General Misc > Misc

Revolutionizing Robot Building

(1/2) > >>

Collaborative robot building

My suggestion for revolutionizing robot building is to tap on the knowledge of the web to build a robot that is comparative to the commercial robots; // the idea of collaborative robot building is taken from this website, it currently produces multiple research papers from observation of the public
Arduino , Linux , Human Genome Project are prefect examples to prove that collaboration is better than going alone;

I believe that generalizing (multi-purpose) robotics platform will be the direction to the future;
Key properties of the Multipurpose Robotics platform (MRP) = self assembly + self sufficient (Taps into renewable energy / voltage source) + self navigation;

The combination of the 2 ideas that i had just proposed may have huge implications;
I have the general idea in mind and i love to share my ideas, if anybody have any suggestion/ comments about the ideas proposed please share and let us collaborate together;
May it be the first Networked Robot to be design and created by all;

Few days ago I was searching around the web for information because I was trying to understand the implications of building a small humanoid robot capable of executing small tasks like self balancing, walking, executing commands or avoiding objects while walking. I found out it's not easy but indeed very complex because you really need a good engineer mind and something like a humanoid robot would involve a big investment ($$$).

I like your idea because I'm thinking the same. I think that the key to building advanced robots like commercial ones is to use the existing resources from the web but in a collaborative way.

Wasn't linux created in the same fashion ? Programmers world wide joined togheter to develop programs, distros and so on.

A networked robot that would be great !
I have so many ideas :)

There are cheap alternative to humanoid robot


I had purchased an humanoid robot (with 996R servos) which cost less than USD 300, shipping fees included.
It worked just fine, i use Arduino Mega to control the 17 servos and up till now i only manage to make it stand.
Next step i am going to add in the 2 mechanical grippers (USD 15) to achieve the some self assembly capability. To achieve full self assembly capability the humanoid arms must be longer and there must be a rotary tool attached to the robot.
Main purpose of rotary tool with changeable tips is to cut grind, screw and many more.
To recognize each tips will need a very strong computer vision program.
The camera module chosen is ov7670. (USD 10)

As for software i had decided to use matlab to control the robot, matlab have very powerful software codes that can tackle any problems but the learning curve is quite steep.
Computer vision can be achieved relatively easy with matlab.

Links      //use google translate to read (Humanoid robot)                         // Mechanical grippers                                          // Rotary tools (USD 2203)                     // Commercial humanoid robot with no self assembly capability

To initiate collaborative robot build i would need guidance/ questions/ problems from anybody so please try to reply and share your ideas no matter how ridiculous it is. I would write it down in the word format and summarize what i had gathered through the forums and share with all of you.

I am going to kick start with my ideas.

Key properties of the Multipurpose Robotics platform (MRP) = self assembly + self sufficient + self navigation

Try to visualize each section as an 3D modular electronics block.
- Display (TFT)
- Power supply ( Solar + Lipo battry + charging circuit + battery level indicator)
- Communication ( Bluetooth + Xbee or NRF24L01 )
- Navigation ( Camera + ultrasonic sensors + lasers + human detectors + IMU + barometric +  photo resistor +IR + sound sensor)
- Interfacing ( Servos connectors, Motor connectors , External sensor connectors, magnets, current sensor)
- CPU (Arduino or STM32F4)
- Motor (L293D motor driver + motor + encoders)

Need not implement all of the sensor, but think of it as interchangeable modules that allows the support for all sensors.
All of the 3D modular block can be removed except for the CPU & power supply , else it will still work.
The problem now is the Interfacing circuits to the various platform the MRP will work with.
There are 3 platform this MRP is planned to work with, Quadcopter, Humaniod and a Truck.
Feel free to suggest more platform , but more platform means more problems to the interfacing circuit design.

Why not Raspberry PI(25$). It's a 700Mhz ARM CPU but I don't think it's powerful enough for computer vision. It handles well blob detection with SimpleCV and python using a cheap(10$) webcam.

Matlab is more powerful but you have to run on a dedicated host computer with wireless(either wi-fi or bluetooth) connection between MRP and the host. Is it possible for the software to run on board the MRP ?


[0] Message Index

[#] Next page

Go to full version