Arty the robot - Design Notes 10/19/10

I had built a robot (Mikey) 15 years ago from spare parts I had and using the 8052 BASIC chip. I built the circuits from scratch using wirewarp and TTL chips. The robot had 16K RAM and 16K ROM and used an 8255 for I/O ports and had differential motor design. Click here for info that that design.

Now, in 2010, I decided to build a more modern robot using current technology and parts as the world of robotics has gone through tremendous change in the last 15 years! These notes go over the design thoughts and ideas of this new robot and hopefully serves as a guideline or better yet, inspiration for someone else wanting to do the same. And let me say right off the top, if you are looking for Star Wars quality robots or craftsmanship that looks like a work of art, you are not going to find it here. But if you want to see what an average person can build using average parts and time, continue on. Actually, building a robot today I feel like I am cheating a little. Back in the 80's and 90's, you had to design and build a robot from scratch. You had to design the circuit board and CPU, you had to wire wrap or make a PCB, you had to write your own code. If you want to say, add a voice, you had to find a voice chip and wire it up yourself, etc. Now instead you buy an already wired microcontroller and carriers cards that have sensors and voice chips and whatever already made. You download for free software libraries and code off the net. There is still the challenge of putting it all together but I feel the emphasis has shifted now from hardware design to software programming. Where once you would have spent a good deal of time just getting the hardware to work, now you spend most of your time programming and debugging your C code.
 

Design Objectives -

a. Liking the differential motor design, I wanted the new robot to have two DC motors which of course would be used to steer and move the robot.
b. I wanted it small enough to use medium size motors but big enough to carry a good size battery (won't need to be charged so often) and big enough to expand later on.
c. run on a 12v battery as I already had a beautiful battery laying around
d. run a long time on a charge
e. use a sealed lead acid rechargeable battery
f. be programmable
g. cheap
h. use whatever parts and junk I had around the house
i. I wanted it to do various things like:
1. run autonomously
2. follow a line
3. solve a maze
4. return to its home base

To those of you who ever thought about and/or built a robot, you know that those things are challenging enough in themselves to start with.
 

Parts -

When I built my first robot in 1980 there was no internet and finding robot parts, especially motors and wheels was very, very hard. When I built my last robot in 1995, the internet was a baby and it was very, very hard to find robot parts, especially motors and wheels. Now, in 2010, it is like Christmas all the time for robot hobbyist! I could not believe all the robotic web sites that exists now. Sites have EVERYTHING you need to build your own robot from scratch or from a kit. Motors and wheel, traditionally the hardest to find, are widely available in many sizes, voltages, RPMs, and so on. You can check around but I ended up buying almost all my stuff from Pololu robotics in Las Vegas. Their web site is excellent for finding stuff and they have forums where you can ask questions if you need help. They ship fast and the prices are good. Unfortunately, I don't get paid for telling you that!
 

Chassis (body) -

For me, this was one of the hardest but fun part to figure out. I looked at the robot bases on line but all were too small (couldn't hold my battery) or too expensive. There is a great web site that has more info on hobby robots than probably any other place on line, the Society Of Robots at http://www.societyofrobots.com/  If you go to no other place on line, that site you have to go to. There you will see photos of many robots built by alot of people. Some built using wood, plastic, metal and on even built on a ceiling tile (I kid you not).

The body is one area where I felt I could be creative, used existing parts from around my house and still be functional. So I decided to use an old 5" disk storage unit I had from the 1980s. What is sad is many people don't even know that floppy disks were 5" when they first came out for the PC. And I bet many younger people are not sure what even a floppy disk is anymore. Anyway, the 5" disk unit has enough room instead to mount the battery, electronics and motors. Since it is plastic, it is very easy to drill and light weight. Plus I was going to throw it out if I couldn't think of a use for it! Lastly, because it is clear, people can see all the wiring and stuff inside and will be impressed 8-). I am thinking about putting flashing lights inside, just like the robot on Lost In Space! If you look at the photos you'll also see some Erector Set parts inside used to mount the battery. There are people who have built the entire robot using Erector Set parts!

If you look on line, you will see that the predominate robot base is rectangular with the short sides being the back and front. My robot is the opposite with the long side being the back and front. You need to decide what aspect ratio works for you. If you are concerned about tight spots, go with the smaller narrower width. If you want side to side stability, go with the wider front. I don't think it makes alot of difference.
 

