Drawing using Inverse Kinematics and a 5-Bar

The primary objective was to autonomously draw a closed square on a standard letter paper using the provided pen. The workspace for this task was defined as a cube with side lengths of 33 cm. The position and orientation of the paper was unknown to teams in advance aside from being guaranteed to be entirely in the workspace and in one of seven possible orientations; [1,0,0] [0,1,0] [0,0,1] [1,0,1] [1,1,0] [0,1,1] [1,1,1]. The pass/fail requirement for this task was that in at least two trials with two different normals, teams had to draw a closed shape with 4 identifiable corners. To count as a corner, a line on the page had to have mismatched tangent slopes at the putative corner location and the shortest side length of the shape had to be at least 4 cm.

Teams were not allowed to manually lock any degrees of freedom, except for the last degree of freedom before the pen, and that degree had to be a 1-parameter screw motion. In the autonomous phase, teams would be given the (X,Y) target coordinates and the scale S. After inputting those values, the team cannot interact with the robot again while it draws a square of the input scale at the input location.

For our final design, we chose our 5-bar linkage design, but the method of both mounting the pen and orienting the linkage are unique. In an effort to reduce the weight being supported at the end of the linkage, we opted to take advantage of the ability to manually actuate our final degree of freedom. Rather than mount the pen on a motor at the end of the linkage, we mounted the pen on a large cardboard screw with the screw itself serving as the axis about which the two middle links rotated. The screw could be manually adjusted to change the distance between the tip of the pen and the linkage, controlling our last degree of freedom.

While our team had by far the most successful robot performance on P-Day, we still fell short of our robot performances prior to P-Day and there were several areas that we could improve upon. In the days leading up to P-Day, we ran tests trying to account for all of the systematic errors we observed. This led to us having three sets of about ten variables accounting for each consistent error we observed. These variables were just set integer values that we added to the initial conditions, but optimally a multiplier or a linear equation would have been better to account for the error. For example, when drawing vertical sides of an 8 cm square, they often measured only 4-6 cm, whereas for a 4 cm square, the vertical sides were about 3 cm. This inconsistency suggested that a multiplier would have better adjusted for scaling errors than simply adding 2 cm. Due to time constraints, we opted for a fixed approximation that ensured our square met the requirement of having sides at least 4 cm long to satisfy the pass/fail criteria

A structural highlight of this robot was that I made a inset screw to alter the pen height... out of cardboard.