I would like to purchase the Magni autonomous robot. But I would like to use an NVIDIA board as the control board instead of the raspberry pi that the Magni robot comes with. Is the code base directly transferable to an NVIDIA platform?
Here is a two part answer.
- We have run the magni from an Nvidia platform as well as Intel board in the past.
- I feel I do not know your own definition of ‘directly transferable’. So I must say ‘no’, a port of some code would be required. If you have an experienced firmware person onboard then this is a doable task. So this part of my answer is being ‘conservative’ by saying ‘no’ so I do not promise you it is a ‘slam dunk’. If you have an experienced firmware person it becomes ‘more of a modest task’ and thus closer to a ‘slam dunk’. We do not have any application node to fully outline port to an Nvidia processor but I assure you it ‘can be done’. We have some customers using other CPUs is my point but it requires some thought.
I should also note that our native raspberry Pi code is all ROS on top of Ubuntu. Your success is also partly based on if you use Ubuntu Linux (thus less possible variables). I myself have even controlled the Magni MCB from as simple as an arduino. We define the MCB protocol and after that it is a matter of coding to that protocol which I should state has requirements like continuing to send speed commands our our dead-man timer kicks in and so on. If you require obtaining good wheel odometry it becomes even more involved but it can be done, just with an experienced firmware person being a suggested requirement.
Thank you for your response. This helps me a lot.