Motors and wheel arrangement -

For years I didn't build my first robot because I could not find practical motors and/or wheels. I would find a motor but no way to mount a wheel to it. Or I would find one but not two identical motors. For me, the two motor and one idler wheel arrangement is the best of everything. With two motors mounted on one end and the idler wheel mounted on the opposite end, you get both directions and the ability to spin in a tight circle. About the idler wheel...listen up. Robotic books are great. I have about a dozen books on robotics, some going back to 1980. However, to write and publish a book, you don't necessarily need to know what you are talking about! Case in point is the idler wheel. I have seen countless descriptions of building a robotic base using 2 motors and idler wheel(s). I feel that 99% of those descriptions are incomplete. I even question the authors of some of these books if they have ever actually built what they are describing.

The two biggest problems with using idler wheel(s) are going over bumps and backing up. Regarding going over bumps, the two motor design where the motors are mounted in the center of the base and idler wheels on both ends has a problem going over bumps. Bumps like the metal floor strip that separates two rooms in your house. With the idler wheels mounted fast on the base, the robot can get caught up on a small bump. As the first idler wheel goes over the bump, it lifts the center drive wheels off the floor and the robot gets stuck. Been there, done that. You could fix with springs, etc, but that's a mess.

The second problem is 99% of the articles I have seen using idler wheels show a SWIVEL idler wheel like you see on all chairs and the like. Big problem. When you back up, the idle wheel swivels around so you don't back up straight. When you go forward, the idler wheel again turns around to face the direction of travel. While it is turning, it acts like a rudder.

Bottom line is the robot never goes straight after a backup and you never know for sure actually what direction it is going to head in afterward. If you just want a robot to roll around your house, you may not care. But if you are trying to solve a maze or follow a line, you do care. The solution is easy. If you robot is light enough, you may not ever need an idler wheel.

But if it does, just use a non-swivel wheel. See my photo. Non-swivel wheels are harder to find. I went to Home Depot and they had a ton of swivel wheels but only 1 or 2 fixed wheels. I had a swivel wheel on my robot at first and had all the problems mention above. I switched to a fixed wheel and all the problems went away. FYI!

At first I tried using a small ball wheel for the idler. Unfortunately it did not work out good, too small. So that is when I went to the much larger swivel wheel and then to the fixed wheel. You need to remember that the rear wheel has to carry some weight and needs to be able to handle the stress of forward and backwards movement and the resistance it will face on carpets, etc.

Motors - www.Pololu.com  has alot of great motors. Figuring out what size wheels I wanted help decide what RPM to choose. Using 12v DC motors was a given for me as I already had a 12v battery and liked the selection of 12v motors. Plus I could run a 12v motor on 6v if required. . The hardest part for me was figuring what RPM I wanted. I was planning on using PWM to control the speed so having a maximum RPM too high is no longer a big concern. For the 3.5" wheels I needed to use for the motor, I selected 80 RPM motors. At full speed, it moves along well and with PWM they can slow to a crawl. See my video to get an idea of how fast 80 RPM is with 3.5 wheels. Following lines and doing mazes, 80 RPM is more than enough. But if I wanted a robot that goes faster at full speed, I would next time go for a little faster, like 100 RPM.

IMPORTANT POINT: Keep in mind also what kind of motor controller you are going to be using!!!! If you use an H bridge, some bridges lose alot of voltage. Your 12v motor may end up only seeing 6 or 7 volts and going alot slower than you expected! So you may want to go to a higher RPM motor to make up the difference. Again, you can use PWM to slow it down but you can't make it go faster (not using 12v anyway).

The motors draw 300ma free running and about 500-600ma normal running. They stall at 5 amps but honestly, mine have never stalled. If the robot gets caught some place, the wheels break traction and just spin in place, the motors have never stalled on me. Each motor has a .1uf cap across it. With my 12v battery, I can run the robot for days on a charge. They are mounted to the base with a pair of very nice brackets that Pololu sells. Since I am not real mechanical, I rather buy the brackets than try to make some at home.

Wheels/hubs - Not being real mechanically inclined nor having any machine shop tools, attaching wheels to the motors has always been the biggest problem for me. Not anymore. Pololu sells the motors and wheel and the hubs that attaches the wheels to the motors so it is a no brainier now. Regarding wheels, I got the biggest ones that Pololu sells and that mount to the hubs, 3.54" (90mm).

Motor Controllers -My first idea was to build an H bridge using N channel MOSFETs. I have built Bipolar H bridges before and they worked pretty good. The MOSFETs are cheap, easy to work with and can easily handle the current. I have the Robot Bonanza book and it has a schematic for an N channel H bridge. It was similar to the circuit I had seen so many times before so I built it assuming it would work right well. Unfortunately it didn't. Once again I ran into the problem where people write books but have no idea what they are talking about or don't explain things well enough.

