Final Prototype Design

The video above shows the operation of the final prototype. Similarly to previous designs, the final version is primarily constructed with laser-cut acrylic links and base plates. However, there are some substantial differences.

The first major improvement is the addition of 3D-printed links on each drive shaft, which are visible as the purple components in each linkage. These links include recesses for fasteners at each joint, which allows for a much more compact assembly. The doubly-supported couplers were changed to singly-supported because previous prototypes found that extra support planes had a tendency to interfere with each other and lock up the joints when load forces were applied. Finally, with the exception of the drive shaft, all of the bearings within the linkage were replaced with loosely-fit nuts and bolts. This change dramatically increases the ease of assembly while also enhancing the strength of the mechanism, with the drawback of additional friction at the joints.

The linkages are all driven by a single DC motor powered by a 9-Volt battery. The motor is mounted near the front and drives the front shaft through a timer belt-and-pulley system. It also drives a second timer belt across the side of the robot, which finally drives the back shaft through a third timer belt. The intermediary belt is necessary because the distance between the front and back of the robot is adjustable, so the two shafts cannot be directly connected. There are various points to attach idlers to the intermediary belt to tighten it for any configuration.

Prototype Evaluation

The final prototype successfully achieved the motion profile we designed it for across all six of its arms. However, it was not ultimately able to climb across the monkey bars as intended. There were two main reasons for this. The first is that the robot is not balanced while hanging on the bars. Having two hooks on the bars at a time does prevent the robot from swinging between the bars, but it can still rotate about the axis created along the two hooks. This issue could be resolved in one of two ways. The hooks could be modified to have a wider grip on the bar, thereby inhibiting the rotation, or the robot could be elongated and a third set of arms added. With three arms on the bars, the robot would be fully constrained at all times.

The second reason is that the hooks are not fully able to detach from the bars at the end of each cycle. This is partly because the lifting motion is less pronounced in the physical linkage than our analysis predicted, and partly because the hooks are fitted too tightly to the bars themselves. If the robot reaches a state where two phases are simultaneously attached to the bars, the mechanism becomes over-constrained and can no longer move. To fix this problem, we would first attempt to increase the height of the lifting motion by modifying the linkage design and/or manufacturing the parts with greater precision. We would also design hooks that allow for some flexibility in how they grasp the bar so that the mechanism does not over-constrain itself.