Older blog entries for jlin (starting at number 6)



In April of this year, I left my job doing machine learning work at Scribd, to join a new startup building robotic toys to teach young kids how to program. We call ourselves Play-i (supposed to be a play on "AI", but unfortunately most people seem to miss the connection). While we had been very public about what the goal would be, we were very hush-hush about the actual product. In all honesty, partially it was because we changed it so many times. We probably went through four completely different complete product ideas before arriving at the one today. Then we worked many long nights to quickly get the prototypes done. And now our robots, Bo & Yana are finally public in the form of a crowdfunding campaign. You can SEE THE ROBOTS HERE.

Our original idea was to actually create a robot where you can build the robot AND program it. But while testing with kids in our target age range, 5-8, we quickly figured out that we needed a fully-functional robot. We will have our own language for kids to program the robots, in addition to being compatible with Scratch, Blockly, and Objective C. There will be an open developer's API as well, which you can gain early access to by pledging at the level of the developer's pack.

We tested Bo & Yana with kids as young as 2 and as old as 17. The younger kids definitely gravitated towards "Yana", the smaller robot, who can detect gestures you make with it. Older kids enjoy the mobility of "Bo", short for 'robot'. Both genders enjoyed playing with the robots. Each robot works as a standalone toy, but they also can detect each other through IR beacons, for more interactive play.

After much demand, we are adding an additional accessories pack to allow connections from Lego Technic and Mindstorm pieces along with other standard 10mm ball attachment toys.

We have a contract manufacturer lined up to create the robots. While we originally prototyped with Arduino, the boards in the final prototypes seen in the video are indeed custom boards that we plan to use in the mass-produced version. We think our estimated BOM is pretty close. But there's still a lot of work to do, particularly with finalizing the firmware and the software. Our campaign is looking pretty good and we've already met our goal, so we are definitely hiring if you are interested in joining the team.

Finally, please feel free to email me personally if you have any questions about the robots. I know that the website doesn't provide very good information, particularly for those who actually know something about robotics.

Thanks everyone for all your support.
27 Aug 2008 (updated 12 Feb 2009 at 14:05 UTC) »

Wow. I have to admit, I feel a little guilty posting this, particularly since I haven't posted any personal entries in a while, but I'm up for the "Hot Blogger Calendar" and as of the time of this post, am 19th. The top 12 will get to partake in a professional photo shoot to be put in the calendar.

Define "hot" as you want. Maybe it's hot that I program robots. Maybe it's hot I got my degree from Carnegie Mellon. Maybe you like my picture (found behind the link). If any of these apply, vote for me.

Thanks!

Robocup at Hopkins has kicked off again. Looks like I will be slowly phasing out of the organizational roll and hopefully doing more coding. The students seem like they will need less guidance and pushing from me.

I just joined the Northern Virginia Robotics Group. It's mostly a group of hobbyists. They currently have a project going on that like a "Minature Grand Challenge". The original design was built from a Radio Shack Hummer, but looks like they are going to start building a new custom robot base using a milling machine one of the members just bought.

Still at HPTi doing non-robotics stuff for a living, but seems like we are getting more robotics people in the company. I am now the coordinater of recruiting CMU students to my company, so hopefully I can get even more and reach my eventual goal of working on robotics as part of my job. I will be at Carnegie Mellon's job fair 9/29/05.

Well I'm back from American Open. Driving 11 hours each way and going negative in vacation days in order to go wasn't fun, but it was worth it.

I arrived really late Saturday night of the competition, while that's about the time my students started driving there. They got to Atlanta around lunchtime and we all managed to stay up to 4:30 in the morning that first night to get things recalibrated and working. But we finally had a working demo. It wasn't what we were originally hoping for (two robots playing "Pong"), but we had a robot chasing a ball, shooting a goal, and knowing not to go out of bounds on the field.

The next morning we set-up for the demo, but the ambient light now present during the daytime made the wall look to be about the same orange as the ball. The robot went crashing into the wall and our whole camera mount crashed. It didn't matter that we brought 4 sets of everything. Before that crash we were already down to our last working set.

