An Experimental Approach to Flipper Geometry
Having developed the methodological design process through formal schooling, I wanted to apply the process to something I could test in the real world. Specifically, the design of high-speed pneumatics in combat robots.
Aug - Sept 2018 / 4 weeks
Role / Skills
Independent Project | Mechanical Design, CAD, Fabrication, Experimentation, (light) Physics & Math
Rhino 3D, Photoshop, Lasercutting, 3D Printing, Machining
Designing combat robots has always been an intuitive process for me. Experience and precedence hone my intuition, but I never truly know whether my design will withstand the forces of battle. In the smaller weight classes, things tend to sort themselves out in curious ways. Weapon belts stay on the pulley long after the pulley flange shears; electrical connectors hold things together when the armor is long gone.
For my latest build, I wanted to push my abilities– create the most powerful antweight (1lb) flipper possible. A flipper jabs a pneumatic arm the length of its entire body underneath its prey and viciously flips the enemy high into the air, inflicting damage through the impact of a long drop. I wanted to launch opponents as high as possible, potentially up and over the arena wall. I had small-scale pneumatics parts from a previous project.
But what happens when intuition can only take you so far? Say, when designing the arm geometry of a pneumatic flipper? Unlike relatively simple energy storage formulas for spinners, a pneumatic system is not so easy to quantify. There is the compressible nature of gasses, and force equations in which the opponent is not hit but shoved.
When the complexity of a problem outmatches intuition or experience, experimentation precedes. I didn’t have the technical know-how to work through the theoretical calculations, so I did the next best thing: try a bunch of flipper arm designs to see what works best.
Frame of Reference
I first looked at flipper precedents for clues to focus my experimentation. I collected notes about the arm geometry of Team Inertia Labs’ top machines, notably “Bronco” and “T-Minus”. While they are in the larger weight classes (60lb +), their machines are consistently the most powerful flippers in the sport. I looked at both the obvious design (use of a three-bar geometry) and the small details (like length ratios between the cylinder piston mount and rear pivot arm, and arm tip length).
Through photographs and video, I mimicked the arm geometry of one of their more documented designs and scaled it down to fit the proportions of an antweight. From there, I created variations off this base geometry. Modeling in CAD helped me visualize the arm movement across start and end positions. Once I had created five different arm geometries with differences in angles, lengths, and heights, I lasercut the arms and support jigs for testing.
When it comes to testing, consistency is key. I mounted my camera on a tripod for a consistent and repeatable frame of reference. After filming launches of a wooden test block from all five arm designs, I overlaid the video frames to see the maximum height, number of rotations, and final distance of each wooden block.
From the video frame overlays (Figure 6), flipper arm #2 provided the highest launch and the least spin. While spin isn’t inherently bad, it diverts energy from the maximum launch height. I was happy with arm #2, so I fit it to my final design, making slight adjustments such as increasing the tip length to better get under the opponent.
However, my end goal was not to figure out the best flipper geometry, it was to create the most powerful antweight flipper. A local competition, MassDestruction X in New York, served as a perfect real-world test with its “plastic ant” class. Competitors in this 1lb class were limited to plastic for weapons, armor and structure. This allowed me to optimize flipper geometry without worrying as much about proper armor against the current breed of vicious antweight spinners.
With the competition fast approaching, I 3D printed the final arm geometry and constructed the rest of the frame with basic machine tools. I finished wiring the robot together in my friends’ Manhattan apartment the day before. The completed robot came in over an ounce underweight, ready for combat.
So, you may be wondering, how many opponents did I bounce off the ceiling? Well, to be honest, the actual performance of the pneumatic system in the competition was a bit of a letdown. I easily flipped all my opponents, often multiple times, but the flips themselves were not as impressive as I hoped. Maneuvering the robot was also more difficult as a last-minute transmitter issue forced me to operate the drive controls in an unmixed, two stick arrangement.
What were the takeaways? I focused so much attention on the flipper geometry that I overlooked other important considerations such as drivability, weight balance, and self-righting. In the end, those considerations became greater limitations than the prospect of how high I could flip the opponent. Then again, my primary focus here was on the flipper geometry. For the next competition, with real spinners and greater stakes, I can focus on a more holistic machine.
However, for the next iteration, there are still significant improvements to be made to the pneumatic system. I haven’t discussed pneumatic considerations outside of flipper geometry, but optimizing the system for maximum air flow is the next focus. I have ideas for dual solenoid values and increased port diameters. Additionally, I am not convinced that the three-bar arm geometry is best… perhaps a four-bar linkage will be better optimized for the given cylinder size. Naturally, more testing is in order.
The beauty of the experimental method is that it doesn’t require a great depth of knowledge for a successful result. All it takes is a fair bit of patience and diligence. A healthy dose of intuition, approximation, and experimentation can take you to great lengths.