This week was focused on 1.) integration of the program in real time and 2.) optimizations to the navigation and path planning

We connected all the components of the project together into a multithreaded program joining the Roomba, the lidar sensor, SLAM, and navigation system. This entailed writing control flow to make sure the program status was being updated as expected, and data was being passes smoothly between each thread. One by one, we were able to add each thread to the main program, resulting in a real time system. The robot can now move while executing a path, reading data from the lidar sensor, processing SLAM, and checking for obstacles.

After integrating the program together, we noticed it was running with a delay between each thread. To minimize processing time, we optimized the search and navigation algorithm to keep track of previously encountered obstacles and choosing farther destinations that would yield more exploration per path. We also implemented a program stop condition, indicating that the robot is done mapping when it is fully enclosed by walls, and has explored the interior.

Some problems that we continuously run into are inconsistent sensor datas from the roomba. Because the data from the roomba is very inconsistent and faulty, we chose to not use it at all for our program. This causes some challenges because we have limited data from the lidar as well, but it’s better than causing a fatal program failure.

This next week before the final presentation, we hope to optimize the program even more. We also will get the robot to transfer the user view of the generated maps to our web application.