Summer Projects, 2017
I recently had the pleasure of participating in the “CURM”, the Convention for Unconventional Robotic Movement, hosted at the University of Texas at Austin, where I demoed the latest version of the Glussbot.
Some of the students there expressed an interest in possible summer projects related to the Gluss Project. This is a list of potential projects. Since I live in Austin, this is convenient for UT students, but I’m happy to work over the Internet with diligent students where ever they may be, although some of these will be much easier to do if you are closer to Austin.
This is work would be closely supervised by myself. I will provide all materials and/or cover all incurred expenses.
Public Invention works transparently, and does not seek patents. All work is published, eventually. Simple work is published at GitHub. When warranted, we try to publish in peer-reviewed academic journals and conferences. Additionally, we attempt to publish in trade magazines such as Servo and Make Magazine.
If you are interested in working on one of the projects, please email me at <email@example.com>.
We would like to place coils of magnetic wire in small vessels of magnetic wire and see if we can make the fluid move. In particular, we would like to place two coils on a plastic straw and see if we can either pump fluid through the straw or move the coils on the straw. This will require a basic understanding of electromagnetism and V = IR.
In-place Assembly Printing of the Song-Kwon-Kim Turret joint
Based on models already produces in OpenSCAD, we would like to be able to print the complex Song-Kwon-Kim joint as a single print operation. This would be done with a dual-extruder head on a 3D printer (Public Invention has a LulzBot Taz 5). The support material would be printed in PVA which can dissolve in water. This will probably require a lot of trial-and-error work with the 3D peinter, and a certain amount of solid model using OpenSCAD.
Mount a Camera on a Glussbot and Figure out Something Useful to Do With It
Mount a camera on the “head” node of the glussbot and discern a practical application.
Use Open-source Software to Perform a Static Structural Analysis of a Glussbot Configuration
These projects are challenging if fully completed. I suspect each might be be publishable or patentable if perfectly completed. Public Invention, however, does not seek patents, but we will attempt to write academic papers if convenient. In a summer project, limited progress on these projects might be expected.
Development of an Optical Linear Sensor
We have already demonstrated in principle how an optical linear sensor using just a CDS cell and an LED could work. However, this project needs to be made practical. This requires simple Arduino programming and basic circuit design. Additionally, the ability to 3D print the sliding sensor housing could be very valuable. This may also use 3D printing using conductive PLA as a material. Finally, making this accurate and “digital” in its nature should be possible and would be extremely valuable.
3D Printing an in-place Gluss Controller
Ideally would be be able to producing a functioning tetrahedral gluss controller as a single 3D printable object. This will combine mechanical and electrical engineering.
Add a Modular ToolHead to the Glussbot
The Glussbot currently does not have an actuator or tool head. We would like to add a “grasper” or some other tool to one of the nodes of the Glussbot. Since the joint is complex, this is not a trivial problems. Additionally, we will have to decide whether to architect the device as a separate micro-processor controlled device reusing the exisiting Arduino Mega shields. Any toolhead should be modular, potentially allow the operator to mount different tools.
Machine Learning of Gaits
The 5-Tet Glussbot crawled. A gait for the 7-Tet glussbot has not yet been demonstrated, but it has 6 pseudopods on the ground, so probably many robot gaits that work for hexapods will function for it. Either by hand or via machine learning, using the existing Javascritp/Ammo.js physics simulation of the 7-Tet glussbot to develop functional gaits for the physical robot.
Develop Mathematics of Tetrahelix Curve Fitting
The current glussbot is based on the geometric structure known as the “tethrahelix”.
Tetrahelices form straight lines in space. Develop the mathematics necessary to the allow the tetrahelix (mathematically, not the robot) to conform to a curve in space. This should be able to answer such questions as: how closely can tetrahelix A conform to curve S? How many tetrahedra are necessary to obtain a given curvature? Can we conveniently use common mathematical techniques, such as splines, to perform this mapping? How can we best characterize a twisting of the faces along a curve (without deviating from the path of the curve)? Can we propagate a wave into a tetrahelix to produce theoretical motion, such as swimming, or crawling assuming anisotropic friction?
Develop Mathematics of Octet Truss Curvature
The Tetrahelix is a length-regular geometric structure that extends in a line indefinitely. There is a also a length-regular planar structure. That is, there is a way to build a planar omni-triangulated geometric structure in which all lengths are exactly the same, known as the octet-truss, patented by Buckminster Fuller in 1961. Write a technical paper and a graphical computer simulation (analogous to Read’s paper “Untwisting the Tetrahelix”) that describe out to form barrel vaults, cylinders, cones, and other shapes from the Octet Truss. In reality, this is mostly high-school geometry — -but very advanced high-school geometry requiring careful computer programming and/or the use of Mathematica.
Develop a Soft, Pneumatically Driven Version of the Glussbot
Soft robots are all the rage at present. The Glussbot concept is in principle supportive of soft robotics, but the current implementation using linear actuators is not. Using the tetrahelix as a model, build the simplest possible (3 tetrahedron) soft glussbot wherein motion and force are driven by off-board pneumatics. Do not attempt to put the switches and control apparaturs in the robot. Possible means of doing this include 3-D printing in flexible material carefull treated with a solvent to eliminate air holes.
Very Challenging Projects
These projects are probably master’s thesis or PhD-level projects.
Silly Putty/Clay Sensing
Figure out a way, using perhaps cameras or sonic sensors using time-of-flight and/or tomography of a lump of clay. The idea is to be able to hand a lump of silly putty to someone and accurate sense the shape into which they mold it, for the eventual purpose of modeling that shape with the glussbot.
Given an arbitrary 3-dimensional shape, perhaps defined as a closed surface in three-space via a mesh, compute an approximation to that shape that can be obtained by an arbitrary omni-triangulated geometry with limited length changes in the members (in other words, a mathematical model of the geometry of a glussbot.)
Miniaturize the Glussbot
Figure out a way to build a Glussbot at a millimeter-scale actuator length. Possibly Nitinol memory wires are a means to do this.