EE C106A

The goal of our project was to have the turtlebot get to a set goal point while avoiding any obstacles along the way. This involved:

• Using sensor fusion to combine indoor GPS data and the odometry and IMU data from the turtlebot
• Using LIDAR data to find obstacles close to the turtlebot
• Solving an optimization problem to find the optimal path to reach the end goal while avoiding obstacles
• Using a path following algorithm to stay on that path

There are many real-world applications, as it is crucial to any autonomous system that the robot be able to keep itself and those around it safe. This project is part of a larger research project in the MSC lab on finding safe ways for humans to work with robots in manufacturing. Due to the limited time frame, we began by making certain assumptions: we only considered static objects, instead of moving ones, and we used boxes which were easier to pick out from the lidar data, instead of trying to recognize humans. In the future, the framwork that we built can be scaled to handle more complex obstacles and scenarios.

This project differs from existing packages such as SLAM in two major ways:

• It does not require a map, which means it can be immediately placed down and told to navigate without needing to form a map of the space, making it more computationally efficient
• It directly solves an optimization problem to find a path, rather than taking several pre-computed paths and optimizing over the set, so that in theory the path followed should be more effecient