inertial guidance using RC type gyros ?

Has anyone done any inertial guidance using one or more of the small RC type gyros available ? I'm talking using it on a land based robot, not air or water, so it would only required 2D movements/computations. The type I mean are like "GWS PG-03 Micro Gyro" or "Head Lock RCE-500X Gyro" found at

formatting link
.

I'm thinking that for many applications a gyro would be better than odometry. Specifically, to determine if a robot is actually moving even though the wheels are grinding away or slipping, such as when blocked or gaining slow ground in mud or other slippery surfaces. On the other hand, can they be used *for* odometery - to measure distance traveled ?

Is there free source code out there for making use of any of these in robots? Got any favorite brands/models of gyros ?

Thanks! JCD

Reply to
pogo
Loading thread data ...

You can do that with gyros and accelerometers, but the accelerometers have to be significantly better than the ones usually seen in R/C gear or even as MEMS parts. Accelerometers give you acceleration; integrating this to get velocity is subject to cumulative error.

Rate gyros give you angular rate. Gyros don't tell you anything about linear motion.

The problem of integrating accelerometers, gyros, odometry, and GPS is hard, and the available solutions are not yet cheap. In 2004, it took a 4U rackmount cabinet full of gear to do it, and that's what the CMU team's Grand Challenge vehicle had. Now it's down to something the size of a thick book at about $7K.

Merely detecting wheel slip, though, isn't hard. You just need encoders on all wheels, both driven and non-driven. That's how ABS works.

John Nagle

Reply to
John Nagle

at

formatting link

Pogo,

I did some experiments with a Futaba Heading Hold gyro for robot navigation but without much success. These are really designed to have a human in the loop for periodic "corrections." For the navigation on my jBot off-road robot we ended up using a commercial IMU from MicroStrain consisting of gyros, accelerometers and magnetometers.

There is a writeup on the integration of that sensor data with the robot's odometry at:

I don't think the hobby R/C gyros like those used on my TREX-450 are accurate enough for robot navigation. I know that Larry Barello's FIRST team in Seattle, as well as some of the more advanced battle bots, used R/C style gyros in order to drive in straight lines, not quite the same as location but still a useful navigation improvement.

cheers, dpa

Reply to
dpa

at

formatting link

The LogicTech 2100T and 6100T are very nice gyros, and the latter has a PC interface IIRC.

One additional used for these gyros on a robot that I forgot to mention in the previous post is outlined in the jBot IMU Odometry article in the section on determining if the robot is stuck.

Basically the technique is to compare the rate-of-rotation of the robot around its center (in radians per second) as measured by the gyro and also as calculated from wheel odometry.

Normally the two should be the same or very close to the same. When they are very different, the robot's wheels are slipping. And when the rate-of-rotation calculated from the wheel odometry is large and that from the gyro is near zero, the robot is probably stuck.

Details of this and other methods for determining and responding to the condition of being stuck, high-centered, tangled in brush, etc, are covered in the last section of the writeup, along with a video of the robot getting itself unstuck from several situations:

Even without the IMU, a simple gyro might be useful in several of these behaviors, like determining if the robot is trapped, that don't require location accuracy.

regards, dpa

Reply to
dpa

If you are using differential steering, I understand the measurement of angle is the biggest problem, while overall distance is relatively accurate... at least in indoor conditions. For example, say you've instruct your robot to travel 120", turn 180 degrees and then return where you started. The odometry might say you've traveled 240", but you've really only traveled 239". This is 1" shorter than you planned, because of wheel slippage. But, you might end up significantly further away from your target if the robot doesn't travel a straight line. When one wheel slips more than the other, it will cause significant changes in your angle. If your path is off by 2 deg, then you will end up over 8" from your target. (it is a bit more complicated because your angle continuously changes)

I imagine a gyro would be a benefit this situation greatly. Even if there is drift, you can reset it any time your robot comes to a stop.

Regarding acceleratometers, I do recall reading about car navigation devices which existed before the GPS system was created. The stand- alone ones, which could not have access to the wheel odometry, worked using accelerometers. They would adjust for drift by resetting reset their position whenever you made a turn. It could do this because it would assume you would only turn onto a road, and could compare your estimated position with its internal map. As long as the drift was less than the distance to another road, it could properly figgure out your position.

Joe Dunfee

Reply to
cadcoke4

You're talking about the original ETAK navigator. I know that system and its designer. It did in fact have access to wheel odometry. The original version had only odometry and a magnetic compass, yet using map data, it was able to correct most errors as the vehicle made turns.

The Etak people did build an inclinometer/rate gyro combo, well before MEMS accelerometers and gyros. It was about the size of a soda can. The gyro was a motor driving a flexible metal plate, with capacitive sensing of plate deflection. The inclinometer was an electrolytic 2-axis device. I used to have a few of these units. That approach wasn't that useful. They also played around with an optical device like an optical mouse, intended to look down at the road and measure movement. They never even tried accelerometers; ones good enough to be useful for measuring velocity by integration cost too much.

Once GPS got going, it wasn't worth the trouble to do an inertial system.

John Nagle

Reply to
John Nagle

at

formatting link

Has anybody got

formatting link
to work with a UAV -
formatting link
what is the cost.

Reply to
backspace

pogo: keep in mind that an rc heli gyro isn't even *meant* to be a sensor, per se; it is meant to be a servo loop controller for a complex system. i have never used such a device, but offhand, were i designing one, i would certainly consider derivative/integral response.

dpa: do you know what kind of sensors the microstrain unit you used contained, i.e. how much of the accuracy was due to optimal signal processing and how much was due to expensive components?

for some time i have been pondering building a 4-rotor electric heli but one of the big unknowns is how much accuracy is even theoretically possible with a tiny mems imu.

-chris.

at

formatting link

Reply to
e c kern

at

formatting link

Hi , I am also trying to get a GWS-PG-03 gyro to connect to my 8051 based land robot , it is not very sophisticated just 4 motors differential drive with IR sensors for obstacle detection. This gyro from what i have read seems to give PWM output thats why i was really interested as my 8051 micrcontroller has the PCA ( programmable controller array) to automatically read PWM input by detecting a rise or a fall in the PWM signal. i am planning to buy a oscilloscope to verify the signal coming from the gyro ... until then this seems to a interesting thread to look out for info ...

Regards Deepak

Reply to
dbbarua

Now *that* is exactly what I'm looking for! That's a great approach to achieving what I probably would have written waaaaay too much code for!

thanks ! JCD

Reply to
pogo

PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.