[HN Gopher] MineDojo - Building Open-Ended Embodied Agents with ...
___________________________________________________________________
MineDojo - Building Open-Ended Embodied Agents with Internet-Scale
Knowledge
Author : theblazehen
Score : 65 points
Date : 2022-10-05 11:38 UTC (11 hours ago)
(HTM) web link (minedojo.org)
(TXT) w3m dump (minedojo.org)
| uoaei wrote:
| "Embodied" seems like a stretch. I would imagine "embodied" to
| mean something involving feedback control using multimodal
| sensory data. Otherwise it's no different from playing Atari Pong
| with a larger state and action space. Moving around in a quasi-3D
| environment, being essentially represented as a point with a
| sprite wrapped around it, is not what distinguishes a "bodied"
| agent from other kinds.
| TOMDM wrote:
| Agreed for the most part.
|
| If we were to be extra generous, then the audio and visuals in
| minecraft could be a toy case of training an agent with a
| multimodal sensorium (can the agent avoid enemies sneaking up
| on them using sound cues, find mobs based on sounds, know lava
| is nearby by the bubbling etc.).
|
| In this case though the agent is given just RGB pixels and some
| structured data about their inventory, a small number of nearby
| voxels, their health etc.
|
| You'd have to be very generous to call structured data like
| that and some RGB pixels a multi-modal embodied agent though.
| giancarlostoro wrote:
| I had to keep scrolling down to figure out what this was. People
| should really do tests to see if a non-technical person can tell
| them what their website represents in under 10 - 60 seconds, if
| they can't you need to rework your homepage so the immediate
| information fed to the end-user tells you what you are "selling"
| otherwise, you'll typically get people like me who close the tab
| and leave.
| learndeeply wrote:
| > you'll typically get people like me who close the tab and
| leave.
|
| You're not the target audience, why should they appeal to you?
| This is a website accompanying a paper about reinforcement
| learning.
___________________________________________________________________
(page generated 2022-10-05 23:02 UTC)