The 422 servo may be ok for the hip (horizontal) and the knee (vertical) but useing a 475 in the hip (vertical) may help.
You are only using 4.8vdc. If you change to 6.0vdc you will notice an improvment in the strength! Using 7.2vdc will provide a dramatic increase in strength compaired to 4.8vdc. It’s almost double the torque. Springs will help if you use them.
I’m running from a power supply at 6V, the 4 X AA is just for loading the robot to see how much i can go.
Thanks will try increasing to 7.2V 
just uploaded a short clip of the walking gait
Can see the HS422 are straining under the weight (chassis + 4 X AA), so am considering to do some major weight shaving off the chassis, screws, etc 
The red/yellow switches are the foot sensors that haev to be implemented 
Here’s the link: s31.photobucket.com/albums/c391/ZeRoC00l/FYP/?action=view¤t=P1000010.flv
I’ve also sent in a job for a new chassis, with weight reduction, and a little smaller in size. So gona fix that up and start on the weight distribution
Gona pause for now, exams are aound the corner. Will be back in Late November/December
cheers
Just a quick thought. As stated earlier you may consider upgrading servos. You have to account for the weight of the camera and other components that are not on the bot yet. Doing the weight reduction on the chasis will help but ultimatly the servos have to be able to carry the whole bot. A little time now to upgrade might save you a ton of headaches in the future?
Yup, I’m looking into that. but that will be my last option as i’m on a tight budget now Should have bought a 647 or similar right from the start. But since I was builing a mini quad, so I thought can save a little by getting something cheaper 
Currently working on the walking algo, together with foot sensors (digital), and accelerometer. Hopefully, my bot can walk over terrains and still maintain it’s balance 
PS: Anyone build a mini quad before? Or know of any links/ readings?
Have seen micromagic mini hex and it’s WAY cool!! And it’s using Futaba 3101 servos (3.6kg/cm) , yet my HS422 can’t do the same thing???
Can someone enlighten me? I’ll try to get some diamensions of my bot’s femur, etc.
cheers
Hi Zack,
I can see that it’s much tougher to write a gait generator for a quad. But I’m a bit confused, you are building a hexapod, right? For a hexapod, you have samples of code on the LM website. For a quad, well, I haven’t found any as yet. I was considering running my 'Bot on 4 legs until I got the other 2 legs. Not so much because I am cheap, but it would give me a try at a quad gait. I’ve even been considering trying out an octapod; although the extra weight may be too much. We’ll see. Different leg arrangements are definitely a possibility, as the 3DoF-C style legs are not to hard to swap out. Different lengths and arrangements of Tibia and Femur are on the task list as well; and adding an additional Femur joint is not out of the question.
I’m going with the Hitech 475 servos that LM offers. I have a bunch of Futaba 3004’s, maybe I can use them for grippers or Jaws.
Foot sensors would be a good addition; I’ve been toying with some Ideas of my own. For a quad, sounds like an accelerometer would be dictated if you intend to introduce body roll to aid in the gait. I was just at the Micromagic website last night, but didn’t see a mini quad as such, but something called “ic”? Ran out of time! Still haven’t seen where I could BUY something on that website!
Yes, it would be interesting to get your leg dimensions, and I’m sure more then a few would be interested in the leg and body designs. Acrylic plastic? Do you have it made for you? Excellent work! I’m destined to stick with aluminum, I only wish I could MILL more leg parts, rather then BEND them! I did get a small (18") sheet metal brake (cheap) from Harbor Freight a week or so ago. That might help!
Cheers!
Alan KM6VV
Central Coast, CA, USA
Hi Ze!
A group of people that have done a great deal of work on small quadrapaedic robots are the B.E.A.M. robotics guys. they tend to employ a more dynamic gait that is reliant on shifting the center of gravity. I think you can pick up a cheap kit pretty easily so you can see for yourself. Check out Solarbotics.net, its a sort of BEAM reference site.
To: KM6VV
I’m builiding a quad My professor commented that everyone is building a HEX, so I’m not to follow the flow… Furthermore, a QUAD gait is not as simple as a HEX as there are only 4 legs to support the whole body… so something of a challenge for us students
Yup the I.C. hex is a REAL beauty!!
