirements
- How to represent spatial locations and regions, etc.
(e.g. in goals for the robot, plans for the robot)
- Apart from planning capabilities, what sorts of reasoning
capabilities will Kitty need, e.g. for predicting effects of
actions, etc.
- How will kitty understand causation? See
presentation on learning about causation
1230(??): lunch
1330: Begin collecting requirements for tools to
support architectural requirements, distribution requirements
and particular components:
- motivation, planning, decision making,
plan-execution,
- high level control (including attention)
- motor-control, proprioception (PlayMate and
Explorer?)
- low level vision, various levels of visual
perception,
- enduring model of the the environment
- various kinds of learning
- interfaces between language and other sub-systems
1530: Switch to discussing software engineering
methodology:
- how will new targets/subgoals be agreed
- formal testing and releasing of software
- evaluation of software by the group
- documentation requirements for software
- how to deal with unexpected delays, or unexpected new
opportunities
- how we should collaborate with other people (as we said
we would in the proposal)
- when and how we should decide on reports to write for
external publication
1930 Dinner at the
Cafe Rouge, Brindley
Place, B1 2HJ
Friday 20th January
Depends who is still here.
