pitch roll measuring

Im working on adding force sensitivity to the gai tof my hexapod. Ive chosen a very simple aproch to this. Simply added 0.1 ohm resistors between servos and ground, using analog MUX and intenral A/D of the 18F452 to measure the current passing thru each servo. This is not going to be very acurate, but hopefully it will be sufficent for what i need.

In addition to this it is nesesary to monitor the real pitch and roll of the robot. One way is ofcourse to use a 2 axis accelrometer and integrate on it to get the tilt . What i would prefer is some kind of module that has this built into it. Ive read alitle about electrolyte tilt sensors. To my understanding its basicly a electrolyte capasitor that has its capasitance changed by beeing tilted (the electrolyte flows between the condensator plates).


Does anyone know of any prebuilt modules, IC's or any other inteligent way to go about solving this problem?

You have to specify what kind of accuracy and resolution you need first. 1 degree resolution? Do you really need a dead reconing of the tilt of the device or do you just need a basic 'losing ballance' sort of switch. Basic tilt switchs will work if that's all you need and are just mechanical switchs that switch if the angle they sit at is greater than say 45 degree's. Mounting the switch at a specific angle will give you a trigger at any degree of tilt you want.

tilt

I want to use a tilt sensor to help regulate the force each of the robots 6 legs pushes against the ground with. I have sensors in place measuring the current each servo draws as it pushes against the ground. In adition to this it is i think neasesery to have some control of the pitch and roll of the thing.

I havnt listed any specs as im not entirly sure what i need. A qualified gues would go somthing like this:

0.5 - 1 degree acuracy
0.5 degree resolution
100 ms response time

powerconsumption doesnt matter at all

I dont need anything very acurate, not like what would be in a presision machine or the like... just a simple feedback of wheter the robot is tilting or not...

Just a small update. I finaly found what i was looking for: a MEM dual axis gyro. sparkfun sells them, if anyone is intrested.

They have the worst temperature drift for bias and scale factor among all the other MEMs gyros though . I wish Silicon Sensing would have an IC_based one.

Probably matters more for an autopilot than a legged robot.

the gyros ive looked at so far all measure rate of turn. Ive never worked with gyros before. Does all gyroes do that or is there anyone that will measure just the turn angle? Thats what i need to know. The rate of turning doesnt matter, just the absolute angle.

Obviusly i can sample and integrate in the microcontroller but if i can avoid that it would be preferable

Any ideas?

All gyros measure angular velocity. You must integrate that measurement to get angular position (and amplify errors at the same time). For ground based tilt measurement, an electrolytic sensor will given you a more accurate reading in 2-axis (for 2-axis sensors). They are more fragile and expensive though. This doesnt work in planes or vehicles that cannot stop since acceleration skews the readings of such sensors, therefore gyros are needed.

Fragile and expensive i can live with. Ive been looking for the kind of capacitive tilt sensor you talk about, but ive been inable to find any prebuilt modules anywhere ... Have any ideas where i can get such a tilt sensor?

http://www.spectronsensors.com/

Look at inclinometers, not the tilt sensors- you dont want to have to deal with the signal and temperature compensation circuitry to implement the electrolytic chamber. The Micro-50 has digital output. They all have varying incliniation ranges.

The SpectroTilt II is $386...that's what I mean by expensive vs. $20 for a mems gyro.

Go for the SpectroTilt II modules for 2-axis measurements with varying supply voltages (like AC, 12V, 5V analog outputs).

400$ is alitle more then i had imagined ... im guesing this is way more advanced then what i actualy need. I dont need temprature compansation. And im using it on a hexapod walking robot, so its realy relative pitch and roll i am measuring. Supose maybe a MEMs gyro is the way to go about this after all... Could set it up with a dedicated micro to read and integrate it, then read that data digitaly. Strange tho, that there isnt any cheeper and simpler inclinometers around. Woulda thought that is somthing more robots could use in their sensor packs.

Thx for all the advice

You can use an accelerometer with a gyro to correct for bias drift. When the robot stops you can use the gyro to find out the actual direction of gravity and rebias the gyro.

Again, this doesnt work on planes because planes are always moving which never lets the accerlerometer readings stabilize to correct the gyro. But it works great for a land robot that can stop to a standstill occasioanlly. Actually, you can just use an accelrometer if you wanted...it's just not very accurate while your robot is moving.

Gyro- accurate short term while moving
Accelerometers- accurate long term when not moving

Good luck.