The goal of our project was to build a low-cost near 1:1 scale compliant humanoid hand with realistic joint motion and single motor actuation. Our final deliverable was able to meet all of our requirements, though we still have many ideas for future improvements.
While the hand could successfully grasp items, the low friction of the plastic fingers led to many items slipping, the simple addition of friction material on the palm and fingertips would tremendously increase the ability of the hand to hold items. Another area of improvement is in the spring stiction, currently the hand has problems with fully opening and closing. This is due to longitudinal forces that the springs can produce from side to side movement in their mounts. This can be fixed with springs better designed for this application and tighter tolerances for the spring mounts. Another solution would be a solution that can be better constrained such as a soft material or pneumatic pistons. Through our tests we found that 3D printed plastic parts were not the optimal solution for the fingers given the loose tolerances and being relatively soft. Using metal parts would improve the quality and durability of the hand dramatically.
A possible experiment for future groups to experiment with would be the addition of force feedback sensors inside the hand that use mechanical advantage analysis to accurately predict the force applied and the fingertips.
We would like to thank our TA’s Conner and Aayush for their help throughout the course, the TIW team for their help and use of their space in the mornings. Finally we’d like to thank Dr. Symmank for her valuable instruction and mentorship.