Pages: [1]   Go Down
Author Topic: General Questions re Robot startup  (Read 1400 times)
0 Members and 1 Guest are viewing this topic.
Offline Offline
Newbie
*
Karma: 0
Posts: 46
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Hello All,

I not sure this belongs in this section, but as it pertains to my Hexapod Robot I thought I would post here.

Currently I use a Spectrum Dx7 and Ar6100E receiver to run things

I am looking at a startup sequence for my bot this consists of the following:

1) Query the Servo controller and confirm Input voltage, Firmware version and of course that I can actual talk to it.

2) Have the user establish the high and low positions for the throttle and directional control.  Basically move the sticks and save the values.

3) Flick a switch to move each the servos as a serviceability test, servo movement is automatic and done sequentially

4) A sensor test to confirm data is being received correctly no long lines of zero - this is a target on a stick that is at a known range it works pretty well.

5) Finally flick the switch back to continue

All requests currently pass through the serial interface, a small LCD with buttons to be added.

My question is this overkill ?  I look at it as a logical startup process but I have been known to over think things.

Any input would be appreciated, I need a rest from IK, it has been way too many years since I had to think that hard.
Logged

the land of sun+snow
Offline Offline
Faraday Member
**
Karma: 158
Posts: 2882
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

All I can say is you start implementing and add what seems appropriate.

IK is nice, but I always thought it was somewhat overkill for a hexapod. You're not doing
pick and place. Do the feet really need to come down in precise locations? Actually, a walker
should have sensors in its feet to provide feedback to the main processor. Eg, some people
found that the easiest way to get a hexapod to walk over obstacles was to sense if the foot
banged into something, and then literate on code saying, to effect

a. lift foot a bit.
b. move foot forward.
c. if collision, then go to a.
« Last Edit: December 07, 2012, 01:02:07 pm by oric_dan(333) » Logged

Offline Offline
Newbie
*
Karma: 0
Posts: 46
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Oric_dan, interesting observation and it mirrors mine, it seems like a lot of math unless you are try to keep your center of gravity precisely set for uneven ground.  I setup an idiot walking system that works fine for forward motion to verify servo range of movement. 

I also have been playing with FSR's as that seems like a good way to determine a safe leg position on the ground.

Thanks for your feed back.
Logged

the land of sun+snow
Offline Offline
Faraday Member
**
Karma: 158
Posts: 2882
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

I'm not sure that IK has anything to do with keeping the COG positioned properly. Most
people implement very simple gaits, like alternating-tripod gait, for hexapods. And
hexapods are very stable in general, as compared to other walker geometries. People
rarely use gaits that control each leg completely separately, that I have seen.

If you really want it to walk over uneven you should do some research into what people like
Boston Dynamics Little Dog and Big Dog do [albeit those are quadrupeds]. Eg, they will
advance a single leg, and then knowing the position of the 4 legs, move the COG into
position between all four legs. A lot has to do with having accelerometers to read body
position. Go to youtube.com, and search on "little dog darpa".
Logged

Pages: [1]   Go Up
Jump to: