#9: Three-in-one newsletter
EOH, leg roll, and solving for the optimal series-elastic actuator
Welcome back to Pintobotics!
Thank you for waiting, we are publishing this newsletter a week delayed because most of the work this time is theoretical, and we want to spend the time to go into more technical detail. Fitting it all into one email would be cumbersome, so in addition to this newsletter there are two extra Substack posts linked below, with a brief summary each.
EOH
We have applied for UIUC’s Engineering Open House! On April 5-6th 2024, we aim to demo the robot squirrel’s abilities on one of the trees on the quad. The plan is to mark out an area near a tree and allow visitors to briefly remote control the robot under our supervision. We will also display prototypes, molds, and test equipment used in our engineering process. Should be fun!
Leg roll mechanism
To attach firmly to the tree, we found that the robot needs clamping forces at the grippers. However, space is very limited on the robot for leg roll actuation, so to make the assembly compact, we decided to connect the two legs together and use one servo to control the roll of both legs simultaneously.
The isometric view shows the legs retracted to normal position, and the front view shows the legs spread to the maximum position for perching on a tree. The actuation of the servo would provide a clamping force that increases the friction between grippers and the tree bark.
The asymmetry causes the legs to move at slightly different rates in the middle of the extension. This would likely be fine because the movement would happen very quickly, and we can even take advantage of it to slightly adjust body roll if both feet are planted.
This extra post is about figuring out how to get a motor-spring system to accelerate a mass to the highest possible speed within a given time and distance, relevant to the mechanical design of our jumping squirrel leg. First, a review of ideas from literature and intuition, then a description of process of generating the optimal strategy given a model of the system.
A gradient-based optimization algorithm searches for the best transmission ratio profiles (how the mechanical advantage changes over time) that achieves the most energetic jump, and there are two different emergent strategies: direct-drive and hold-release. Direct-drive does not use the spring to store energy and immediately accelerates the leg, while hold-release keeps the leg unmoving to wind up the spring then suddenly releases it to convert elastic energy to kinetic energy.
The preferable strategy depends on physical constants and constraints: hold-release is preferable for systems with powerful motors, lower mass, longer time, and shorter legs, likely because they are more likely to hit the leg extension limit.
This extra post is about implementing the hold-release strategy using a latch to keep the leg in place and a (8-bar?) linkage to optimally wind up a spring.
Prototype of the latch mechanism:
Possible linkage to stretch a spring using nearly constant motor force, involving a twisted string:
Can't wait to see this at EOH 🐿