Legged robots have to be some of the most interesting robots out there- intriguing people with technical know-how, and those without. However, much like the humans and animals that they resemble, they tend to become unstable after suffering injury to a limb and/or losing a limb. However, if we expand our search to the "creepy crawlies" we find that a six-legged configuration is far more common then two or four-legged. These also tend to suffer stability issues after limb loss, but it gives us a good pattern to continue- more legs.
This research study take the eight-legged configuration of spiders and aims to design a prototype capable of maintaining dynamic and static stability after leg loss.
Stability:
We can see why spiders, and the whole of arachnids, are an ideal candidate for this by looking at their support polygons (left top). Support polygons are a crude way of picturing how much area the center of mass can reside in to maintain balance. Outside of these shapes means you are now off-balance. The minimum stable shape that we want to aim for is a triangle, so having three legs on the ground at any given time in our gait pattern. By replicating the gait shown in the top gait sequence (where all eight legs are working) and removing leg number eight (shown left middle), we see that the first "step" with group one lifting off still has a triangle of support. If we repeated this process with a quadrupedal gait pattern (left bottom), the gait would have to change based to maintain the triangle.
Due to the nature of the project being a Capstone, there were basic constraints set:
Eight months to deliver prototype
Must cost no more then $2000
Only 5 team members
All components must be custom made or commercially available
In addition, all team members were also actively working on finishing their degree, and so the time able to commit to the project varied. Because of these constraints, the following rules were set:
Robot has no sensors
All custom components must be 3D printed
These rules made sure we kept a realistic design in mind, since most of us had little to no previous robotic experience. In particular, I wanted the team to have time to fabricate and test multiple prototypes.
Original Mock Design
Over the eight month timeframe, the physical design of the robot changed drastically over time. Originally, the robot's fourth joint was suppose to be a linear actuator, but the extra strength was determined to not be worth the complex joint design, and was scrapped. In addition, the 3D filament was changed to ABS because we could print parts and glue them together better (and cheaper) then PLA. This added significant weight, which changed our servos.
Once the major design changes were made, the electric system was then stress tested. Since the motors could draw a high amount of current at stall, we had to make sure the wires were not being overloaded. Once safety was ensured, the robot found its place on the floor, being tested for static stability. Lastly, was dynamic testing. We found that because of the quality of motors we were able to get, the robot would lose its strength when the command shifted. I.E. if joint one lifted, joint two would collapse when that command was sent from the Arduino. A support jig was made with tethers, so the robot was moving under its own power, but wasn't constantly falling due the aforementioned issue.