While there are many ways that CleanBot could be improved, we are proud to present our hard work.
Presenting CleanBot with its various face:
Figure 1: Final product (CleanBot with its 8 eyes saying “Hello World!”
Figure 2: Front View (“Can you feel the stares~”)
Figure 3: Back and Side view with various components (“I’m Shy!”)
Swapping Mechanism
Video 1: Showing how a user would have to swap the cleaning mechanism.
Autonomous Motion
Video: Motion of the robot autonomously. The first part of the video shows the actual speed of the robot. As the video progresses, the video is sped up to not bore you.
The cleaning robot uses an S -shape cleaning motion, which is the “most” efficient cleaning method available. While there have been research showing that long runs of Random Walk algorithm and S-Shaped walk would yield similar results as well as Random Walk providing slightly more efficient cleaning, S-shaped cleaning motion would require lesser correction and intervention, as well as having a more evenly spread out heat map as compared to Random Walk algorithm (Source 1)
Ultrasonic Sensors HC-SR04 is chosen as the main sensor to reduce cost. There has been experiment performed that shows that this form of sensor is capable to doing obstacle detection. (Source 2) Our project has also shown than it was able to avoid clearly defined obstacles easily. (Seen in the video as the robot avoids colliding the wall)
Cleaning Algorithm Design
The design of cleaning algorithm is as such:
The initial idea is of S-walk algorithm, mapping and path finding. Rationale of S-walk algorithm is as above, being the “most’ efficient method that meet our needs. Mapping allows for the robot to identify its surroundings as well as recognise its location, prevent unnecessary trips. Path finding allows for the robot to find its starting point, allowing for the entire room to be covered, even if the robot does not start at the edge of the room.
However, due to the complexity and time constraints, path finding was not implemented. Mapping was not fully implemented as well, with only the most basic structure formed. The robot is unable to simulate autonomous pathing, which could allow for customs maps to be used (Such as in the form of excel)