Testing Done in Autumn Lab

Claws on RoboToe?

Follow up to the discussion on whether large preloads are needed to attach claws.

Andrew ran some RoboToe? tests over the weekend with steel claws on various surfaces, and came up with some very interesting results.

1) Claws can be very effective on some surfaces with tiny preloads (less than 1/10 preload:shear ratio). Other surfaces present significant obstacles to claws. Some materials may require setal adhesives, and/or very high claw preload impulses. We are still testing surfaces.

2) Concrete: claws can generate very large shear forces with minimal preload (less than 0.1N). We observed stick/slip dynamics caused by local failure of the concrete. This underscores Bob's point about the advantages of multiple attachment points (6 legs), and my point that the robot will need to be engineered for slippage as a common occurrence.

3) Plywood: Claws can generate nearly arbitrarily high shear forces with low preloads (less than 1N). RoboToe? reached 12N shear, and ripped the plywood off the sensor.

4) Hard tree bark (fallen, dry wood) is very hard, and lacks concavities present in concrete. Most trials failed to attach claw, or caused delamination of bark. Some trials yielded respectible shear forces (8N) with relatively low preload (0.7N). Andrew will run some more trials using bark from live trees.

5) Painted wallboard: surprisingly hard for claws to penetrate. Even very sharp claws with large preloads (2N) failed to penetrate the paint.

6) Design ideas: Consider claw deployment mechanism internal to the foot instead of using leg forces. This could be similar to a staple gun, where a small motor can wind a stiff spring to be fired when needed to generate large claw preload impulse.