#gopigo3 — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #gopigo3, aggregated by home.social.
-
Two ROS 2 Robots Are Alive Again!
TB5-WaLI has been "alive" continuously for the last 18 months, but Dave has suffered six annual re-incarnations as new versions of ROS 2 and Ubuntu are released.
This year Dave's re-incarnation from "Kilted-Dave" was complicated by an improved #GoPiGo3 API installation into a #Python #VirtualEnvironment
I am feeling so happy after two weeks of updating and debugging to have Lyrical-Dave autonomously "fully alive" again.
-
What Does A Robot See In The Mirror?
I couldn't resist comparing the #MoonDream 0.5b and 2b models looking at images of Dave or Wali.
The 2b model recognized Dave has a Minion character and is a robot.
The 2b model recognized that Wali is a robot but did not recognize the WALL-E character.
The 0.5b model simply hallucinates stuff, just like Ollama local models did for text queries.
#MoonDream #GoPiGo3 #TurtleBot4lite #Robot #Vision #LanguageModels #RaspberryPi5_8gb
-
Just had to try the #MoonDream Vision Language Assistant on my 4GB #RaspberryPi4 #GoPiGo3 #robot and my 8GB #RaspberryPi5 #TurtleBot4lite robot:
TL:DR; Pi5 with 2B model might be useful, but the 0.5B model will not be useful regardless of running on Pi4 or Pi5.
-
Dream has been to have my robot wander (using vSLAM) taking pictures of “unknown interesting objects”, (<-hard part), segmenting and learning the objects for #OpenWorldObjectDetection
(All local processing for privacy)
#MastersThesisIdea use #GoPiGo3 or #TurtleBot4 robot with #Oak-D #StereoDepthCamera, #ROS 2 distributed processing for #RTABmap #vSLAM and rejected RTABmap frames (unknowns) for #UnknownObject Detection and incremental model #MachineLearning
Reference Paper in comment ->