hmartiro / project-thesis Goto Github PK
View Code? Open in Web Editor NEWxJüs, the hexapodal robot with a passive-backbone to improve behavior over harsh terrain.
xJüs, the hexapodal robot with a passive-backbone to improve behavior over harsh terrain.
By proposal date, we should have some specific options we could put together in a budget
Use this page: Motors
There are lots of advantages to getting the EduBot running.
Waiting on Hal to give us code and CAD files, as well as the Ubuntu login/password.
Such that the units all match up, everything is realistic, and actual numerical testing can be done
Ask them about motor/gearbox/encoder package
Tell them that we want a position and speed controller that is light and see if they have any ideas
This is still what needs to be done to finish up the hardware portion of the test cart
Hayk & Chris, Be ready to talk about pros and cons of different controllers/sensors we will meet at 3pm the day of the meeting to be on the same page.
Brendan, Greg & Tom, Be ready to talk about general mechanical design, (step vs normal) normal motors, materials (strength : weight)
Goal of meeting:
(1) Have some idea of cost associated with project. This will be critical for Oct 9th deadline
(2) See if Hal has any suggestions for adjustment to basic design, also, does he warn us against anything we are thinking of trying?
(3) How do we get our robot working?
Justin Siepel from Berkeley just pointed us to the Canid, a quadruped with an actively actuated spine that Kodlab is working on. Along with it, there is a paper that has a ton of information, not just about the Canid but about all the RHex designs they have built.
http://kodlab.seas.upenn.edu/uploads/Main/spie12.pdf
It includes tables that compare all the different types of RHex designs they made, including dimensions, weights, motors, batteries, mechanical and electrical info, and exact parts they used. I think everyone should read this before we go any further with designs. This is the best guide we have so far by a lot.
Goal is to get ubuntu 12.04 on the 32GB sd card.
Learn and utilize finite element analysis in Creo.
Model bending and twisting on joint, check stresses and deflections
Here is my current response to Hal's comments. I'm gonna take a look at the papers before I send, but any comments or edits?
Hi Hal,
Thanks a lot for your comments - it's great to have your help. Reading through the remarks, we agree with your take but have two questions for you:
You recommend going for a quadruped design because it would be cheaper and simpler. However, we're concerned that the quadruped design would be a higher risk for us due to the decreased stability. I can't find any example of a quadruped using RHex-type locomotion, and Pei-Chun Lin's design, which seems to be this one looks like it has more complicated mechanisms with prismatic joints. To accomplish that seems like a harder feat than dealing with the added complexity of two spinal joints and six legs, especially since we are much stronger at design and manufacturing than at dynamics/control. We chose the RHex because it was relatively simple to control for a legged robot and it seems to us the four leg design would add difficulties that we may be less prepared to handle, versus the marginal cost of building another segment, which we think is much more doable. What do you think about this concern?
The other question we have for you is what part of this design project do you think will be the most difficult and time consuming? Is it the design, the mechanical construction, the electronics, or the software and control? It seems that you have some reservations about the feasibility of this project, so we're trying to get a picture of what the risks are. Since you've worked on the RHex, you probably have the best idea of this. From our current perspective, the design and mechanical construction seem within our reach in terms of time and budget. The implementation of controllers for the legs and the RHex gaits is more foreign to us, but we're hoping it is feasible, mostly because of the simplicity of actuation and abundance of papers on the topic and help from you and Prof. Holmes. What should we be worried about and budget extra time for?
We have until Friday morning to determine the goal of our project, time to legitimately brainstorm.
My idea:
Add a storage compartment to the body and maximize the robot's load carrying capability, perhaps focusing on its load to weight ratio.
Application: (Military) Carrying small but essential supplies such as ammo, medicine, or food between positions in a hostile environment. Reduces risk of human life.
What's new: Requires a new body design capable of storing or attaching a reasonably sized object. Have to deal with the issue of minimizing the body weight while still ensuring the legs/motors and structure are capable of carrying an additional load. Varying loads will probably also affect the control system. Boston Dynamic's Big Dog is already designed to carry supplies, but this is not meant for battlefield conditions for obvious, lumbering, reasons.
