This week we were able to start connecting some of the individual parts of our project. We’ve starting to link the frontend and backend of our web application, and export the odometry data of the Roomba to our processing and navigation algorithm. We ran into quite a bit of trouble installing Cartographer ROS onto the Pi, because there were many unresolved dependencies that are poorly documented. We were aware of Cartographer’s package problems before spring break, and spent an additional 5-6 hours trying to resolve them. After re-installing multiple times, we decided to switch over to a Python package we found called BreezySLAM. Although it doesn’t seem to give as accurate or granular data as we hoped, it works so we will continue with it for now. We also realized that iRobot’s Open Interface for programming on the Pi is in unreadable opcodes, so we used a Python wrapper called Create2Control to collect and structure the data from the Roomba.

Aditi was able to physically assemble the robot with tape for now, and it seems to hold up. This set-up is okay for now for basic testing, but will probably have to be more secure in the future. We’ve also started using live work sessions on Zoom in addition to the individual segmented work we do alone. This has been helpful in making collective decisions. In the next week, we’ll be pair programming even more, and be able to better evaluate our progress. Next week, we’ll be writing a basic navigation algorithm to control, given lidar scans and testing this on the robot.

For the in class demo, we plan to be able to show some form of control based on lidar and robot odometry data. Even though we spent/wasted many hours trying to install the Cartographer ROS packages this weekend, progress seemed to flow faster after we moved past it.

Updated Gantt Chart here

Updated Risk Management here