Ghost Rider
By: Anirudh Kuchibhatla, Collin Allen, Nicholas Glaze, Michael Angino
Team DISSECT
ELEC 491
Project Goal
We set out to explore the emerging field of autonomous driving and the complex technologies that make it possible. While most people immediately consider cars and drones as targets for this technology, we decided to attempt a new challenge and implement self-driving capabilities on a functional RC motorcycle. We ultimately hope to design an autonomous motorcycle capable of reacting to its environment.
Technical Challenges
There were many aspects of the project that we needed to consider when working on this project.
One of the biggest challenges we faced was ensuring the motorcycle was able to travel at sufficiently fast speeds to remain balanced while still being able to react to its environment. This was especially problematic while testing our motorcycle, as we had to drive it slow enough to keep it in control but also fast enough to remain balanced. We have temporarily addressed this issue with training wheels, but will need to create a more permanent solution in the future.
Another challenge with the motorcycle is that it has a low weight capacity. This is another issue we have not addressed completely.
The limited onboard computing capacity is also a challenge in creating a self-driving motorcycle, however, we have not faced this issue yet. This is because we are currently relying on a simple distance sensor to recognize when to stop the motorcycle, and thus, do not need much computing power for this task. The limited amount of computing capacity will be a big hurdle in the future, when we utilize more complicated machine learning techniques to recognize different objects as the motorcycle is driving.
Project Accomplishments
The first major goal we achieved was to understand how the original remote controlled motorcycle functioned. We discovered that the remote control sent signals to an RC module, which sent instructions to a servo and a brushless motor which controlled steering and throttle respectively. Furthermore, an internal flywheel in the back wheel of the motorcycle helped stabilize the motorcycle.
We then were able to bypass the RC module by directly connecting the servo and throttle to the Arduino and were able to successfully input a simple path for the motorcycle to drive on.
Finally, we were able to successfully integrate a distance sensor to cause the motorcycle to stop within a certain distance in front of it while it is driving.
Next Step
We hope to next integrate cameras on our motorcycle to record video and take images while it is driving. We will then implement machine learning techniques to identify various objects and scenarios to make the motorcycle more intelligent. Some sensors we hope to implement are LIDAR, radar, and motion sensors on the motorcycle.
Picture
Ghost Rider in Action
Comments
Please log in or sign up to comment.