Newbie, sensors buying advice.
mugur
Posts: 105
I need advice for buying some sensors to start with.
i want to build an autonomus rover, and i will start with the propeller robot board. For having autonomus navigation from point A to point E (going thru B, C, D) and obstacle avoidance (not line follower for now) what sensors should i buy to start with ?
I know that in time i will buy a lot of sensors and modules, but for know i have to experiment on a small 4x4 RC car inside my house. Once the project will start showing results, all will be moved on a Traxxas E-Maxx 4x4 truck. Thank you for your help
i want to build an autonomus rover, and i will start with the propeller robot board. For having autonomus navigation from point A to point E (going thru B, C, D) and obstacle avoidance (not line follower for now) what sensors should i buy to start with ?
I know that in time i will buy a lot of sensors and modules, but for know i have to experiment on a small 4x4 RC car inside my house. Once the project will start showing results, all will be moved on a Traxxas E-Maxx 4x4 truck. Thank you for your help
Comments
You should include IR sensing. In addition to its use for object detection, you can use it for remote control. Different objects work with different detection methods. Some things are very reflective to ultrasound, but absorb IR. Some things reflect IR well, but absorb ultrasound. Some things don't work well with either. If you have an IR sensor, you can also use IR beacons to mark off areas. This is often suggested for navigation within a house where you can have an IR beacon in each room sending out a different code and/or other information.
One is to navigate. I use a compass for direction and a U.S. Digital encoder attached to one of the motors on my E-Max. For short distances (<400 feet) on level ground without buried iron, this setup works well. Tilt compensated compasses are more expensive and may be difficult to use on rough terrain. For longer distances you will need to add something (I can only think of GPS) to bound the unbounded error growth of odometry. Some people use only GPS for RC car based outdoor robots, but GPS is not that accurate. If money is no object, the IMU used by Dr. Anderson on JBot is great.· If you don't have something to backup GPS make sure that you robot stops when it loses the GPS signal.
Another is to avoid obstacles: I agree with Mike Green that using at least two types of obstacle is necessary. A third type is mechanical switches.
I would highly recommed having a way to tell what your robot is doing. Seeing the robot do something and trying to guess why is a frustrating experience. I use an XBee to radio data to my laptop to be captured for later review. You could log to EEPROM or serial RAM. Another option would be a 4x20 LCD.
John Abshier
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Samuel
www.RobotBASIC.com
·
type of sensor to do a reasonably complex thing... it will fail under some real world condition.
My best advice is to use at least two different types of sensors for any mission critical task.
Range is also something you need to think about. Some sensors work great... but with the wrong range.
One of most interesting things about the Prop is the growing number of different types of sensors that are supported... eventually you will want all of them[noparse]:)[/noparse]
Rich
I feel a little bit better now. I am sure you are all right and i have to think of many types of "world sensing". Like the living creatures do, a good robot will have to rely
on more sensors to interpret the world. What i am trying to achieve is to make small stepts, with indoor first, and then to move them on e-maxx. I do have a quite acurate GPS module (Ublox LEA4H) but as i said, i have to get there before i use this. I was also thinking about intertial measurement but this is a bit too far for me now, i have to understand the things right, then to proceed accordingly. The money are not an issue. Spending them wise it is...
For John: I hope i will get that far as you are ... I do have another question: is there a way to be in permanent comunication with the robot? I mean, using x-bee i will be able to communicate live with it, and able to see on my laptop what actions are executed in real time? log recording is also nice.
1. is it possible to use this compass for a start and change it later (since i already have it and it is I2C) ?
www.robot-electronics.co.uk/acatalog/Compass.html
2. i also have the ultrasonic ranger SRF02 www.robot-electronics.co.uk/htm/srf02tech.htm
and a USB-I2C adaptor (this i dont know if it is of any use)
it is possible to use it with proppeler? i know is not the best performer, but tested on my PC with it's USB adapter, the precision looks pretty much decent to me...
John Abshier
Post Edited (Mike Green) : 1/3/2010 5:54:22 PM GMT
Be sure to get the xBee that has the 802.15.4 firmware like www.sparkfun.com/commerce/product_info.php?products_id=8665 or www.sparkfun.com/commerce/product_info.php?products_id=8742. This firmware is the easiest to use for a simple wireless serial link.
Is there any chance for me to get xbee from Europe instead of US? Any european member willing to help me with this info?
a 4x4 chassis may not be the best (read 'simplest' ) starting point as a course change isn't 'instant'.
For that a BOE would be better, or just about any tracked vehicle.
IR Sensors. If money is an issue, then I believe SparkFun has IR emitter/receiver pairs for as low as $1.
(At that price, it doesn't matter that much if they end up being tossed away as unsuitable)
Their gyros and accelerometers(don't know about their compasses) all use voltage levels for output data, so can be read using a simple AD-routine (The Propeller is capable of doing a lot of AD operations/second) the same way as the IR sensors cn be read.
Compass...
An electronic compass may not be very useful indoors unless you have a very large open room with NO POWER CABLES under the floor.
Also, the model you mentions does not like tilting...
Instead, I'd suggest a gyroscope and reading compass info from your GPS if you find that you need it.
(You use the Gyro for short-term course corrections - aka keeping it on a straight line - and the GPS to calculate the next leg of the road)
Accellerometers.
These can be rather ficcle, and the math involved for calculting speed and distance can be rather involved.
It may be easier to nick the optical sensors from an old mouse with a ball inside, and attach those to a drive axle and set a COG or two to count...
GPS... Don't expect better than 10' resolution even on a good day, it'll be more like 20' - 30' most of the time, with aa slowly 'improving' fix when the vehicle stands still. This all depends on how many and which sattellites it can 'fix' onto.
You really don't want to know about vertical resolution...
If your vehicle is to find a specific target it'll need some sort of 'marker' in the target area.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Don't visit my new website...
Yes, my vechicle will have to find something, you are more than right.
But this is the very last thing on the list, being the whole purpose of this.
Oh, and i am already considering a small tracked vechile base for the tests.
Thank you for your input.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
images-team.ro
Post Edited (mugur) : 2/20/2010 2:32:13 PM GMT