notes-computer-robotics-roboticsNotes

notes from https://news.ycombinator.com/item?id=6904793 :

drone 9 hours ago

link parent flag
 I've found the real cost factor is not so much the robotics platforms when you want to make big+specialized things like this (because the robotics platforms specifically designed for one's particular problem space are often few or non-existent) but the hurdle of tool acquisition. That is to say, most people I know working on these sorts of things as hobbies build as much as they buy. A decent small metal lathe runs about $600-1000, a mill drill can be put together for around $1k, CNC router capable of cutting decent size panels of aluminum for around $2k+.

You can buy inexpensive robot platforms, but they're small - there's a reason for this. Heavy, powerful components cost real money and someone has to make margin selling them to you, given they paid margin to buy them. High power and high current control systems (motor drivers, motors, etc.) are more difficult and expensive to design than small, low-power units.

But, there's nothing stopping you from seeing a design you like, and making it from the raw parts, except a lack of tools (in most cases).

> a simple little robot platform moving around my house sensing tape on the floor. Even that will cost me $1000

Huh? You can buy line-following robot kits on Amazon for $30, and there are even cheaper ones out there.

> When can I build a robot that will perform real, useful automation that will improve my life, even in some marginal way, not just seem like a cool toy? How can I build a robot that will fetch me a beer or let the dog out or get the mail or something even minor that will make me feel like I am doing something with purpose and a real goal and not a homework assignment to get a robot to perform some meaningless task.

Right now, buy some motors, buy some metal stock, bearings, shaft, pulleys, etc. and start designing a machine around them. Long before you get to the algorithms of running a robot, there are the physics of it, the basic electronics knowledge, and the design and machining.

> What are some robotics projects that I can build and learn from while and actually use and enjoy, today, relatively cheaply?

Small robotic arms are cheap these days - OWI and Lynxmotion make fairly inexpensive arms for learning, small wheeled bases like the Dagu Wild Thumper can carry a small arm for around $250, and are well regarded. You can throw an Arduino or Raspberry Pi in for control put together a Lynxmotion arm and a Dagu base, and for less than $600, you have a platform that can lift small objects and move them around. When you want bigger, make it yourself from what you've learned.

mikepurvis 7 hours ago

link

> Right now, buy some motors, buy some metal stock, bearings, shaft, pulleys, etc. and start designing a machine around them. Long before you get to the algorithms of running a robot, there are the physics of it, the basic electronics knowledge, and the design and machining.

This is very much a need. The Turtlebot is a low-cost ($1300) way to get a lot of functionality out of the box, though the lack of manipulator does limit what you can accomplish.

If you (or your lab/company) does have a five-figure budget for hardware, there are a number of very attractive off-the-shelf options which exist to prototype on: http://wiki.ros.org/Robots/#Portals

(Disclosure: I work for Clearpath Robotics, which design and manufacture several of the machines on that page.)

reply

You might be surprised at the software components which do actually exist:

http://moveit.ros.org/

Check out the videos on that page. There are a lot of obstacles to inexpensive robotic manipulation; the main ones are power-weight ratio and power-cost ratio of the actuators.

robotresearcher 17 hours ago

link parent flag

ROS and Gazebo are Open Source, are used by professional researchers, and are all you need to make a real contribution. You can't afford a robot, but you can simulate a very good one for free.

http://www.osrfoundation.org/

esonderegger 8 hours ago

link parent flag

I highly recommend watching a talk given by Erik Nieves from Motoman at the most recent RosCon? because he talks about the state of the robotics industry and why this "gap" exists in a very accessible way.

http://vimeo.com/67044628

Two reasons why I think the gap between getting started and being able to build something useful exists is that robot parts are still much more expensive than we expect them to be and the fundamentals of robot movement are much more difficult to learn than we expect them to be. Unfortunately, I think those two factors discourage a lot of engineers early on and is why the size of this community is so small.

I feel very fortunate that I have a Kuka YouBot? to hack on. The company I'm trying to get started with two friends wrote a proposal that got accepted into the sponsored track of a competition Kuka is running. (http://www.kuka-labs.com/en/network/innovationaward/) After setting it up, it's clear why this is a 20,000€ robot. But before you can actually DO anything useful with it, you have to be able to understand kinematics and dynamics as well as how to get path planners to talk to controllers. And then the really powerful stuff becomes possible once you incorporate computer vision, which is also not an easy topic to understand. It not surprising that most of this work is being done at universities by grad students.

The state of ROS today feels a lot like what Linux felt like to me back in the late 90's. There are lots of issues with dependencies and programs not compiling correctly. Documentation is severely lacking. But there's lots of enthusiasm from very friendly people to help each other work to make things better.

What makes me worry is that many of the advances in ROS came from Willow Garage and Google is buying up many of the companies that spun out of Willow whereas a lot was made possible in computer vision due to the sensors made by PrimeSense?, which was just acquired by Apple. I hope the Google vs Apple rivalry doesn't slow the pace of progress in open source robotics that was making the dawn of non-industrial robotics seem not so far away.

tlarkworthy 11 hours ago

link parent flag

yeah proper robots is absurdly expensive. But its also multi-faceted. So if your interest is in machine vision, you can do alot of experiments with a web cam (which are really good now!)

I built an optical feedback robot arm for under $100 [1]. At work I use a robot where an individual sensor costs $100,000

The algorithms are much the same though. Doing control on cheap hardware is more challenging, so if you can do something on cheap hardware, you really have done something noteworthy! Expensive hardware pays for reduction of noise, it treats the symptoms of reality but does not solve them.

To truly solve the robotics issue of environmental noise, you have to solve it in software which is really hard but can be investigated cheap on cheap hardware. Robot projects fail because of software, not hardware normally. You should always build the software stack before spending lots of money on hardware.

[1] http://edinburghhacklab.com/2013/05/robotics-adaptive-contro...

Here are some cheap ideas:

1. Improve the robot arm control I outlined above. There is a huge market for cheap robot arms you could scale the ideas prototyped in [1] upto.

2. Active face recognition, put a camera with a zoom on a pan and tilt. Try to zoom into peoples faces for better recognition. Animals exploit shitty eye hardware with active attention in a way our current robots don't exploit.

3. voice activated inventory system for electronic components. Every hackerspace can't find the components they own. Popout a component tray via a push gantry behind it.

4. 3D hardware localization system. Actually determining the position of something is hard. Its the building block for gather training sets to do more challenging problems

dangrossman 13 hours ago

link parent flag

You might look at the Neato robotic vacuums. Unlike Roomba's random wandering, the Neato bots have 360-degree LIDAR on top and a wall-finder on the side to do SLAM. They also have a USB port on the side that supposedly provides raw data from the sensors as well as direct motor control. They get a good 1-1.5 hours of run-time per charge with the vacuum turned on, probably longer with it turned off. All that robot for under $200 if you watch for sales.