First problem is driving the circuit. I was using the output of my microcontroller (TTL level) to drive the gates to the controller. Didn't work. To turn on the top MOSFETs requires alot more than 5 volts. So I added a CMOS gate which runs off of 12v. Now the top MOSFET gate was getting 12v. Second problem is unfortunately 12v isn't enough to fully turn on the top N channel MOSFETs either! Gee, why didn't the book tell me any of this? Bottom line is the motors were only seeing about 6.8v because even though the bottom MOSFETs were turning fully on, the tops weren't and they couldn't unless you could feed them with much more than 12v. These things you don't realize until after you have spent alot of time building and troubleshooting the circuit. Been there, done that. Needless to say, I am using the Robot Bonanza book as a paper weight today.

Although I liked building the H bridge (I felt like I was not cheating here), it wasn't going to work unless I redid it from scratch using P and N channel MOSFETs. But I got lazy after all this and I went out and bought from Pololu a dual motor driver board. It can handle the stall current of the motors, has braking, is about 1/4 the size of my H bridge and like the H bridge can handle PWM and directional control. There is very little voltage lost with it. It works very well and has nice LED lights which show you what it is doing (adds to the wow factor). It doesn't require any heat sink surprisingly because the MOSFET have very little resistance when turned on.
 

Microcontroller -

I think the biggest advance in the past 15 years is in the robot brain, the computer. My last robot used the 8052 with on board BASIC. It was a great chip. I built the 8052 circuit using my own design and ideas from magazines. I had to wire up everything, the I/O, the buffers, serial port, the memory, the CPU, everything. It took up 2 boards and took a while to wire up and troubleshoot. Then I had to write BASIC code for it to do anything. There were also the 6800, 68000 and 8085 CPUs that were very popular but all had to be built from scratch. And those chips had no BASIC so you were really on your own, having to write in machine language. And to program them I had to use a serial port from a PC running at like 300 baud.

Today it is awesome. Today you can get a microcontroller already made which has EEPROM, RAM, I/O and even built in USB for alot less money than you could buy the parts for. I decided on the Arduino Mega microcontroller which has alot of I/O ports and much memory. And you just plug the Arduino into your PC's USB port and you are off programming! It can be powered off of 12v directly and supplies 5v to other stuff. Plus it has DIGITAL pins and ANALOG pins and pins already setup for PWM and communications. And the ANALOG pins have built in ADC! It is unbelievable what these modern microcontrollers can do. You would really have to search far and wide to make a board that has all of this stuff built in in the size of this and at the low price.
 

Software - not finished

At first when I read that most microcontroller used "C" programming language, I stepped back and thought "C? I had been exposed to it years ago but never programmed in it. I always liked BASIC. But once I got the microcontroller up and running, I found "C very easy to learn. It is all in the syntax. And there is so many "C" programs and bits of programs on line that anyone can program in "C". "C" is powerful and well suited for robots. Plus robots don't need or use the fancy, more complicated aspects of "C" so it is easy to pick up. What is super nice is you don't need to buy a "C compiler or anything else. There is a free Arduino programming environment that runs on your PC where you write your "C" code and then download it to the robot. It compiles and checks for errors and all kinds of neat stuff. I will be the first to say I am not a super programmer and I don't write programs to use the least amount of code or to be the most elegant. I actually write most code on the fly. With that said, attached are a couple of programs I wrote for Arty. The run around the house code. The line follower code.

Battery -

Since I already had this very nice 12v, x amp hour, sealed lead acid battery, I designed the base and robot around the battery. The battery weighs about 6 pounds and can power the robot for a long time; I don't have to worry about recharging after 30 mins or having to replace alkaline batteries. It works for me. If I want to go very small, I would have used smaller everything, motors, base, battery, etc. If you do use a heavy battery, it would be better to use a robot with a narrow back/front and wider length as the battery placement is important for weigh distribution. If you have a short and wide robot like Arty, if the battery isn't placed right, the robot can flip over if it stops too fast. If you notice on my design, the rear idler wheel sticks out past the back of the robot to give me enough stability so I can mount the battery as far back as possible. Unless Arty tries to climb up on something, it is a very stable design.

Power Distribution -

The microcontroller has a built in 5v regulator but because I plan to add lots of things to the robot, I am not relying on that. The microcontroller has one 5v and one ground pin for external circuits so I added my own 5v 7805 voltage regulator on a separate board. See photo. The 7805 can give 1 amp, much more that what I need currently. I added filter caps and pin connectors on the board and it feeds all the 5v requirements except for the microcontroller itself. As you can see from the first photo I didn't have a heat sink on the 7805 and it got HOT. So I added a heatsink later and it cools it off nicely.

