Question 1: What was the general strategy or plan behind your bot design?

While I was designing and building the pyramid we used for this event, it occurred to me that the chance of being toppled by another robot was pretty high. This led to a desire to build a robot that was less vulnerable to a roll-over situation. My first ideas were to build an invert-a-bot; a robot that can be flipped over and still operate as if nothing has changed. However it quickly became apparent to me that the largest wheels I owned would not allow enough space for me to build a robot of this type. I played around with the idea of having 2 separate drive trains, one on the top side and one on the bottom side, this got complicated and heavy in a hurry so I scrapped it. Then I remembered seeing several "robots" on the t.v. show Battle Bots that had various means of recovering from a flip. I decided to try using a mechanical arm. This whole line of thought led me to think that I needed to design and build a custom sensor that would detect roll-over. Once I got this design in my head, the strategy followed; climb the hill and stop on top, if you roll over, recover and climb again.

Question 2: What technologies/mechanisms did you use?

I designed and built a prototype home brew tilt/roll-over sensor, although in the end I did not use it. To replace the faulty sensor, I built a simple pendulum and attached it to my rotation sensor. This was used to detect fore and aft tilt and well as level ground.

The chassis was a modified version of Jennifer Clark's Skid Steer Loader chastise and drive train. The original has a width limitation of 8 studs which I attempted to stick with. Having worked with this same design before I knew how tough it was to get the whole gear train to fit in that tight of a space. Ultimately I decided to split the left and right sides of the drive train into their own chastise. This turned out to be a good thing because it gave a natural location to mount the lift arm assembly. This configuration is what inspired the name 'Scorpion'.

The lift arm assembly is fairly straightforward. It has its own motor which drives a worm gear which in turn drives a 24 tooth gear that has an axle that connects to the arm. The trick to getting the robot right-side-up again was to make sure the drive wheels were spinning against the direction of the arm and not with it since the robot would stand on end when the arm was fully collapsed.

Question 3: What was the most challenging aspect of construction?

This would have to be the home brew sensor I built. The idea is simple enough, take the circuit used in a 4 switch multiplexor wrap it around the inside of a brick, and use a ball bearing to make the shorts instead of switches. This would give you bottom, top, fore and aft contact detection as the ball rolled from one end of the circuit to the other, or if the whole thing rolled over. What turned out to be so tough about this was the assembly, and by assembly I mean the soldering. Lucky for me I happen to work with a bunch of electrical engineers who like challenges like this. After the prototype sensor was put together (by my co-worker) it was time for me to test it with the RCX. To my dismay the readings I got back were not consistent with the location of the ball in the circuit-cage. I brought the unit back to my co-worker who did some tests on it and in the end we concluded that the material I was using for the rails was not conducting the electricity around the circuit. Since this discovery was made the Thursday before the event, I shelved the whole idea and quickly built the pendulum instead. I think this sensor type is very viable and I plan on continuing to develop it. Ultimately I would like to see it detecting all six sides of the cube formed by the circuit-cage.

Question 4: Are there any special features you care to talk about?

I was particularly fond of the pendulum based tilt sensor. It was so simple and elegant. Also I would like to point out that with the pendulum detecting tilt, and the shrouded light sensor watching for the black plateau I still had the ability to detect roll-over. The light sensor could have detected the brightness of the ambient light in the room. I say could have because I hadn't actually thought of it until Rob pointed it out to me on event day.

Question 5: What surprised you about how your bot behaved/performed during the event?

Going into the event I thought my robot had no problems with climbing the hill, as it turned out this was not true. The robot would list a bit as it neared the peak of the slope it was climbing. This turned into slippage by one side of the drive train which in many cases turned into a self induced roll-over/tumble back down the slope. I was able to adjust for this a bit by shifting the RCX over towards the side that was slipping by 1 stud, yet I still had trouble getting onto the plateau.