I’m making a self driving RC car for my senior project. It will be guided by GPS waypoints and utilize GPS data as well as inertial movement readings to accurately localize itself. I am also currently toying with the idea of cameras and distance sensors so the car/robot won’t be so “blind”. But that is a stretch goal far beyond my current capacities (need to get the car to move first).
My previous (brief) experience with Arduino equipped me with a false sense of confidence when I bought a Beaglebone Black and started my project. Embedded Linux is something I have absolutely no experience with; but regardless, time for the fun to begin.
For the project I have:
- Beaglebone Black
- Sparkfun’s MPU-6050 Breakout Board
- Adafruit “Ultimate” GPS Breakout
- and a used toy RC Car
A lot of this project is inspired by Jason Dorweiler’s Autonomous Car.
Gameplan (rough, semi-overlapping, and in no particular order)
Get the sensors working with the BBB. I want to be able to get proper readings as well as play with some of the extra features of the two sensors.
Set up ROS. I want to use ROS (Robot Operating System) to handle all the scafolding for the “brains” of the car. Also there are some nice testing tools that I don’t have to write myself… hopefully this will save me time in the long run.
Navigation, localization and driving algorithms. Use Kalman Filter, dead reckoning and PID control to let the robot figure out where it is going.
Get a toy RC car and modify it. The steering in toy-grade RC cars is often just a motor that goes hard-left or hard-right, replacing it with a servo will give me much more control over the robot.
Integrate sensors, navigation/planning algorithms, and driving interface into ROS.
Tune PID and…
Maybe even have cameras and distance sensors to give the robot “eyes” and do obstacle avoidance…