Hi Jennifer,
Thanks for all of the positive feedback! It was great seeing the students embody the bees as part of their learning experience. From our observations they really enjoyed the process of transforming into the bees as part of their scientific inquiry into how bees pollinate flowers and collect nectar.

Regarding how OpenPTrack is able to track individual movements simultaneously, it is through a distributed computer vision network which is able to identify an individual through multiple depth sensors installed in the classroom. What this means in practice is that multiple Kinect One sensors (these are the sensors that are used for the Microsoft Xbox) are installed strategically around a room, then are networked together to allow each of the cameras to communicate with one another in real-time allowing each of the cameras to contribute to the person detection and tracking. Each person is tracked individually through first identifying them through a depth cloud that is generated by combining all of the sensors data in the network, then using machine learning that is part of the Robot Operating System (ROS: http://www.ros.org/) and the Point Cloud Library (http://pointclouds.org/) which are libraries that OpenPTrack is built upon to determine if a shape in the point cloud is a person. If the algorithms identify a shape as a person then a Unique Identification (UID) is given to that person. Then through the UID each person is able to be identified as an individual, and tracked separately. It is also noteworthy that OpenPTrack and all of the libraries and the Operating System it runs on (Ubuntu) are Open Source. More information can also be found on OpenPTrack’s Github site: https://github.com/OpenPTrack/open_ptrack and on openptrack.org

Thanks for the explanation. I will look into some of these just for information because I am curious about the technology. Great work!

Teresa, so glad you liked the project. I can’t tell you how enthusiastic our whole team is about this project. You ask a number of really good questions. I’ll try to answer them in different responses.

“Has all the research gone smoothly?”

The instructional implementation side of the research has gone incredibly smoothly due to constant communication and open collaboration. We meet weekly with our collaborators at Indiana University to share ideas, challenges, and successes. At both research sites, we also meet regularly with our amazing collaborating teachers who helped us to anticipate or adapt to any surprises that did come up. So even when we missed several days in a row in Indiana due to snow and freezing rain, the teachers helped us adjust the curriculum to work with their new schedule. Snow was less of a problem here in Los Angeles.

The technical side of the project has also gone smoothly mostly thanks to the efforts of Inquirium (http://www.inquirium.net/index.html), Open Ptrack (http://openptrack.org/) and UCLA remap (http://remap.ucla.edu/) who worked together on all the technical aspects. But the STEP project is a complex socio-technical system and their contributions by no means merely technical. What made this project work was a shared vision for how the technology would fit with and extend social and physical relationships.

“What have been your biggest challenges?”
One of our biggest challenges has been technical—helping the system to keep track of who is who. What is new about our work is that we are not tracking one or two children but are tracking up to 12 students at one time in a relatively large space. Our tracking system is based on visual recognition, so occlusion (when one child walks in front of another child for a given sensor) can confuse our tracking and have the bee avatar of one child swap with that of another child. We have been making good strides towards solving this problem, but currently our work around is to limit our designs so that everyone plays the same type of character—in the video you’ll notice everyone is a bee. This way if the system gets confused and swaps one child’s bee for another bee it is not too big a problem. This limits our design, but still leaves us plenty of room to design interesting and engaging activities.

“Are you working with students from underrepresented and underserved at all?”
We have two sites one is the UCLA Lab School and one is a public school in Indiana. The ethnic diversity at the UCLA Lab School attempts to mirror the demographics of California. I will let Joshua Danish, the co-PI on the project, chime in on the population of students at Indiana he has been working with.

“How have groups of students varied?”
One way the students vary at UCLA is that 2 of the 5 classrooms were dual immersion Spanish-English classrooms. We have not seen any difference on our pre-post results from these classrooms. More importantly, a core belief of our designs is that our activities are intended spark and support productive classroom discussions. We are very interested in how, if at all, these conversations differ when they occur in the context of a dual immersion classroom. And, of course, the students in Indiana are at a completely different school, in a different state. The Indiana school is also a public school, and thus faces very different constraints.

