Research topic & research question
My research is into the effect of maximum spring torque (the rated specification for springs on McMaster-Carr) on the throwing distance of a spring-based torsion catapult. As the torsion springs are slightly non-linear, changing the angle the arm is fired from, and thus the maximum spring torque, will likely not result in a linear correlation between angle and throw distance.
Discussion of related work
There are various articles in the literature about the design of catapult mechanisms for both throwing objects separate from the catapult (as is traditionally found in siege engines) and throwing of the self (found in jumping robots). Some previous work includes:
The work cited above looks extensively into catapults and their performance, be it for throwing objects or the catapult itself, but do not focus on the maximum torque of torsion springs. It instead relies on ancient techniques using torsion bundles or on a combination of torsion and linear springs, all of which have rated spring constants.
Methods used to investigate the research question
For my experiment, I used my group’s robot as the test setup to correlate maximum spring torque versus throwing distance. As the robot will have a fully automated and repeatable throw, it will be able to mitigate the effects of other variables on the outcome of my experiment. I will be able to adjust the maximum arm angle of the catapult to vary the maximum spring torque available and record the throwing distance for each angle. This angle is measured from the angle of the arm while resting against the crossbar as zero. In reality, the angle of the arm while resting against the crossbar is approximately 45 degrees from horizontal; with my angle measurement, an arm parallel to horizontal would be approx. 45 degrees from the crossbar position. This way, the larger amount of angular displacement of the arm, the farther the throw.
Evaluation Criteria
The throw distance will be compared at five different arm angles: 0, 15, 30, 45 (arm is horizontal), and the maximum angle of 52.5 degrees (arm is slightly below horizontal).
Figures summarizing results and description of all results gathered
I have created a graph that correlates the throwing distance of my robot and the maximum amount of spring torque available to the robot. It will likely be somewhere between a linear and exponential increase in distance versus the maximum spring torque.
Interpretation
These results show the correlation between maximum arm angle, measured from the crossbar, and the throwing distance of the robot. These results agree with my initial hypothesis, including some nonlinearity. The data is surprisingly linear between 15, 30, and 45 degrees, but tapers off slightly by 52.5 degrees as expected. This data is shown below.
Results during competition
The robot did not perform quite as expected, but not as a result of the results of this experiment. In previous trials, we had been reliably able to throw 11 feet; here, we found that we were more towards 10'6". As a result, we consistently landed beanbags at exactly the foot of the board, but not on it. This was due to the tensioner rubbing on the belt, which introduced sufficient friction to reduce our throw distance. However, the beanbags were repeatable enough that we were able to land beanbags on top of one another on several occasions.
Conclusion and future work
In summary, this experiment was used to characterize the throwing power of our robot, such that we were able to achieve repeatable throws of the nearly the correct distance during competition. Knowing the ideal angle to from which to throw our beanbag was incredibly useful during competition, as it got us very repeatable shots due to the limit switch sensor we implemented. Similar experiments would involve the height of our "cross bar", which stops the throwing arm, as this determines the trajectory of the beanbag (high arc vs. "cannon shot" straight throw, at the extremes). This might
This could be useful for applications such as jumping robots, which often use a sudden release of spring tension to “throw” the robot through the air. This is similar to work being done at Berkeley in the Biomimetic Millisystems Lab of Ronald Fearing: