CE Project: CE Bot

[xheavenlyx]

This is one of the few ongoing Crazy Engineers projects.

Project Goal (General):

Design and develop a medium sized robot with the given specifications.

Project Objectives (Listed in Priority + Timewise):

1. Discuss the specifications of the bot, physical size and its use. What is the bot for? What functions will it perform? The Why and What. TIME: 1 week.

2. Draft a rough model. List what all modules would you like to see on the bot? Processor? Motors? Motor drive? Power source? Webcam? Ethernet? Parallel com, Serial com, USB com, etc TIME: 1 week

3. Put the modules together, each module must have specific Input Output relation with the main on board processor. TIME 2 weeks.


We will first finish with these steps and then follow through. Anyone who has an idea can take part but remember the format!!

Give your idea, then support it with good reasoning, then point to any sites and/or documentations to further support it. At the end of the 2nd objective, I would like to see the serious contributers make a powerpoint presentation (you can make it on Google Documents) and upload it here or on google docs.

I will add more when I see enthusiastic response to this project.

xhx, hope you don't mind me adding some stuff here. -ash
Please check http://www.celab.tk to view the progress

Current contributors: xheavenlyx (project leader), ash, bayazidahmed, gohm, sanih, raj87verma88, joe3204

[xheavenlyx]

So this is our first step:

Project Objectives (Listed in Priority + Timewise):

1. Discuss the specifications of the bot, physical size and its use. What is the bot for? What functions will it perform? The Why and What. TIME: 1 week.

Must finish in 1 week. Anyone interested can add for now.

[ash]

Great work, man. However, maybe we should extend the deadline a little 1 week is a little to soon given the relatively low activity here. How about 2?

I've got number of different ideas. But since we MIGHT have some interest with the Google Lunar X challenge, here's my initial draft.

1.
The bot will be a rover-like vehicle, able to traverse over rough surfaces to get from point A to B. It'll be able to detect objects with the use of image-processing and other proximity sensors. Cameras will be used to record data and wirelessly transmit data to a computer.

On the computer, the data can be processed to create some kinda terrain map. 3D perhaps?

The bot itself can be autonomous or remote control. If autonomous, the bot detects some beacon at point B and goes to it while avoiding obstacles (obstructions, holes, traps?) and taking data.

The size is relatively small, maybe the footprint would not be larger than a standard magazine.

So, with that, we have all sorts of engineering disciplines involved. I'll update it once I get more ideas.

[xheavenlyx]

That is an excellent idea, specially about it transferring the video data to an off board computer for it to be processed into a 3D terrain map.

I was thinking, we must not concentrate on the moon rover. This other team which will be joining soon will be bringing their own design and they just want us to help. Anyway, moon rover is no easy task, we can talk as much as we like but the reality is very different. I am pissed off too after realizing this...like take off, landing and survival in Xtreme temp -173`C to +100`C! Making a rover is a cake walk compared to the other tasks.

Anyway, we can make a small rover which can move into small places and need not have a plain ground, it can also "traverse" into a hole vertically. Why this? Firefighters and rescue teams can use it many cases. In India itself we have many incidents of ppl falling into holes, or building collapse, or floods etc.


I agree with the size too. Has to be small.

Proposed Ideas:

1. Time for each task is 2 weeks now.
2. Possible components under Sensors: Camera, Wireless transmitter (Ethernet, not bluetooth- need range), Temporary storage.

3. What do you think about most of the process being taken care of an offboard computer? Should it be manual or auto or semi-auto controll?

[bayazidahmed]

Yup you are right. Moon rover is not an easy job but not impossible.

It would be a hard time for me to decide what to do with this robot, but if others have a definite plan I can help with tasks 2 and 3.

Compnents:
1. Camera: Necessary for self navigation of the robot
2. Serial/Parallel or USB port - for connecting to the robot
3. Wireless: Bluetooth has a max range of 100m (Class I) and uses less power (the biggest advantage). Other possible option is WI-Fi.
4. Processor: A good microcontroller will take car of all the jobs. I could be a very good help while programming this thing.
5. Temporary Storage: It oculd be anything, from a small memory card (SD/MMC) to a complete hard disk. Nowadays storage is pretty cheap, so we dont have to worry much about the cost factor. When connected vai the ports or wireless, we should be able to acces the temporary data stored here like log files, raw data that is being captured thorught its mic or camera, etc..
6. Arms: Ateast two arms so that it can pick up things or move objects from its way.
....
.....
......
and let the list grow..


In btw, why the heck there is very little activity in CE?

[ash]

Great contribution bro!

