The Stabilisation Device
After building the ping pong ball launcher, the next small project I started working on was designing a mechanical attachment to stabilise the Tumbller. Yes, the robot still hasn’t arrived and it is looking increasingly like it might not at all. This however doesn’t change much as I can still implement my design by adding it to my CAD version of the Tumbller. The original plan was to have the robot travel through an obstacle course which would be quite difficult simply using its self balancing functionality. This is the reason for the stabiliser. The brief was kept very open so there are a load of different possibilities for this device. The only limitation being that it should be attached to the clear plastic platform on the front side of the robot. Here is another image of the Tumbller for reference in case you had forgotten what it looks like.
On a short side note you might notice that this render looks a bit nicer than previous ones shown here. I used Visualize rather than Photo View 360, which can often give a higher quality render.
Before looking straight into what attachment I wanted to design for the Tumbller I felt it was very important to get a sense of how the robot was actually going to move around. This of course is a lot harder to do when you don’t have the robot in front of you though. The method of motion that the Tumbller would use is called differential drive motion. This is where there are two wheels, mounted on a single axis, independently powered and controlled with two different actuators (motors), one for each wheel. In this way the robot can be turned by simply letting one motor run faster than another. By spinning the wheels in opposite directions, it is even possible to turn on the spot.
Knowing this, it is important for a stabilisation attachment to be versatile and able to turn sharply, otherwise the robot might run into difficulties with mobility. As the main purpose of the attachment is stabilisation, there is no need to for it to be driven and so it can simply be an idler wheel. This is very common in three wheeled differential drive systems. Bearing this in mind, there were a few different options that I considered implementing including a caster (swivel wheel), an omni wheel, a ball (rolling device) or a fixed rounded part. Examples of these are shown below.
Any of these attachments would likely be able to achieve what we need from them but there were some that I thought would be a lot more suitable than others. The easiest one to eliminate was the fixed rounded part for the obvious reason of there being a lot of friction with movement. It does have the advantage though of allowing movement in any direction and the frictional effects would likely not be too sever over a smooth hard floor so it might not even have been that bad an option. One problem I foresaw though was it getting stuck in grooves in the floor and having real problems getting over any sort of irregularity which could hinder movement a lot. The second option I eliminated was the omni wheel. Although I really like the idea of it and think it could work well, it is also very complex. It might prove very difficult if I did end up having to build it myself or have something 3D printed from solidworks. This is because all the smaller wheels are individual components that would have to be made and then attached to the main wheel. I wouldn’t even know how to go about that now so it could end up being a huge amount of work. This left me with the final two options, the caster and the rolling device. These are both very good options but each also have their own specific disadvantages.
A caster wheel can turn in any direction, unlike the driven wheels which just have one axis of motion relative to the robot. Caster wheels are also naturally self aligning to the direction of travel, when negatively castered. This is when the top pivot point is farther forward than the vertical center line of the wheel. This can be seen in the brass caster wheel shown in the image above. A common application of this is seen in the wheels of shopping trollies. Turning is made very easy as the wheels simply align themselves to the direction of movement. One disadvantage though is that if a reverse or sharp turn needs to be made there is a noticeable, unwanted change in direction while the wheels align themselves. This is called caster steer and is also common experience when using shopping trollies. Caster steer is more excessive the more the wheel is out of alignment. This is an important issue when precisely tracking the position of the robot with methods such as odometry. In our case accurate positional monitoring is not a high priority but rather simply getting the robot from A to B. For this reason caster steer might not be a huge issue.
A ball roller would not have this problem as it doesn’t need to align itself at all before moving in any direction. However, in the same way as for the fixed rounded part attachment I could easily see a ball roller getting stuck in any sort of groove in the floor and having real difficulty dealing with any irregularities. As well as this, if not on a smooth, hard surface but on carpet for example there may be a large amount of frictional effects. If the ball roller were to ‘sink’ into the softer surface, there would be a very large surface area in contact with the ground causing a lot of friction. This wouldn’t be as severe as with the fixed rounded part but could cause problems.
For these reasons I feel that the caster wheel is the safest bet and will likely perform the best in most situations out of the options mentioned above. There are still a lot of things I will need to play around when designing my caster wheel to make it suit the Tumbller. For example it would of course as well be possible for me to attach two caster wheels on the robot. This is something I will have to think more on and will definitely mention in the next blog when I actually show off the final CAD design. Hopefully I will be able to get that done soon.
Thanks so much for reading through my blog post. If you did enjoy it you should check out my Instagram page where you can get more regular updates on everything that is going on here. Bye for now.
