As far as I understand, AMT is only available to US users, wheres I am an NLP researcher based in the UK. I am currently working on a small project to build an initial seed corpus of English robot commands. Annotators will be shown up to 1000 scenes consisting of paired before/after images of a robot arm in various positions with a board of blocks, and asked to annotate a natural language command that a robot might be expected to execute to rearrange the board from the first image to the second image. Once this data is collected, the corpus will be made freely available to researchers under the GNU public license. I am looking to collaborate with someone in the US who has access to AMT. I already have funds to pay for this task. I'm hoping to work with someone interested to design and implement the data collection task. As a second (future) step, a statistical model may be trained using the annotated corpus that we collect in order to see how hard it is to train the robot based on this data in order to execute spatial commands. So anyone with experience in this kind of work would be a bonus, but not essential.
Anybody interested in collaborating? I already have the images and will happily contribute the funding to pay for AMT. I'm looking for someone with experience in AMT to make some significant contributions to helping me desk the AMT task. I don't think that this should be too time consuming given that we already have the images and that the task should be more or less clear to annotators.
Looking forward to hearing back from you on this corpus construction idea!
Kind Regards, -- Kais Dukes