Yeah, we are not really working on an actual Lunar Bot that will be sent to the moon, but a normal rover that does similar functions (ie navigate over a rough surface, collect data, transmit data etc). xheavenlyx, thats a good point about using it for rescue purposes. If we plan to go vertical, we might look at the use of legs instead of wheels (like a spider). Reminds me of the topic we discussed here:
Rescue the boy from the borewell - better way?

Low power consumption
We'll try keeping most of the processing offline, so that the bot does not consume much power. That would also include using a wireless transmitter with a good range to power ratio. Low power = small battery supply = reduced weight and size. But remember guys, the transmission channel may not always be perfect! You must consider the fact there is interference, and that the bot is not stationary.

If we plan to use microcontroller, we must design the circuitry as efficient as possible. Now we need a balance between cost/efficiency like proper voltage regulators, etc.

For storage, we can use the a flash storage like bayazid suggested. It'll probably act to store robot instructions as well as a buffer for data (since we are transmitting it off later). However, if you want to change the bot's role to collect all data first, then we may need a bigger storage solution.

We need a rough sketch of the bot. Very hard to visualize with just words. Later, mechanical/mechatronic CEans can draft them on CAD software

In btw, why the heck there is very little activity in CE?

We need a critical mass of active CEans, I suppose. However, Big K is putting a lot of effort to get more members and encourage participation.. look out for that!

[xheavenlyx]

Good Ideas! So now we have these components as @bayazid and @ash have pointed:

Compnents:
1. Camera: We can see later what kind of a camera, and how it will be connected.

2. Serial/Parallel or USB port: We can keep all if he make one concideration, see below.

3. Wireless: Bluetooth or Wi-Fi. But since we need more range we can have Class 1 BT or a Wi-Fi, that will depend on the processor.

4. Processor: IMPORTANT CHANGE. See below.

5. Temporary Storage: A hard drive can be used, as a temp storage and as a buffer depending on the time and need of the data! It can also be used as a log when something goes wrong in a rescue operation.

6. Arms: Very good suggestion! We can have one or two. Lets see how the design proceeds.

4. Processor: Now this is an important factor. I was thinking, why not use an old laptop, which has all the important things! Parallel port x1, Serial x1, USB x2, A hard drive, an expansion slot for wireless or bluetooth (swappable) and a custom linux backed firmware for operations. All programming can be done in C or Python. Can have more AI then what a normal microcontroller can support. We can remove the screen for power saving or keep it for status of machine, testing etc etc. I have an old acer with me and never found a use for it

7. Motion mechanism: How about something of the likes of this, but not the same:

http://therawfeed.com/pix/bomb_robot.jpg
http://www.nato.int/pictures/2003/031007/b031007ab.jpg

Ash, programming the legs can take up a lot of processing power! Since even in a vertical drop, a legged robot cannot be more productive then a wheeled, because anyway we will be hoisting it down with a rope! If we design a legged, then we have to concentrate on the mechanism and programming. We can have another project in the future with a snake like moving ability or legged, even I want to do something different than wheels, but for this one we can have a tank type drive. What do you think?

And as for the outer design and chassis, we will do that after we finalize the components

[bayazidahmed]

Hi CEans! Call me ahmed.

Coming straight to the point.
Laptop?? Are you sure? Do you know how much your acer laptop weighs?
Even if we keep the issue of weight aside, there are some other problems with using a laptop.
1. The major problem is with ports: Your Laptop has only one parallel port. Parallel ports are a great help in designing things. For instance, take the case of wheels. For wheels we will be using stepper motors, stepper motors require 4 pulses. So two stepper motors for the two sides, that amounts upto 8 pulses or 8 pins from the parallel port. Now, the parallel port on the laptop has only 8 i/p/o/p pins! How will we configure other things like the arm and other components that will be attached to it? Camera can go to the USB. I think we need atleast 32 bits to program this beast.
2. Weight
3. Size: Its quite bulky
4. Customization: another biggest disadvantage, you cant customize it to fit exactly your needs. All ou can play with is programming, and that take half the fun out.
.....
......

Having said all this, I am not completely disagreeing with the suggestion of xheavenlyx using a laptop. There are many advantages to it as well.
1. You have everything assembled! All you have to take care of it is programming.
2. Programming is much easier. With a microcontroller you wil have to go for ASM though there are some compilers available fro microcontrolles hat convert your C program to ASM.
3. Communication part becomes easier with the LAN/bluetooth/Wi-Fi of the laptop. With the microcontroller we will have to go for a serial port (I call this an advantage, but I am in favour of a serial port, it looks professional )
4. Crazy: We are turning a laptop into a robot, and this falls into the category of crazyness!
.....
......

[ash]

I'm down with fever, so I can't think of any new ideas to contribute laptop sounds good, but remember we are doing most of the data processing offboard.

