I have been a member here
I have been a member here for 3 years. I haven’t ever published a project because to date I haven’t really done anything that remarkable. I have done robots that are autonomous, won’t fall down stairs etc, but nothing really, really cool that I think is outstanding and stands by itself. I am not a hardware guy, but the past three years I have learned a lot thanks to the many folks here at LMR.
Where I have been helpful to the community is my ability to write software since I have 16 years experience in the field. I did do several blogs that I feel are interesting.
https://www.robotshop.com/letsmakerobots/node/39818
This is an AI framwork that one can use with an Arduino. I taught a robot what the best courses of action to do were when it ran into an obstacle after about 10 minutes. What I have since done with this is to extend it to work in C# and attempted to make it much more generic rather than the specific implementation I published here on LMR. I have left it so it will be easy to port to Java or C++ if I decide to do that. The framework has extended from just a few hundred lines to almost 10000 lines of C# so it will do advanced event pattern matching, reinforce its learning etc. There are still some bugs with it so I haven’t put out another update on the code and what the library can do. I have it working with a Pong game I found online and after 20 minutes it beats the algorithm of have the paddle follow the ball whereever it is vertically. It also can learn complex behaviors on its own by randomly trying different actions and then verifying them as successful or not against its motivations. It also recognizes patterns from its experience that it hasn’t encountered before and will then work to try to find best ways to meet its goals the next time it sees the event pattern. I have found the work on this incredibly frustrating and fun at the same time.
https://www.robotshop.com/letsmakerobots/node/37339
From an architecture viewpoint, this is interesting and was fun to build. Ultimately, it was an interesting but failed experiment. I learned a lot about embedded development from the experiment. Knowing more about Arduino that I do now, I would use FreeRTOS to get the same multithreading, move it up to an Arduino Mega but maybe keep some of the ideas like Algorithm in separate class etc. I also learned a lot about memory management with C++ on an embedded platform.
I have in progress a blog to help people to deal with encoders on an Arduino so one can tell it to move forward an exact amount, backwards, turn so many degrees. I noticed there are a lot of questions on this issue and feel it might help. My code is kind of hard to follow so I have been trying to clean that up.
I never ask to do any of these reviews because I feel I am lucky to work in the US as a software engineer at a job I love where I make a good living. I know we are an international community and there are members who might struggle to buy things for their builds. But I am asking for a chance to review the Pixy CMUcam5 Image Sensor not for me but for my son’s robotics team at his high school. He goes to a small high school in NH who doesn’t have a robotics team. I have volunteered to be their coach although I have no experience doing anything like this. We will be starting in a few weeks, and my hope is to work the rest of the school year on getting kids excited about robotics and getting them some experience with robotics. We will be building very basic robot kits to help get kids up to speed on the basics of robotics. Next year, I hope the kids will be ready to compete in April 2017 for the Trinity College Fire Fighting challenge.
http://www.trincoll.edu/events/robot/
I am planning on using an Arduino Mega as the controller for the project. The Pixy CMUcam5 Image Sensor would be an excellent solution for the challenge since one knows ahead of time what kinds of things to look for like the flame from a candle, stuffed animals, etc. I will have a total budget of about $200 for the complete build which would push the Pixy CMUcam5 Image Sensor pretty much out of budget.
I hope you will consider giving this to my son’s robotics team. If you opt to do so, you will be helping the Parker Academy Robotics Team, and you will also be helping a number of kids to learn to work in a team, how to program an MCU, and how to apply algebra, geometry, trigonometry to solve real world problems. You will be helping to fire these kid’s imaginations, helping them along the way to be the leaders building cutting edge technology that will heat our global economy and ultimately be the technology that makes our world a better place.
I will have the kids from the team do the review although I will work with them to ensure professional quality writing and a scientific approach to putting the device through its paces. I will work with them while testing it, but ultimately let this be a learning experience for them. They will need to write up a test plan and follow the plan after reviewing literature about what the device is supposed to do. They will need to work in a team, figure out what to do to really test the hardware, effect of different lighting conditions, identify test plans, etc. and also have to come up with the review itself. During the summer and fall, we will be doing our build, and we will be sure to publish our results here on LMR, give robotshop.com credit for your generosity.
Thank you for reading this and considering allowing the Parker Academy Robotics Team to review the Pixy CMUcam5 Image Sensor.
Regards,
Bill