Eventually through much fiddling with finicky dongles, we got the robot up and running, ran our demo, and took a nice video of it (which I still have yet to put online). Enough of a success to make us happy and we learned a great deal from going through the whole ordeal.

Meanwhile I also got to referee and assistant referee a few of the small-sized games, since I used to be on CMU's small-sized team and am an experienced referee from past RoboCups. We also watched quite a few AIBO and Segway games as well as went to the talks. The craziest AIBO game for me was the finals, with CMU versus UPenn. I am of course loyal to my alta mater, CMU. UPenn scored within the first 10 seconds of the game. The game is made up of two 10 minutes halves. The game remained scoreless until the last 55 seconds when CMU finally scored. They went into 5 minute goal-to-goal which pretty much just means "sudden death". CMU scored almost immediately.

I unfortunately could not stay to watch their game against Dortmund, the winners of the German Open, since I had such a long haul back and had to go to work the next day. But it was good to see all my fellow-RoboCuppers again. Maybe next year I can go again, and maybe actually as either staff or a grad student at some university this time instead of having to use vacation days. =)

8 Mar 2005 (updated 11 May 2005 at 21:06 UTC) »

I keep meaning to post again but just have been so busy!

My team's now registered to be at American Open in May. Unfortunately it is during the students' exam period so not many of them will be able to come along.

We have a website up now at http://www.smokinjays.org. The videos and pictures are from 2001, but hopefully we'll have new ones soon.

The students have been very self-motivated, much more so than I expected they would be. We hope to get a demo ready by after their spring break.

The really really amazing thing is we did end up getting qualified for this year's RoboCup in Osaka. The students would really like to go as do I, but we're going to have to work really hard on getting funding to go since we didn't expect to qualify. We seem to be a doing a good job of staying on schedule to be able to have a team ready, though.

My personal goals is to have a more sophisticated strategy system. Coming from CMU's small-sized team, I really enjoy working on the multi-agent learning and cooperation. I really hope I can introduce that to mid-size as well through Hopkins's team. I hope to write a play learning module for the team. I'm also trying to coordiate cooperation between these two teams through Jim Bruce, creator of ERRT path-planning and CMVision.

Well, it looks like I did it. I have gotten enough interest to restart a RoboCup team at The Johns Hopkins University.

We will be doing the mid-sized league. At the meeting last night, there were around thirty students who showed up. I tried to emphasize that it would have to be a decent time commitment, but most of all it requires a lot of self-motivation to set aside time to come into the lab yourself and work on stuff. At the end of the night, I ended up with twelve undergraduate students and two graduate students who decided they wanted to commit themselves to the team. They seem to mostly be CS people but there were also a few MechEs and ECEs, all of which I find necessary to compete in this league.

The next step for me is to secure funding for the team. We will certainly need it for parts for the robots, travel, entrance fees, and shipping the robots. (Any suggestions?) Supposedly we already have some old equipment and even a large lab space left over from the last time JHU entered. I wonder what condition everything is in... My expectations are not high, but it would be nice to have something that was at least working, no matter how slow or outdated it is.

Also, now I get to figure out how to balance a full-time job, take a class or two for part-time grad school, and do this. Advising a dozen undergraduates by myself and probably doing a large portion of the coding for the team is going to be interesting...

22 Nov 2004 (updated 2 Dec 2004 at 15:00 UTC) »

I have now entered the real world, but sadly, at a job that doesn't currently involve robotics. I am hoping my company will eventually bid on robotics projects (my boss is pretty supportive of me pursuing robotics even though we don't work on it yet) or I will be able to switch to a company that does robotics, but meanwhile, I am trying to start a RoboCup team at The Johns Hopkins University.

I miss doing robotics research. I miss doing RoboCup. Luckily, I was able to find Prof. Greg Hager, who was interested in restarting RoboCup at JHU. The last time they entered was in 2001(?) in the mid-sized league. At that point the only two mid-sized teams from North/Central/South America were CMU and JHU. Now there are none.

To try to get interest, I gave a talk at JHU last Thursday. The students seemed decently interested. =)

I am hoping to convince Hager that we should start in the Legged League instead of reentering mid-sized, but I'm not sure if it's going to work. We'll see. But either way, I think I will end up doing at least some robotics research somehow.

X
Share this page