In terms of weight, we can just use the motherboard.. no need for the laptop chasis. You'll just have one big ass PIC board then. What else.. um, as ahmed said about the parallel port, I guess we need to multiplex it We can also use other I/Os; the mic input/sound output can be used as a ultrasonic detector, with modifications of course. We can safely rid the screen entirely, as long as you have a VGA output!

I like the tank track idea. Looks cooler We can still use a microcontroller, I guess. Connect it to a serial or USB, hehe.

However, im still an advocate of using low power. What processor does the Acer laptop have, xhx?

[xheavenlyx]

Wow! Both of you have raised amazing points!!

Ahmed, thats a good point. Here are a few things I would like to clear:

1. Yes, ports can be a problem (Refresh: 1 PP, 1 Serial rs232, 2 USB), but I wanted to include something very vital in the project. Similar to the lines of what ash pointed out. I will talk about it in the end.

2. You are right about the weight, but we can try our best to strip it down. It uses a P1 processor. 128 MB RAM. Still have to see power consumption. As ash said, we can keep the VGA-outpt and remove the screen. I also think, for some field jobs we need good traction, if its okok light then it can slip off surfaces and be a bit ... flimsy. Dont know, lets see. From robot competetions I have seen on TV. A 200 lb (about 90+kg) can run good on affordable batteries.

3. Above

4. Customization: Another important factor. But what does the laptop replace? Only the mC. Still we have the motors, sensors and the power management to handle. Well, we have the most flexibility with a powerful processor with "INFINITE" inputs This is the fun part, the processor and the inputs! See next post on Processor and Input/Output.

[xheavenlyx]

Firstly here is the links for the ACER Extensa 500 laptop.

Extensa 500 On-line Help: Main Frame

• The Processor is Intel Pentium® processor with MMX technology.

• 166, 200, 233, 266 MHz

• MMX technology provides 57 new instruction sets to improve processor performance in traditional digital signal processor [DSP] applications, including the graphics, audio and voice processing capabilities now emerging as value-added features in high-performance embedded products. MMX technology can potentially eliminate the requirement for DSP chips in embedded applications such as video kiosks, telecommunications devices and POS terminals.

Why I support using this processor instead of a PIC? Well, we have huge potential to expand the on-board functionality! I mean, we are doing so many tasks, in a controlled manner, and on the way helping in the manual control even if its not fully automated! Like, motor drives, 1 (or 2) arm(s), camera, IR, sonar (ultrasound), temperature, and optimizing the power usage.

Ok, now the problems of I/O.

Two solutions exist, in my head, you can add in too!

1. Using multiplexers. This can get a bit complicated (havent thought on this one) but is possible to get 32 or 64 bits from 8 bit I/O. One disadvantage is that, the mux are non-intelligent devices. How? :

2. Using simple PIC mCs connected to the PP and Serial: Think of a multinational company. We have the board members (main processor), the managing directors of each division (PIC's connected to the Proc.), the workers (modules, sensors connected to PIC's). This way we can have more ADC's (Analog to Digital Converters), more Inputs and more outputs. PIC's have interupts to stop a processes in emergencies. They can either do their own calculations and take an action or simply report to the processor.

Software wise we can have C functions specifically written to program the PIC's whenever needed through the Serial port!

Now the tough part is sitting down and designing the protocol and methods. I am willing to help on that.

What do you guys think? Of this looooong and tiring post

***
*EDIT*
***

Oh, yea, and another method is using a network of PIC's connected to each other and forming an intelligent network. Each one can take own modules to control, using their own head, or talking to each other to know the status and take actions.

[bayazidahmed]

Multiplexers is not an option

Connecting teo PICs, one to PP and one to SP will solve the problem but make the complete thing quite bulky. One Pic can control the basic functions of the robot like wheels, arms, cameras, etc... and the other can be used to mointor the various sensors and for any additional functions apart from the basic ones.

A network of PICs is a very good idea, it will be a bit complicated. Lets not concentrate on this for the time being and keep this feature for version 2.0.

[gohm]

I was thinking treads for locomotion as well. Oversized so it can still run if it gets flipped over.

What about a gas detector? Can detect the threat of carbon monoxide, natural gas leaks, etc. that could be a risk to people entering area afterwards (scientists, rescue workers). The vehicle itself would need to be spark proof.

temp. probe or camera using IR.

A bulldozer type front blade to push objects out of the way to the side? That way you only need one grab arm. Arms would be more likely to break in a fall or tight places.

Do you want a speaker/mic/sound card for basic speech communication?

What degree of water resistance would this require? none, resistant or proof? I would vote for resistant.