Newsgroups: comp.ai.neural-nets
From: Leon@lfheller.demon.co.uk (Leon Heller)
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!pipex!demon!lfheller.demon.co.uk!Leon
Subject: Re: NN for trajectory recognition
References: <ee6384a1@p61.f1.n2000.z242.fidonet.org>
Organization: Home
Reply-To: Leon@lfheller.demon.co.uk
X-Newsreader: Demon Internet Simple News v1.27
Lines: 34
X-Posting-Host: lfheller.demon.co.uk
Date: Sat, 10 Dec 1994 13:40:05 +0000
Message-ID: <787066805snz@lfheller.demon.co.uk>
Sender: usenet@demon.co.uk

In article <ee6384a1@p61.f1.n2000.z242.fidonet.org>
           Daniel_Swertz@p61.wharfrat.fido.de "Daniel Swertz" writes:

> Hi NetLand!
> 
> For a medical application, I have to train a NN which can detect differences in
> movements, e.g. of a hand. The system samples the 3D-position of several points
> over time, then translates this into angels relativ to body, arm, hand etc. The
> modell used is very similar to the standard modell describing a robotarm. 
> This data (trajectorys) should be fed into a NN to detect different movements,
> e.g. grasping a ball and grasping a pen.
> 
> Now the questions:
> 
> What type of net do I have to use? Recurrent networks? Kohonen-nets?
> What papers/books do I have to read? 
> Are there similar NN, e.g. from Robotics?
> Will the world still exist tomorrow?

I supervised a student project a couple of years ago, which used a NN
to recognise gestures made by someone wearing a Dataglove. The NN used
was a standard back-propagation network.

A published paper worth looking at is:
Fels, SS and Hinton, GE (1993). Glove-talk: A neural network interface
between a data-glove and a speech synthesiser. IEEE Transactions on Neural
Networks 4, 2 - 8.

Leon
-- 
Leon Heller                       | "Do not adjust your mind, there is
G1HSM                             |  a fault in reality": on a wall
Email: leon@lfheller.demon.co.uk  |  many years ago in Oxford.
Phone: +44 (01734) 266679         |
