Bipeds

[mattallen37]

...Also I would not use sensor multiplexers. If you use digital sensors you can chain up to 127 sensors to one sensor port. Just make sure that you can change the I2C address when you buy a sensor. (Hitechnic sensors do not qualify)...

Why not use them? I have had great success with them. I don't think I would ever buy one, but only because they cost more than they are worth to me. I have made several very successful ones though (I have two built right now, one for the RCX's and one for the NXT's), and they cost very little to make.

Actually, there are 127 addresses, but because of the pullups and cable length required to connect them, 5 devices is typically about the limit.

[doc222]

What I hope to do near term:

I am planning an obstical course for X2b.
Needs:
to be able to navigate a simple but not straight path. To be able to know if an object is to tall to climb. To know if it is gong to fall off a plane)stepping into a hole so to say. To be able to judge how high a step is or low it is, depending on if accenting or desending.

Issues now: The nxt brick is full right now with 4 NXTmmx and abc in use. each mmx has an open sensor port, so i have room with out a sensor mmx to play with 4 sensors.

I have a prototype dual funcion light sensor on one foot now. what is going on here is the light sensor has two jobs to detect black(dark color or a light color) the dark color is a crash bar in the foot that is the front area of the foot if it kicks anything it will close against the light sensors and that part the touches it is black. This way if it hits something it didnot "see" it will stop the walking cycle. in the bottom of the foot is a rod if the foot touches down on the ground this rod moves up and blocks the light sensor with a light colored plate. This is for if say it is about to step into the abis and it thinks its walking flat ground. just before it would shift weight to the lead leg the push rod must be pushed or it would clcye back 1 half step to a standing possition.

This would use up 2 ports leaving two, I have a sonar for now(not the greatest to measure with but I will start with that until I get a better one from Mindsensors. This is in the top and will scan a given distance and pan to the floor by bending at the waist. If done right I should be able to detect walls or steps to a point anyway. 3rd port


4th port? not sure i have a accel and gyro, compass but unless the first two are are in the feet the wont work for anything but balancing standing still. The trouble with this biped is it has no real body just a pair of legs with belly. The shorter the body as comparted to the legs the more a body must counter the movments of a leg this makes the upper bodly out of vertical alot and move alot as well, not sure how I would work these into it for now. I been thinking and its nothing like Legoway, it can not react fast enough with legs to over come a tipping issue once it happens, the reaction time needed is very quick. meaning the leg would need to move faster than gravity will over come it, "my thinking here out load, may be wrong too? Segways if say you push it, that action would tend to make fall back but the wheels go with your push until it rights its self. a biped would have to take a step back to stop the falling back, leaning into it wont work unless the robot is heaver than that of the pusher. The whole thing is just so complicated I find it daunting . So to keep on the task at hand and not to get discouraged I just take it one task at atime. taking what i can do and trying to make that work and learning what I do not know as I go(and thats most the time)

two distance sensors would be nice too. one looking forward and the other down or in a foot to measure low objects and the bot would not have to look down to much.

anyway alot to test and to work out. I am sure about everything I planned will change or bring up other issues as well. Its not always one step forward, at times it takes a few backwards.

Doc,"still moving ahead staticaly"

[aswin0]

Doc,

I do not see any use in gyro's or accelerometers for your robot. I think the main issue is balance control, this can best be solved by using force sensors in the front and back of the feet. From differences in force one can calculate the position of the center of gravity. A gyro could only tell you ho fast your robot is falling over, but then it is too late already. The accelerometer can give you tilt info, but to find the center of gravity from it requires a sensor on every joint.

[mattallen37]

...Issues now: The nxt brick is full right now with 4 NXTmmx and abc in use. each mmx has an open sensor port, so i have room with out a sensor mmx to play with 4 sensors...

...This would use up 2 ports leaving two, I have a sonar for now(not the greatest to measure with but I will start with that until I get a better one from Mindsensors. This is in the top and will scan a given distance and pan to the floor by bending at the waist. If done right I should be able to detect walls or steps to a point anyway. 3rd port...

I don't know if you want to deal with electronics prototyping, but this sounds like the perfect platform for a PICAXE 20X2 program I just made. It would allow you to read 10 ADC's, with 10-bits at pretty high speed. This would be like expanding the NXT to have 10 more analog sensor ports (or more, depending on how many 20X2's you use).

As far as the IR distance sensors, I know of a place you can get them really cheap (less than $11 even for the long range ones), and shipping price is pretty good.

Using these, you should be able to have an almost unlimited number of the IR distance sensors on your robot.

[doc222]

Doc,

I do not see any use in gyro's or accelerometers for your robot. I think the main issue is balance control, this can best be solved by using force sensors in the front and back of the feet. From differences in force one can calculate the position of the center of gravity. A gyro could only tell you ho fast your robot is falling over, but then it is too late already. The accelerometer can give you tilt info, but to find the center of gravity from it requires a sensor on every joint.

I agree the force sensors would be the best way.


As slow as it is mechnicaly I could use the ecel/gyro to have set off a scream for help as its falling over Then as its laying on the ground twiching the legs would be trying to finish what ever reaction they were trying to overcome

[mightor]

As slow as it is mechnicaly I could use the ecel/gyro to have set off a scream for help as its falling over Then as its laying on the ground twiching the legs would be trying to finish what ever reaction they were trying to overcome

Make it display a helpful message on the screen:
TODO: fix balancing issues

Perhaps the dflex sensor would be able to help you with this. If you have a "toe" on the foot with a spring and the dflex, you could perhaps see how far you're leaning over.

- Xander