[robocup-humanoid] Significant concerns regarding the proposed humanoid league roadmap
Oskar von Stryk
stryk at sim.tu-darmstadt.de
Mon Feb 3 03:55:57 EST 2014
Hi Sean,
When you apply for a fundamental research project you will never get
it approved if you do not outline fundamental research questions.
Robot soccer does include many fundamental research questions
if the environment and conditions are set very close to human ones
(including kinematics and perception conditions like legs, arms, limited
FoV etc.).
If you just want to have a machine to play soccer you can use omniwheel
drives and omnivision.
However, this brings up a valid point which seems to be missing in the
current roadmap draft, it is the restriction of the environment
and setup of the robot soccer challenge not only to human-like playing
field but also to human-like motion and sensing restrictions for the
robots. These needed for a "fair" gameplay (i.e. human-like body plan,
kinematics and sensing restrictions (limited field of view etc.)),
so that the machine intelligence has to overcome the same limitations
as the human intelligence.
Therefore, I suggest to the TC to consider adding such statements in the
opening paragraph of the roadmap draft.
Best,
Oskar
On Thu, 30 Jan 2014, Sean Luke wrote:
> I would like to ask the technical committee to address some concerns I have about the proposed directions for the humanoid leagues as outlined in http://www.informatik.uni-bremen.de/humanoid/pub/Website/Downloads/HumanoidLeagueProposedRoadmap.pdf
>
> I have participated in RoboCup from 1997-1998 and from about 2009 to the present. I've never been interested in winning per se (which is good, since given our budget, we never will!) but rather using RoboCup as a challenge problem for cutting-edge research, and to give students an opportunity to participate in a world-class robotics venue. To make clear my bona fides regarding this position, I was the winner of the 1997 and co-winner of the 1998 RoboCup Scientific Challenge awards.
>
>> From this vantage point, I think the most important reasons for RoboCup are: advancement of robotics research, promotion of robotics teaching in universities, and promotion of robotics in the worldwide public eye. One reason I think is *not* an important goal is to produce a robot soccer team which will defeat the World Cup team by 2050. This has always been a fun slogan, but it’s not a goal which in and of itself helps the field in any meaningful way. I am concerned that the other more important benefits seem to be demoted in order to promote a goal of dubious value.
>
> In this vein, I have three concerns I’d ask technical committee to respond to:
>
> 1. It appears that this is something only being required the non-SPL humanoid leagues. I would like to know if other leagues are being asked to produce similar documents which outline how they will be continually physically modified until they produce robots capable of defeating the world cup team by 2050. If not, are these leagues expected to be phased out? Otherwise this document clearly produces a two-class situation, where I am subject to a steep rate of physical changes and cost simply due to my choice of research platform.
>
> 2. The proposed document is heavily aimed towards the mechanical and control aspects and capabilities of the robots, and does not touch significantly on the enormously problematic computational side of things. Given the makeup of the non-SPL humanoid leauges this is an understandable but large oversight. It is true that soccer presents a simpler computational environment than many other real-world scenarios, but even so, I see few milestones or technical challenges in the document which directly address the vision, multagent coordination, learning and adapting, planning, probabilistic reasoning, or speech recognition tasks necessary to create a viable team by 2050.
>
> 3. The proposed changes are extraordinarily expensive. Few institutions will be able to afford the kinds of robots and continual modifications that this document is demanding. The new requirements will be so expensive that my team, and I am sure many others, will not be able to follow them without appealing directly to funding sources. RoboCup already has a difficult time with funding, even funding on the side, due to perceived lack of seriousness (NSF Program Manager: "You want money to play SOCCER? Get real."). *Direct* funding at this rate is, I think, ultimately implausible. If we want to use humanoids as our platform, our only option will be to form consortia; and those who cannot will be left out. This works directly against the value of RoboCup in encouraging research, teaching, and promotion of robotics, and I cannot see how it can be beneficial to the humanoid research community. Is there some valid reason why these cost increases are being required special!
ly of this league, and why there are no alternative league options proposed for those people with existing kid-size platforms and limited funding?
>
> Sean Luke
> Department of Computer Science
> George Mason University
> sean at cs.gmu.edu
More information about the robocup-humanoid
mailing list