Crab robot

---

When I first started working at the crab lab, the only robot was a klann-linkage robot driven by 4 actuators. For the first several months, that was the robot I developed until we started working on a fully actuated hexapod. Once those were introduced, the klann robot took a backseat for several years.

Recently, we decided to revisit  the idea. In the past few months, I have worked on two different versions of the klann robot. My goal has been to use a simpler and more powerful actuator that is easily waterproofed, and to control more legs with a single actuator. My first iteration was a modular belt-driven robot, where one actuator controlled one pair of legs (front and back), and these leg pairs could be easily coupled to other pairs to adjust the number of legs. While this robot did work fairly well, I wanted to see if the robot could be lighter and more compact, so I started a second iteration.

After researching other 3D printed klann robots, I found that many had a system of 4 legs per half of the robot driven by a single central motor. I decided to implement this general structure on the existing klann leg setup, and after several gearbox and leg coupling iterations (I had much to learn about 3D printing gears), the full robot took form and a few weeks later it walked under water in a lake!

After designing the gearbox to be fully enclosed and water-jet cutting all the flat panels out of aluminum, the robot can now scuttle along beaches with ease! For this version of the robot, I am currently fabricating waterproof enclosures for the motors and electronics.

My next big idea for this robot is to add an under-actuated joint to the middle of the robot so each set of legs can be raised or lowered.

 Over the years, I have designed quite a few 3D printers. My first 3D printer was designed to be a small, high precision desktop machine. One main goal was to design the chassis with as few mechanical components inside so that it could eventually be thermally enclosed, allowing for a wider variety of materials to be used. Over the years, I have gone through several iterations of this machine as I learn more manufacturing and design techniques. Currently the machine is working, and I am designing a new frame that will be much more rigid, compact and easier to build.

My goal for the most recent machine was to have a very small modular lead-screw driven axis so that I could mount three together to make the machine. This is a project that I ended up revisiting several times over the last year, but due to COVID I was able to find time to make some meaningful progress. The final linear actuator contains only a few non-printed parts, yet remains rigid enough to cut wood.

In addition to making 3 of these linear actuators, I also decided to build my own spindle and collet system as the commercially available ones are too expensive for a prototype. In the end, I built it using a brushless RC airplane motor, 2 bearings to keep things centered, and 3D printed the rest. The collet is printed in a high-strength nylon to withstand the machining forces. While this does actually work for machining wood, I will probably switch to a more powerful full system eventually. Regardless it was a very fun side project.

 My goal with this project was to create a precise, high torque gearbox that could be easily applied to any medium/large scale robots I may want. While one can buy one of these, the price is much too high for a prototype, so I wanted to design a 3D printable one.

 While there are some 3D printable harmonic gearboxes out there, many of them are large or require expensive bearings. My goal was to make it as compact as possible, while eliminating the need for expensive components.  

This project is still in its early stages; so far I have built two prototypes. For the coming designs my goal is to continue shrinking the gearbox as much as possible with the constraints of the filaments I am using, while continuing to utilize inexpensive components. Additionally, I want to redesign the central spinning mechanism for more reliable torque delivery and higher precision.