Willow Garage Announces New Open Source Robotics Foundation 9
An anonymous reader writes "Willow Garage has announced the launch of the Open Source Robotics Foundation. 'It's always been the intention of Willow Garage to create an independent body that can take our initial work in open source robotics and see it grow beyond the confines of a single organization,' said CEO Steve Cousins."
OK I suppose (Score:3)
OK, I suppose, but being Willow Garage, it'll be OpenCV. It's got lots of nice algoritms implemented as modules, but it's very black-boxy and is (IMO) a terrible computer vision library, especially for implemting new stuff. Very clunky images, poor type safety and horribly slow numerics. Not exactly a fine example of C++ design.
Re: (Score:3, Informative)
The slow part is certainly true, though - but OpenCV is still a great framework for playing & experimenting with computer vision (and I'm not aware of any alternative that even approach its level
Re: (Score:2)
Okay, sure, OpenCV is terrible. But, what do you reccomend as the alternative? I think it's similar to ffmpeg in a lot of ways. Sure, the API has grown like a cancer, but as inelegant as it may be, what else has the same functionality and reasonable licensing terms that you can use to pull off the shelf and start hacking? Complain about the speed if you want, but if I use OpenCV, it certainly runs better than if I had to try to reimplement all the code myself! IMHO, part of the problem is scope creep.
Finally an end to byzantinism in this area? (Score:3)
A few of the participants seem to be from companies that also sell educational kits, while others are more academically focused. Seems like a good mix. It is good to have some level of "industry support." It seems to have been the case so far that many vibrant robotics communities have formed around particular product lines (e.g. Arduino), and this technical tribalism has served as a wall between groups that have much to offer one another. Having people from many of these companies in the same room along with those more inclined to develop generic APIs might help move things towards a more coordinated base platform.
Still, I think they should really see if they can tap someone heavy from the RTOS community. Most of the systems I've seen have been pretty heavy on statistical multiplexing in an environment where RT systems are pretty essential.
Meanwhile, in Japan... (Score:2)
But don't you need a robot? (Score:1)
This is great. But when a fully assembled robot costs 400,000 as in 400 thousand dollars, more than a house in a major metropolitan area, I am not sure how useful open sourcing it is.