For the past 2 months we have been running an experiment on The Zooniverse's project builder platform, called Agent NEO. We had 2 aims: 1) to improve our code for finding NEO candidates and 2) to shorten the time from observation of the candidate to reporting the result to the Minor Planet Center.
To do this we made a Zooniverse project, where each day we would upload thumbnails of 200 pixels around each candidate. We then ask Zooniverse volunteers if there is an asteroid in the cross-hairs.
After 2 months we are taking a break to evaluate what we have learned from all the measurements the volunteers have made. The volunteers were also very helpful in making suggestions to make the documentation clearer.
Some problems occurred during this first phase of the programme, some were due to our inexperience with using the Zooniverse project builder and supporting such a large number of volunteers. We have been able to clear up those problems, and learned a lot in the process.
We are taking a pause now for a few reasons:
- Monsoon in Arizona. Catalina Sky Survey is based in Arizona and has effectively shutdown for the summer because of monsoon. This is one of our 2 main sources of NEO candidates to follow up on. Without this we will have fewer subject sets to show to Zooniverse volunteers.
- We are down on team members. Our core team is going to be away in bits and pieces for the next month, and we are losing 2 members of the team between now and Thanksgiving. That means fewer people to support this program.
- Our code is misbehaving. Based on the volunteer classifications we need to do a lot of work on our code for finding moving objects. We don't want to waste volunteer's time, classifying things which are definitely not NEOs.
- Our turn around isn't fast enough. We are finding it really hard to get the subject sets to the Zooniverse, have them classified, then download the results faster than a couple of days. This is longer than it takes us to do the work ourselves. Its a really valuable lesson to know that, and it was one of the experimental parts of this project. It also meant that the classifications came in too late for us to be able to give volunteers credit for identifying candidates, something we really want to do.
We are going to take the next few months to update our code and consider how we move forward with Agent NEO.
We would like to thank all the volunteers for the efforts they put in to making this project successful. Your contributions have been extremely valuable and will play a big part in what we do next!
