by piotr5 » Tue Nov 24, 2015 12:55 pm
mainly it's because I found and because andreas pointed out . (obviously in china such people like the one in the audio-thread here in fpga discussion are quite common.) the kind of robot in the first link is small, it fits into a hand. maybe with snickerdoodle you cannot build such a tiny robot, but it isn't much bigger either. also with drones I see the trend towards miniaturization, as long as you can put a camera on it. with a controlling unit for the robot there is a size-problem though. if you're too small you wont fit the batteries for smarter processors, the motors wont manage to lift them. in other words, I see economically an opportunity opening up for small controlling-units which consume little power. i2c/spi are great for sensors and such, cpu is great at triggering actions from sensory input. but fpga has to offer an extremely high bandwidth for communication and the flexibility to plug in stuff which doesn't have i2c/spi. further I noticed gentoo has added a lot of rtos stuff for robotics. so obviously there is some demand for robotics-hardware. with fpga it's possible to have a single chip for many tasks, instead of using one shield per task. all this makes me believe an addon-board implementing lots of typical robotics-duties with the help of fpga in the cpu-chip would fit the current trends. if with minimal effort and weight such board could take care of quad-copters or lineform-robot-modules, that'd be a success story. on the other hand, the market for computation-units is quite saturated, and chip-developers are bad customers for fpga since they only need a single board for their prototypes. I see several trends converging, the desire to build something which could become an economical success, the trend towards modular robotics, the trend for data-acquisition and AI, the customers' wish to have a single device for lots of applications, software-defined radio emerging from that trend too, and a desire for multiple devices connecting together temporarily to form a bigger device with more capabillities and spacial reach. the cummulation-point of all these trends are modular robots which you can connect and disconnect, preferably which can connect to eachother autonomously. so far adjuno is battling on that front, but it's lacking one important component, software-defined communication as is possible through fpga. and the modularity it offers is too big for creating hand-held devices. at least my search didn't come up with anything compact, maybe there already is. either way, more competants are needed in that area, or else we wont seen any useful DIY devices in the spirit of modular robotics...
IMHO the goal should be to push open hardware right in front of what is comercially available, the open hardware movement should offer some innovative stuff which you cannot buy elsewhere. take a bill of materials, some 3d-printer for the static components, and create robots for cheap. there are printable quad-copters. would they be banned nobody could enforce that banning. would they be expensive, poor countries would have them too (by replacing 3d-printers with actual hand-work and good image-processing as an adviser). and the same for any kind of autonomously moving small devices. billions of people are on earth, great opportunity to put them into some constructive use by allowing all of them to contribute to open hardware. the reason why you cannot is that machines are required for some stuff, so let people build those machines!