I’m always thinking “What Would Make A Better Robot?”.
Raspberry Pi has remained the best choice for most GoPiGo3 users. There are low cost versions, and there are big and fast versions for more demanding users.
For basic robot control, the Raspberry Pi family has added a series of microcontrollers (RP2040). The GoPiGo3 has a custom microcontroller with far more capabilities than the Raspberry Pi microcontroller products.
We are, though, entering a new computing age where “old” linear and even multi-processing programming techniques, on processor architectures with 4 to 20 cores are being augmented by massively parallel programming techniques on “Neural Processing Units” (NPU) and “Graphical Processing Units” (GPU).
The Raspberry Pi has attempted to meet these new needs with a couple “AI Hats”, that provide serious users with the NPU capabilities, while maintaining the flexibility and support of the Pi OS ecosystem.
Recently I saw a user declare he was done with Raspberry Pi, and had bought the latest Arduino Uno Q board which combines a “standard multi-core processor” with a small NPU, and a microcontroller on one board. At first glance I was intrigued by the completeness of the board for introductory robots, but today I read that he was hitting the main reason I will probably never leave the Raspberry Pi family - to update he had to flash a new OS, and lost everything he had on the existing OS card.
They will obviously improve, but being on the “bleeding edge” is not the dragons I want to be fighting.
When I originally bought my first GoPiGo3 robot, (Charlie), I also bought an original version Jetson Nano, imagining that I might want to implement a GoPiGo3-type robot with a more advanced processing platform. Unfortunately, that didn’t happen and, recently, the Nano SDK has been depreciated, making continued research there more difficult.
I still believe that the GoPiGo3 platform is still viable, both within the Raspberry Pi ecosystem and also within other ecosystems, (i.e. Jetson Xavier, etc., or other Pi-compatible pinout systems.)
In my own case, I am still learning about the particular details of the GoPiGo robot using the Dexter libraries within the Raspberry Pi platform. I am curious about other processor platforms, but I haven’t had the opportunity to thoroughly explore them yet, and I do not know if that will ever happen.
However this doesn’t invalidate the other platforms or the research this might inspire.
What would invalidate it is when the hardware platform capabilities become insufficient. The older member - Thomas - decided to abandon the GoPiGo platform because he was interested in a larger, more robust and capable robot platform that could run systems like Donkey Car, not because the GoPiGo platform had shot its bolt.
The GoPiGo platform is one among many. And I believe it still has a place, still provides a rich ecosystem for experimentation and exploration, limited only by the user’s imagination. Unlike other platforms, it still exists and still exists at a price-point that is still reasonable when measured against other robotics platforms.