Thank you Noel. What terrific work! And thanks for letting me know about your work with dual immersion Spanish-English classrooms and schools that mirror the demographics of CA. I will followup with the websites you shared to learn more as I am very intrigued and love what you are doing! Thanks for making a difference and engaging young students so meaningfully in STEM!

Thanks for being so forthcoming with your challenges; I’m a firm believer that that is where learning lies in the design process! Good luck addressing the few that have come to light. I like the solution you have proposed above. Kudos on a GREAT project.

In the case of the IN implementation we saw similar discussions in the walk down the hallway. We also didn’t do a lot of work outside of the technology, but our teachers discussed seeing some real potential for that. In the next round of implementations, they have indicated that they hope to plan further integration with their other forms of activities such as creating artwork and presentations, visiting content experts, etc. We have set aside time to meet as a team (researchers and teachers) to implement this next time (as part of the PLAE project and other future projects).

Love it. Play begets more play!

Hi Noel,

In terms of analyses my preference is to start out with exploratory data visualizations. The spatial coding seems tricky… I’m guessing you’d need to look at both allocentric and egocentric measurements. For the temporal analyses I’ve had some success using Markov and Autoregressive models. Please feel free to email me (rogertaylor@gmail.com) if you’d like to chat some more about it.

Hi Brian,

Thanks for your comments and question. One of the things our partner teachers in IN noted is that even when we didn’t explicitly introduce them to bees in advance, the students already know about bees and see them on the playground and at home every day. They believe (and I agree) that this easy familiarity is one of the reasons that bees are so compelling.

Also, in Bloomington our wonderful Science Museum (Wonderlab) has a bee exhibit where you can actually watch the bees dance inside the hive as real bees return from their trips to pollinate the grounds outside the museum. Many of our students have seen this in the past, and a targeted trip to visit this exhibit is something we’ve discussed for future iterations.

Thanks Leslie! Great question. My feeling as we dive into the data is that the embodied play experience helps the students to explore aspects of the system and it’s interactions in ways that make the system-level properties far more visible to them. They also see how these system-level properties are important in ways that I think are not clear to them in “traditional” classroom activities. For example, when we’ve told students in the past that bees don’t collect nectar on purpose, many memorize that fact, but I’m not sure they appreciated it. However, when in STEP we have them collect nectar to get food for their hive, they settle into that role and really attend to nectar collection, forgetting about pollination. When we then turn on the pollination “layer” of the interface, they see how they didn’t actually need to try and pollinate because the flowers that drew the most bees were pollinated quite successfully. I think they begin to truly appreciate at that moment how these systems can interact productively without any need for centralized control.

The other big difference is that we are really focused on systems level thinking whereas few classrooms explore content in such a manner. Instead, the bulk of classrooms that we have seen exploring bees are focused on features such as their body parts. In some of my earlier work, I focused more on those distinctions (http://www.tandfonline.com/doi/abs/10.1080/1050...). In the STEP work, I think we really take this to the next level by using the students’ interactions with each other and the software to appreciate all of these complex dynamics by acting in fun and natural ways.

Thanks! If you like, you can also see our first publications on the STEP platform in the context of children playing at being particles as they explore the states of matter. The citation is:

Danish, J. A., Enyedy, N., Saleh, A., Lee, C., & Andrade, A. (2015). Science Through Technology Enhanced Play: Designing to Support Reflection Through Play and Embodiment. In O. Lindwall, Häkkinen, P., Koschman, T. Tchounikine, P. & Ludvigsen, S. (Ed.), Exploring the Material Conditions of Learning: The Computer Supported Collaborative Learning (CSCL) Conference (Vol. 1). Gothenburg, Sweden: The International Society of the Learning Sciences.

Or if you email me I can send you a copy!

Joshua

Could you email me a copy (rogertaylor@gmail.com)? :-)

Sent!

Thanks Sylvia! I am sure we can come up with lots of ways that your kids could work with our system. We’d certainly love to work with you again too!!