Quote:
Originally Posted by lynca
I like the idea of building off current technology. Adding a PC104 or mini-itx computer to interface with the current IFI controller would be a great way to keep the low-level drivers and expand on high level programming. I thought the Adam-Bots Co-Processor platform was a start in the right direction.
In terms of the actual hardware, Royal Assualt's (357) "Schubox" PCB of a board to inteface to IFI RC is a great idea. I think FIRST should sell a simplified PCB design (like the VEX controller). Letting teams do their own PCB design would bring down the cost of Electronics tremendously. |
Having teams do their own pcb design is not the way to go. Maybe 10% of FIRST teams could pull that off, and I'd say not all of them reliability. What do the rest of the teams do? As much as I love flexibility, they need a solution that doesn't prevent teams lacking resources to compete, while allowing other teams to expand as necessary.
I say give me some embedded 32-bit processor running an os (vxworks or embedded linux among others) with an FPGA on the IO. None of this x86 stuff - we simply don't need that power nor the complexity. The FPGA can do timing dependent stuff (count encoder pulses and time ultrasonic devices) and have the equivalent of the "master" uP in it to disable/enable the outputs. This leaves the processor open to the teams to do whatever they want with. This is similar to how the CMU Qwerk works, and even how a gumstix/robostix combo works (although yes, the robostix is an AVR and not digital logic). I think the Qwerk is actually a pretty good starting point, although I think it needs quite a bit of SW development before we could use it for FIRST.
As far as wireless between the OI and RC, I don't know that an off the shelf solution such as zigbee or 802.11 will work for this. With 802.11, things such as binary exponential backoff I just don't think will work for our situation - as it isn't impossible to end up with large latencies if the channel is congested (which aren't great if you robot is heading towards a wall, or worse, a person). Remember we need at least 6 robots with dedicated communication between each other. I haven't really put enough thought into it to really make some strong statements on this, however I think the communication between the robot and the OI can be focused on at a later date. I think enhanced processing power and an operating system is more important.
If there are really going to be a year and a half of development time on this, hopefully it will have some serious thought put into it and be a rock solid product.