Performance Evaluation
Our stair vacuum could not climb or clean any stairs, so we could not provide any quantitative performance evaluation. There were four performance specifications detailed in the StairVacuum Specs document. Our vacuum could not visit all stairs, vacuum maximum amounts of dirt, traverse each stair without leaving the test bed, or enter a mission complete state. This stair vacuum did however not cause any damage to any devices or the test bed.
Our robot was well below the 10 pound weight limit and stayed under budget. Out of all 4 of the 2022 StairBot designs, the maximum score achieved was a 0, while the other 3 groups were in the negative due to score deductions.
Strengths and Weaknesses
In the end, we constructed a robot that accomplished our main locomotion design, and we believe it to be not far from accomplishing stair ascent. Our robot is lightweight when compared to other groups due to our choice of materials. The robot also has quite quiet operation, and this is due to the locomotion method we designed. By only moving with one set of wheels in contact on the ground at once, this creates very little operating noise. The chassis design is fairly robust, but could be improved by reducing its size.
The major weaknesses of our design is the reliance on traction for stair ascent. This is extremely difficult to design and iterate upon, and even if done successfully, can be an unreliable way to traverse varying stair geometry within peoples’ homes. The rack is also not made out of sufficiently rigid material, and causes problems due to flexing. Our wire management was done in very little time, and is extremely messy. We would also like to mount the motors using standardized hardware instead of custom 3D printed pieces in order to keep the motors perfectly steady during locomotion.
Lessons Learned
This was the very first robot that either of us had ever designed or fabricated. We worked very well as a team, and all of our lessons learned center around the fact that iteration with hardware is extremely difficult. The one advantage of being a two person team is that communication was extremely easy, and this was the saving factor that allowed us to accomplish any progress at all through such a difficult semester.
The number one lesson we repeatedly learned throughout the entire semester was to start fabrication earlier. Every estimation we made for fabrication ended up being completely wrong, and it made it extremely difficult to deliver on weekly deadlines. In hindsight, this makes sense, as neither of us have a lot of fabrication experience, so we should have been generous with our timelines due to the lack of a reference point. This point plays into our second lesson, which is to just build and iterate as early as possible. The long fabrication times made this method much less approachable, but this is something that both of us will work towards for future projects.
In regards to mechanical design, a big lesson learned is to always use the most universal standardized hardware possible at all times. We used a rare miniature aluminum extrusion, built our own motor and sensor mounts, found obscure high-traction wheels from a website with very little customer support, and many other little difficulties throughout the design. Every single one of these decisions lead to hours of work to design and fabricate replacements. While doing this once might be trivial, when there are a dozen parts and multiple of them break, this sets us back by days. Another important note for mechanical design is to create a detailed CAD model early on which includes connection hardware, such as exact bolts, hex nuts, and tolerances. This might be an obvious point to someone with experience, but we never gave this a second thought until we realized it was a massive problem.
In a more positive light, we also learned that it was a very rewarding process to attempt to design a more novel mechanism and go through all of these seemingly unnecessary challenges. We feel that we did so much more than just approach a design challenge to make a stair robot - we learned how to approach engineering problems as a whole by going through this extra difficulty.
Future Work
There are a few easy iterations to be done to the design of the existing robot that will have a significant impact on performance. This includes replacing the panels that bolt the elevator to the aluminum chassis to be more rigid, as the elevator currently tends to flex slightly at higher tilting angles, forcing a lot of pressure on the spur gear. We would also opt for larger lateral wheels with a more porous texture, as these current wheels tend to have unreliable traction and accumulate dirt. We did not have enough time for final wire management, and this proved to be a problem when the elevator extended.
We believe that this mechanism is still applicable for stair climbing and cleaning with some modifications. The first iteration we want to do is to replace the front wheels with ones that have much higher traction. We even ordered, and have on hand, a set of very compliant wheels with a 35A durometer rating. These can squish as they press up against the front of each stair and further increase traction. It would also benefit for the chassis to be more compact, making it easier for the front wheels to maintain traction as the moment due to gravity is decreased. We also want to further understand the effect of moving the center of gravity further forward or backwards on the robot chassis. As the center of gravity is moved forward, this makes it easier for the front wheels to maintain traction, while making the elevator mechanism less effective. More comprehensive tests on multiple robot models could reveal some interesting characteristics. Since this robot relies heavily on a tilting mechanism, integration of an inertial measurement unit (IMU) is a natural next step to achieve more reliable locomotion.
Beyond the scope of this project, it is an interesting mechanism that might have applications outside of a mobile stair climbing robot. The elevator motion can be accomplished using arms or other mechanisms while still utilizing the idea of maintaining front wheel traction for vertical traversal with only a single, non-centered lifting mechanism.