Faster than before for the same price again. This time there is ram upgrade options.
4 Likes
2 Micro HDMI ports instead of 1 normal HDMI, 2 USB 3 ports, and USB3-C charger instead of micro USB. Also they flipped the locations of the USB and RJ45 ports.
1 Like
I was hoping they would add POE without the hat, and I am disappointed in USB C as an iPhone user. LPDDR4 is a big upgrade also. That said, it looks like a solid upgrade, and you can not argue with the price.
Why don’t you like the USB-C?
It’s not like they would put a lightning connector on a raspberry pi.
8 Likes
I love it, I just only have one cable. Not suggesting lightning haha. I have a ton of micro USB cables lying around.
1 Like
You could use the ones that come with the spark maxes if you have any of those
What is a hat on POE?
if only there was some use for this lightweight, durable, inexpensive, $35 piece of hardware that could be used to run Java/Python/C++ code, connect to a CAN bus, and connect to a local area network
i am glad we instead have the National Instruments roboRIO Advanced Robotics Controller
24 Likes
Much as I like Pis on robots for vision, they’re not exactly rugged. They’re picky about input voltage, the busses are fragile, you can ESD them as easy as look at them, and the SD card can be corrupted if you pull power. I much prefer having a robot controller that can take a beating.
3 Likes
Real-time processor, integrated safety schemes, protected I/O, quality/reliability test protocols applied to the integrated controller, dedicated customer support, global hardware team backed by a large corporation with resources to Make Their Deadlines, global regulatory/trade support…
We pay for a lot more than just the processor & connectivity. Five years on the same hardware, it “would be nice” to see a small price cut or chipset specification upgrade, but we’re really paying NI for much more than hardware and I would rather have predictable durability in the core competition components than have shiney new things.
Is Pi a great teaching tool & offseason platform? Definitely.
7 Likes
Sure, I agree that there’s a lot more to it than the hardware, but you make it sound like the Pi Foundation isn’t a multi-million dollar global operation. I don’t think we should act like it takes $1 billion to fund the FRC control system.
And the “real-time processor” is a dual-core disappointment where one core is consistently occupied by the netcomm daemon, but it’s not like anyone knows what that’s even doing.
My point is that everything that isn’t the hardware is not worth $400-$35=$365. Plus the cost of having LabView instead of Python as the beginner language choice, but that’s another discussion for another thread.
I wonder how this will compare to the Jetson Nano for vision. I’m sure the Jetson will be more powerful, but will it be worth the $64-$44 extra?
Market is ~4,000 teams.
About ~4,000 units per year, top line on that $365 is ~1.5m, about enough to pay for a cross-functional team of ~8 US-based employees (supply chain, logistics, product manager, hardware, firmware, customer service…), or more likely a smaller core team in the US with part-time support from other functional groups (some overseas), and it’s probably still overall sponsored by NI after accounting for all the other “little things” that scale nicely for a rPi and don’t for a 4,000 unit device.
We need to scale the program massively to see the level of cost amortization that the rPi team leverages.
3 Likes
This will likely vary for every team, so let’s take a look at the key differences.
Jetson Nano | Pi 4
CPU:
Quad core ARM Cortex-A57 | Quad core ARM Cortex-A72
Advantage Pi
GPU:
128-core NVIDIA Maxwell @ 921MHz with CUDA acceleration | 500 MHz VideoCore VI
Advantage Nano
RAM:
4 GB of LPDDR4 | Pi up to 4 GB
Where the Nano really cleans house is in AI tasks. If you wanted to do deep learning, go with the Nano. That said, the Limelight runs with Pi 3 hardware, so the Nano is generally overkill GPU wise for FRC. I would say more optimized code will be more beneficial than the difference in boards, but I also prefer the approach of throwing more power at things. Also, the Nano has a regular HDMI port, a big selling point because it is likely one less dongle to forget.
For info on the different models of Pis, I recommend this article.
1 Like
I don’t think that’s too big of a deal for the Pi when using it for FRC stuff. Once you have SSH set up, you won’t really need video output if it’s just going to sit on the robot and send over some values.
Definitely a valid point.
I could see Onshape teams filling their shops with these things. It looks like these will now support WebGL.
2 Likes