How do we provide high-level feedback for steering and navigation? Looked at some papers which I uploaded to the research section, but since Holmes wrote some of them its probably easier to talk to him. Incorporate into simulation to follow a target.
An account has been created for the group to use, with Jo-Ann's credit card
Username: Mugatu
Password: xjus2012
I think this is basically the founding paper on the Rhex, from Kodlab:
http://kodlab.seas.upenn.edu/Kod/IJRR2001
Kodlab released this as research in 2001, and Boston Dynamics got hold of it and commercialized it in 2003.
It's a really useful paper and very understandable (even for ELE pleebs), so I think all of us should read it to get some background on the motivation behind Rhex, the pros/cons, and the dynamics involved, so we don't sound like retards when we talk about it.
Torque and size requirements
Where to purchase them
How to attach them with ability to replace them
size up everything, get an idea of the physical layout, etc
Place all on-board components
Determine dimensions of all 3 segments based off of on-board components.
Discuss making middle section longer than front and back, in order to increase control over pitch stiffness
Meet with adviser this Monday to discuss preliminary system
While the CNC machine is free we should begin to machine the parts for the test cart. Many of these parts will be useable on the final robot, assuming that we don't discover any issues in our model. I'll start by CNCing the motor mounts.
linux drivers, etc. we need the beagleboard connected.
-Determine material with appropriate properties
-Find source
-Get quote
quite annoying, needs to go away before videos are made
Confirm material/shear modulus to be used.
Check stresses and deflectioin etc.
Update CAD model accordingly
Arduino, PIC, psoc, gumstix, raspberry pi
Have a good choice by the proposal date
Use this page: Microcontrollers
I think we need to be using the issue tracker and wiki a lot more to stay on top of shit. Honestly we've been getting real sloppy in the last couple of weeks, and we need to do better.
Can everyone just take a few minutes to go to the issues page and open some issues for yourself based on what you're doing? They should be small and specific things, like stuff that would take a few hours max. "Make base plate for test cart" would be a good one, but "Finalize CAD model for robot" would not be.
Also, honestly just make a page and write stuff on the wiki about your progress. Think about how much easier the paper will be if we have some shit about the legs, the cover, the joint, the batteries, the computer, and whatever else we do. Remember, we don't have a girl in the group to write the report..
Why is it coming up for the interpolation mode?
Looks like the best choice is the BeagleBoard XM, based on the research here
We should probably go ahead and purchase this. @boleary134 do you know how we purchase things with Jo Ann's card?
By the end of this wknd
Talk to relevant professors this week and get it figured out.....I just got approval from Andrew Houck to be my adviser. ( http://www.princeton.edu/ee/people/display_person/?netid=aahouck )
Need to fill out SEAS form, make changes based on Hal's comments, and then turn them in
Tell him about our project and ask for
based on the motor controller documents
I've got a simulation of our robot w/ a flexible design via torsional springs, which can be made rigid. I want to put out a set of videos on Youtube that show the effects of each degree of freedom vs the rigid version on a simulated obstacle course / natural rough terrain.
See here
Our plan is to control the robot through remote desktop or VNC
Begun writing goals and design requirements, then solution
I tried to commit a large file ~400MB and it broke git for the beagleboard, now it runs out of memory while pulling. Need to somehow erase the file from its history.
Start with Greg's outline here and from what Nosenchuck told us today
I think 1-2 people should write a draft on that wiki page and then others should edit it. Who wants to start?
dsfdsf
I am putting the finishing touches on the test cart CAD model.
Hello
Analog line 1 and 4 need to be manually pulled from the cable and attached to (+) and (-)
to apply smoothing for the damping
Important Dates:
October 9 - Funding Proposal Due - Needs to have advisor's signature.
Oct 15 - SEAS Funding Due ( ~$600 max)
October 22 - Marshall and McKinzie fund proposal
November 12 - Supplemental fund proposal
Info on this page: Funding
Will look into Keller Center to see if the cough up some dough.
Leg construction consists of 4 items:
*Will a 24V switch slower than a 12V powered motor via PWM
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.