These actually look pretty fast and viable, and I've heard that people use using them pretty effectively.
I often recommend the TM220 series to folks interested in cutting to the chase and slamming out boards. Warts and all, it gets the job done. Dangerous Prototypes reviewed it in depth some time back. It's also a closed system. In the OpenPnP group we wanted to incorporate their PnP heads into our own machines but they weren't so readily available then and we were leery about committing to a single supplier. And that sums up our concern with the TM220 series. Other teams such as TempoAutomation are also going the proprietary route for cheap turnkey PnP. TempoAutomation has vision.
The goal of FirePick Delta is different. Our goal is to create an open, hackable PnP+ machine to engage the broader creativity and innovation bubbling throughout the world. An odd fact about the FirePick Delta Hack-a-Day team is that we are not co-located. We are scattered throughout the world and most of us have never met in person. We are therefore required to build our individual FPDs according to shared knowledge and cooperation. With three FPD machines under active construction (Florida, Netherlands, California), we're already seeing healthy discussions about design and implementation. For example, Christian notes that in the Netherlands, aluminum waterjet fabrication of the FPD base plates is cheaper than 2D laser cut plastic in the US (and I'm using bamboo for my base plates, just because). All three of us share the same design DXF. We even have differences about the frame footpads. I like modelling clay for its sticky anchoring, self-leveling and vibration damping properties. Others are opting for rubber feet. In a few months we'll no doubt get out our stop watches and have time trials. I'm betting on modelling clay.
This commitment to communal design evolution touches all aspects of FirePick Delta design and documentation: mechanical, electrical and software. For example, Simon and I have friendly arguments about the best way to estimate the pixels/mm resolution of the camera from a picture of 8mm component tape. He advocates Hough, I advocate direct use of DFT with a bandpass filter and peak detection for angle detection. Both are incorporated into FireSight software, and we will get out our stopwatches eventually to find what works best. Since we're all contributing to the same design, the benefits accrue to the team immediately.
This emphasis on collaboration and open design also make FPD ideal for schools. We foresee that a high-school team or student could put together an FPD and hack it to do interesting things. For example, an FPD is also, effectively, a robot microscope with computer vision. Cell counting, anybody?