Search:

Home Page


News

Hypercube Open Source Boards

Hypercube Software

Hypercube Cloud & Data Center

Hypercube Training

Hypercube Geometry

Open Nano

Parametric Drones

Hyper Neural Net


Fiber Laser Metal Cutting Machines £34,999+

Hyper Car


PCB Design

3D Printed Parts Design

Virus Scan
Penetration Testing



Industry 4.0

Technological Singularity

Digital Quantum Computer

Embedded C Training

Sharjah Tour


Employment
Six Transistor Cache
Investors
Launch Date

Contacts


Twitter
Twitter



Terms and Conditions









AI Synthetic Hand
The rational behind a Synthetic AI Hand Project is straight forward. The entire human race civilization is built by a pair of hands duplicated over and over again doing the little things that collectively becomes civilization. So if we could make a pair of really good AI infused synthetic hands, only one pair of hands would ever get built, and the rest are all just duplicates.
  • Steps:
1. The infrastructure for Synthetic AI Hand Project begins with Hypercube Tesseract PCBs that make all the relevant peripherals connected to the CPUs that drive the actuators.
Several of the boards needed for the actuators is designed already ( Open Boards ).
2. A second piece of the infrastructure is the software for designing the AI hand 3D printed parts which is scripted and written in Python running under Linux. The purpose of printed parts is to create complex wax mold copies from which the final parts are cast. The wax is dissolved by applying heat and recycled.
Three 3D printers as above have been acquired and a large one capable of 400mm x 400mmm x 500mm prints with dual head. The smaller printer prone to jamming because the linear bearings are impossible to assemble correctly aligned due to a design mistake. An option is to design and print replacement parts for the linear bearings before trying to assemble the remaining printers. The large printer arrived with borosilicate glass plate broken. Local replacement supplier found that can make the 400mm x 400mm x 3mm thick borosilicate glass. The dual head printer was chosen because one head can print ABS while the other prints PVA, a water soluble support material. The support material is dissolved away with water speeding up the 3D printing process which would otherwise require a lot of time to remove support material. The software running on the PC is Repetier running under Linux which connects through USB to the controller board on the 3D printer. The control board is an Arduino. The power supply that comes with the printer was determined to be weak, so a bench power supply capable of sourcing ample current at 15V was used.
Sample parametric chucks printed on 3D printer. These were specially developed in OpenSCAD to make custom machines easy to fabricate with custom profiles and chucks. The opening to the chuck can be customized to any size round shape or a polygon shape such as those on a hex bit, or any profile such as an L shape. The chuck uses an angled helix screw thread so that the angle of the screw does not line up with print layers which makes the screw and nut assembly extremely tough to pull apart. These 3D printed PLA nuts can be tightened with a spanner :) - may be we try to print some olives as well to see if they can be used for cold tap water copper pipes at 3 bar pressure (hot water will soften the PLA and fail under pressure). (Ordinary threads may peel away when the nut is tightened if print layer is parallel to the thread.) We can also now do custom angled gears with a chuck lock to lock the gear wheel to any shaft. It is an interesting way to lock gear wheels to shafts and prototype new mechanisms rapidly if the mechanism is too awkward to print.
3. A third piece of the infrastructure is the test bed that takes feedback from different hand designs and alter the scripts to make incremental improvements with each iteration. The test bed is driven by Linux programs written in Gambas with a GUI interface to make it function easily.
4. A fourth piece of the infrastructure is to make the test bed portable by making it work on ARM SoC boards like the Pi so that it can be fitted to Androids that can transport it to places of testing and work, so it may be tested in situ under real work conditions.
5. We intend to use AI infused Hypercube Neural Net to control the hand.
Login
Page last modified on May 04, 2018, at 07:24 PM