Navigation -

One of the things I wanted the robot to do was to be able to roll around the house autonomously. Senors help do that. The robot has three sensors used to move about:

a. Sharp GP2Y0A21 IR Receiver - the robot has two of these IR sensors. One mount on top using a servo to sweep a large area and on below to detector low lying objects like chair legs. The Sharp sensors have an analog output where the voltage goes lower as the objects get farther. Conversely, as the voltage gets higher, the object is closer. So, for example, a voltage of 1.4v says that an object is within 10". The problem with sensors in general is getting consistent readings. To minimize this, most people don't take just one reading but many and average the results. Then if one reading is off the wall, it won't adversely affect the average. The top sensor is on a servo which sweeps about 100 degrees looking for objects up front and off to the sides. Since the robot is so wide, this is really needed. For narrow robots a smaller sweep may work out better. The bottom sensor works in parallel with the top and if either detect something, the robot takes action. There are examples of code for the sensor all over the web.

b. Maxbotix EZ0 Sonar Range Finder - when IR fails, hopefully the sonar unit won't. Sonar has greater range but not as accurate. It has an analog voltage output which increases as the distance increases, 10mv/inch. So if you want to detect something 10", you look for 100mv. It works with the IR sensors and if any of them detect anything within range, the robot responds. The sonar unit is mounted on the front so it won't interfere with the servo sensor on top.
 

Servo -

To detect a wider ranger, the top IR sensor is mounted to an RC plane servo and it sweeps back and forth. My servo was an old one I bought years ago but rarely used. It is relatively big but it makes no difference on my robot. The sweeping is controlled by software. It is super easy with a microcontroller to tell it exactly how many degrees to sweep, how wide to sweep and at what resolution. It all can be changed in seconds. Plus the sample code to do this is included free with the Arduino software environment.

TV Remote Control -

One of the most unique things about Arty is I can control him via a TV remote control! (Actually a VCR remote.) A few years ago Radio Shack sold an IR receiver made to work with IR remotes. I found one in my junk box and wired to the microcontroller. Next I found some software that runs on the Arduino which receives the signal from the RS IR receiver and displays the code that was sent by the remote. I had an old Sony VCR remote not being used so I used it to display codes for like STOP, RUN, FORWARD, etc. From there is was easy to write a program that uses the codes to control the robot. Since it learns codes, the software will work with any remote control.

Line Follower Sensors -

To follow lines (and mazes) I added the Pololu QTR-8RC Reflectance Sensor Array. It is 8 pairs of IR LEDs/detectors used to follow lines (like .75" black electrical tape). The sensors are .375" apart which means 2 are .75" apart, ideal for black tape. The 8 sensors don't all have to be used and you can break off the last 2 leaving six which is what I did. My sensor is the "digital" one which means you can read a zero or one (as opposed to the analog one). However, it is not that simple to use. To use it you have to write a "1" to each sensor, wait a certain time, and then read the sensor. You will then get a digital readout such as 001100 or 110000. If you are building a line follower, that is really all you need to do. But for a maze follower/solver, it gets much more complicated. With both software and setup. Regarding setup, the height of the sensors above the floor is very important. Too high and the read angle is too wide, causing 3 sensors to pick up what only 2 should. Too low to the floor and they can strike the floor or tape on high spots. I found around 1/8" to work pretty good.

Regarding the software, I have seen a few ways to work the code. Most seem awful complicated. The code I came with is very simple and seems to work well for my application, .75" black electrical tape on a white tile floor. What I do is very simple. There are 6 sensors connected to 6 digital pins. First I program all 6 pins to be outputs and I write a "1" to all. Then I make all 6 pins input pins. I wait 2ms and then read the pins (it is very easy to set delays in Arduino). I get 6 bits of data like 000000 to 111111. I wait 2ms because I put a scope on the pin and found that 2 ms gives time for the pin to respond and to get a stable reading. So with 6 bits, if the tape is directly centered, you get 001100. If the tape is off to the left, you get 011000. Way off to the right would be 000011 and so on. The robot then responds based on the results. To follow a maze is alot more complicated. On a maze you can have a 111111 which is an intersection going left to right, 111000, a left hand turn, a 000111 right hand turn, 00000 if you go off the line completely, etc. My software section goes over this is more detail.

LCD Display -

What good is a robot if it can't communicate while it is running? So I added a 16x2 LCD display. Again the code to drive it is available off the web. I use the display to show me parameters like sensor readings while it is running. The display can show both numbers and characters. I have it mounted inside the body but because the body is clear plastic it works out great. I got the display from http://www.hacktronics.com/LCDs/16-x-2-LCD-Black-on-Green.html?redirected=1.

Homing Beacon -

Another thing I want to add is the Pololu homing beacon IR unit. It is actually 2 units, each with IR beacons at each compass heading. With one as a base, the robot will look for and home to the base unit.

Voice -

My last robot had a great voice synthesizer, you would send it a text word and it would pronounce it. They don't have anything exactly like that now but they do have circuits that can record and speak x number of words under different conditions. I will experiment with that once all of the above is working.