Learn more
Support Salvius
Source Code

The Year of the Robot (2025)

9/15/2024
Its been a minute, hasn't it? Recently I noticed the sunlight streaming in through a cellar window, perfectly gleamed off the collections of spiderwebs that have accumulated around Salvius. It's been about eight years since I've really focused heavily on anything robotics related. School, work, and just life in general took me in more directions than I can reasonably enumerate in a casual blog post like this one. Within the last few months my main projects have been a photo booth (complete with its own app, and hosting its own wireless hotspot complete with local DNS), as well as a separate photo sharing app for events (check it out at nimbusqr.com if you're interested). Both of the aforementioned projects are going to have their debut at my wife and I's upcoming wedding.

As of late, I've been think that it's about time I dust off some of my old projects and start to revisit them in complement to some of the newer technologies that have emerged in the last few years. LLMs and the AI space in general are undergoing a technological renaissance the might utterly unblock some of the challenges I had left these projects to rest with. For bipedal locomotion for example, could we create a cluster of AI agents (I imagine maybe each one running on a dedicated Raspberry Pi) where each agent knows that it is a "knee joint", "ankle", "hip", etc. With instructions from a primary node it would be interesting to see how efficiently the cluster could coordinate and perform actions like taking a step.

At this point I'm aiming to kick off the next generation of my work in the open source and robotics space probably starting early in 2025.

No comments:

Post a Comment

Salvius the robot

Older Post

Dive into this project's history.
Humanoid robot Robotnaut

Extraordinary Humanoid Robots

Robonaut was the first humanoid robot in space with the goal to one day help spacewalkers make...
Chatterbot python machine learning

Chatterbot

Machine learning makes it possible for this python library to reply to users and learn based on past conversations.