Successful Kickoff of AI Racing League

Image for post
Image for post
We had an incredible first event for the AI Racing League. The audience included stakeholders from the AI community, researchers, teachers, and students.

Yesterday we did a successful kickoff of our new Minnesota-based AI Racing League! This was an intentional “soft launch” with minimal publicity and marketing since we didn’t want to have our first event be overwhelmed. Priority was given to influential people in the STEM leadership in the Twin Cities area and we put a focus on inviting people that could get access to women and minority communities. This was also during a workday so we realized that many people couldn’t take a day off from work which limited participation for many people.

The event was held in the gymnasium of the International School of Minnesota. Special thanks to Zach Sheffert for making the connections. This was a great facility and we had plenty of room to spread out and test the incredibly popular DonkeyCars. Note to self: teach people how to limit the maximum speed before they start driving!

Our sponsor was Optum Tech University (OTU). Both Paul Malley and Nicolle Swanson of OTU did a huge amount of work to make this event successful in a very short amount of time. OTU purchased the parts for 10 DonkeyCars, got the track printed, did all the logistics and even paid for coffee and lunch! And they did this all in a two week period!

We drew on volunteers from across Optum and other companies. Many staff from the Optum Advanced Technology Collaborative helped assemble the cars and be our first round of mentors. We had active participation from our summer interns and members of the Optum Technology Development Program.

Our goal for this first event was to introduce people to the ideas around AI in the classroom. We wanted the event to be fun and flexible, even if teachers and stakeholders could only stop by the event for an hour. We worked with organizations like CodeSavvy and MNCodes to make sure we connected to influential stakeholders.

When participants came into the gym they had the options of going to a set of “breakout tables” or they could directly join one of the teams getting a DonkeyCar to work. We had the following breakout tables:

  1. Hardware — we had two tables showing sample Raspberry Pi and Nvidia Nano hardware as well as sample servos, motors, batteries and visualization demos of pulse width modulation.
  2. Python — table with laptops used to teach Python with
  3. UNIX Shell — table with sample UNIX shell demos.
  4. GPU Server — table to show how our GPU server is used to do the training from images.
  5. Jupyter Notebooks — table with PCs to show people how to write Jupyter notebooks to analyze their image files.
  6. Computer Vision — table with many demos of computer vision such as face recognition. The Nvidia Nano runs these demos in real-time.
  7. Machine Learning — table with overall diagrams of the steps in machine learning and how the TensorFlow libraries are used.

Although we had a lot of activity on many of the tables — some of them lacked enough trained mentors and materials — so we still have plenty of content and training to work on.

We also had tables where participants worked on calibrating the DonkeyCars, took the cars for test drives around the track, attempted to gather training data and train their models. To be honest, we had several technical challenges and only had one car get fully trained. However, everyone seemed to have a good time and I think many people learned a great deal about what these events could offer.

We asked participants to fill out “Concept Card” forms where they could name a concept and describe what the learning objectives and resources could be. We are gathering these to start to build reusable learning components for other events. We are now in the “debriefing” phase and will be putting together a list of task to make the next event even better.

Special thanks to everyone that participated, especially Jon Herke, Rob Rossmiller, Parker Erickson, Sean Leary and Paith Philemon for their hard work making this event happen. I should mention that Rob’s team was the first car to complete the first lap, so they are technically the “winners” of our race to learn!

We also want to express our sincere gratitude to the members of the DonkeyCar community. Without your hard work and your willingness to share your code, your documentation, and your experiences this event would never have been possible! We hope to give back by creating a set of portable concept resources that could be loaded into a modern learning management system (LMS) that uses AI to recommend fun ways to learn AI. Kind of like that recursion!

We are posting resources on the CoderDojoTC GitHub repo for the AI Racing League here. Pull requests are being accepted! Feel free to add items to the task list for any new concepts ore resources you would like to see.

Written by

Distinguished Engineer with an interest in knowledge graphs, AI and complex systems. Big fan of STEM, Arduino, robotics, DonkeyCars and the AI Racing League.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store