Petoi's New Products and Vector as a Math Tutor
Petoi's New Products and Vector as a Math Tutor
Some cute videos
Meta likes Vector
First, an exciting update. Folks at Meta AI have commented on one of our demos showcasing how Vector can benefit from the Llama 3.1 series of Large Language Models (LLMs). Here is a snapshot of their quote. You can find their post on X here. If possible, please like it, or repost it. Thanks for the shout out, Meta!
Bittle gets an arm
Petoi is getting ready for its new products and accessories for the holiday season, and one of the exciting additions is a new robotic arm. This is a 2 degrees-of-freedom (DoF) arm which fits instead of Bittle’s head. From the pre-release video, it appears that one servo moves the arm up and down, while the other can move the gripper. Additionally, the arm can be moved sideways like Bittle’s head. All servos can be programmed from Bittle’s firmware. With the addition, Bittle equipped with the arm will become a 11 DoF robot. Here is Petoi’s video demonstrating the arm.
You can check out the new instincts file to support the new arm at Petoi’s Opencat repository here. You can also follow the discussion on Petoi’s forum’s here, including some recommendations of possible 3D printable accessories to help Bittle balance better.
Some of the activities the arm can be used for are to pick up and throw a small object, and to clap. It is definitely going to be interesting to examine how Bittle can balance itself while holding an object in its arm. You could also put Bittle to some use by making it carry small objects from one room to another at your home. The arm definitely brings a lot of motivation for young programmers to experiment with it and learn more. We are eager to put our hands on this accessory and try it out.
Vector as a Math Coach
We also worked on an example to showcase how Vector robot could be a great math coach, thanks to Wirepod’s new abilities. I like this video a lot, because I believe it shows how desktop robots can be a valuable addition in our lives, and teach us a lot. You can easily reproduce this example with the following steps.
1) Clone Wirepod from my repository. I am assuming that you are familiar with the process of installing a Wirepod Server and connecting your Vector robot to it. If not, please look up Wirepod’s installation notes here. Please note that I have made a few tweaks to the official Wirepod repository for the purposes of this demo.
2) Get a free DEV account with Sambanova here. Sambanova offers free access to the Llama 3.1 405B model required for this demonstration. Sambanova’s API also delivers the fastest token rate for the Llama 3.1 405B model according to this dashboard, which is very useful for Vector robot’s real time performance. At Sambanova’s cloud portal, you need to go to the APIs section to generate a new API Key (Click on Generate New API). The following screenshot will provide you an example.
3) Configure Wirepod Knowledge Graph Setup to use Sambanova’s API as per the following screenshot. Please fill the API Key as is provided to you by Sambanova once you sign up. Please also check how we changed the LLM prompt to help Vector become a great Math coach.
Here is a video of the demonstration of Vector as a math coach.
Conclusion
I hope you enjoyed the updates on Vector and Bittle. Please stay tuned for more exciting stuff as we extract the best value from these robots, and learn along the way.