Intelligent Autonomous Robot for ME218B
In Winter of 2011 I teamed up with two other mechanical engineering students to build this intelligent autonomous robot as part of the ME218B Mechatronics class at Stanford.
The robot had to be able to navigate an 8' x 8' playing field autonomously and knock down plastic targets as commanded by a wireless central computer. It consisted of a three-story modular architecture with the power train, embedded processor and circuitry, and Nerf ball tipped robotic arms occupying each tier respectively.
If you'd like to learn even more about our robot, please visit this site which was created to document the project.
|
The project was coded in C and ran on a Freescale E128 microcontroller. Its logic functionality was handled using a state machine, and depended upon significant filtering and sensor fusion in order to allow the navigation system to correctly estimate its position. Overall, our system was able to request a target from a central computer, select a navigation path to the target, and then traverse the field and attack that target without any external help. The system proved very robust and could reliably acquire, find, and attack eight consecutive targets within a span of two minutes. The final implementation of our navigation strategy allowed the robot to dynamically recover knowledge of its position even if knocked out of its expected location by an external bump or collision with another robot. |
 |
The bot kept track of position using an algorithm that included data from IR range finders combined with dead reckoning. This sophisticated navigation strategy, coded from scratch, allowed us to produce a 2D visual map of the playing field and the robots trajectory after each match.
In order to meet the competition requirement of knocking over both high and low targets placed around the arena, we equipped the robot with two servo controlled attack arms that functioned together to hit targets at multiple heights.
Our system's overall structure was designed in SolidWorks and then the build platforms for its main modules were cut out of masonite using a laser cutter. The images below on the left show the correlation between our CAD model and final system; at right, the images show close ups of each module in our three tiered system. These consisted of the robotic arm module, microcontroller and electronic circuitry module, and power train module respectively.
   |
   |