I’m using acrylic currently (3mm) and it’s kinda too thin… I notice some flex especially in the femur (horizontal) piece. I’ve sned for a new pieces cut in 5mm now, with some modification for weight reducing and shortening. My Quad has problem standing up from off state.
The accelerometer is used to retain the horizontal balance of the quad. Even with foot switches, there’s no way to determine the base will be horizontal most of the time. As I’ll be mounting a camera at the front so I need the base to be as stable as possible regardless of the terrain.
I hope to get my hands on some aluminium blocks and have some CNC parts!!
Aluniminum adds a whole new level of professional look to the robot.
To: dreadluthier
Yup, I’ve been to that site. Thank you
However, there is too much body movement. I need a stable plateform to mount a camera and such. My bot is autonomous so it’s gona have a little bit more sensors to “see” for itself.
And I need my bot to walk over rough terrain… so a BEAM is quite out of the question.
Thanks for you suggestion 
Hi Zack,
Hey, a challenge for ANYBODY! I’d be tempted to add a shift-able weight (the battery?), that way the load could be shifted from one side to another as legs were moved. NOT CONFIRMED, but I’d like to give it a try.
I think LM used TWO pieces of Lexan for a leg, or 1/8" aluminum. I don’t mind working in aluminum! I have CNC’d my Sherline mill and lathe, and as long as the stock is not too big, I can make it. The BODY is too large! I’m working around that, 'tho.
I can see how the accelerometer might be used to assist in balance. Have you considered the “Inverted Pendulum” problem in balance?
If the body is rocking back and forth as the 'Bot walks, then a camera is not going to be too steady! I suppose you could gambol mount it, and make it a stable platform (like a guidance capsule), but that’ll be heavy and a task in it’s self!
Idea; You might try for a “trot” first, I understand it may be a little easier. Both legs on one side move at the same time. See the “Big Dog” robot trot!
Alan KM6VV
Wouldn’t the bot fall if you’re moving 2 legs on the same side?
The weight shifting (using battery pack) have been mentioned before, I’ve yet to test it out as I’m concentrating on shrinking(overall size and femur/tibia) and weight reducing for now. I am trying to get my bot to stand up by itself IF possible without getting new servos.
My new bot’s diamensions:
body reduced to: 4" X 5" (97mm X 129mm) ==> drilled for weight reduction/ allow routing for servo wires
Femur: centre of servo horn to servo horn is 1.5" (35mm)
==> min distance for my bot to prevent the femur and tibia servo from touching
Tibia: 3.5" (86mm) ==> might manually reduce if I find it too long
From the vid (s31.photobucket.com/albums/c391/ZeRoC00l/FYP/?action=view¤t=P1000010.flv) I posted, there is a very slight dip with each foot lifting up. So I thought I’ll only do the real test once my new bot body is ready.
The CG is way too high in the high stance, so I’m thinking of walking with the femur horizontal to the ground, or even reversed so that the body is very close to the ground. The bot will walk in mid/high stance ONLY in rough terrain, it will choose which to use base on a sensor that determine the condition of the terrain.
Hi Zack,
The idea is apparently to shift the weight to the opposite side, so that the two legs can be moved. Actually not just pick up the legs, but actually “push off” with them (which also shifts the balance). Bipeds, Quadrapeds, both basically have a gait that is lifting one leg, then the other (or a pair at a time). Much harder to model then the hexapods with their stable tripods.
That may help, But certainly complicates the process! How about off-loading the uP and battery for the present, to give it a chance to jog at least. You can put them back on later, or after new servos. The hip Vertical servo seems to receive the most demanding loads. Maybe you can get by with less on the others.
A little hard to tell your gait from the video. Are you using a sequencer program or something similar? You probably need to move much the legs much faster. CG will certainly have an effect.
As for my friend’s home built 'bot that I’m trying to get code running for, it’s still got “goofy legs”. the servos orientation doesn’t correspond to LM’s CH3-R, and I’ve still to sort them out!
Alan KM6VV
Erm… I still don’t see how can my bot walk when 2 legs are lifted up on the same side. I don’t have a big foot print (1 PushButton ~ 1cm X 1cm), so unless i increase the foot print then probably i can get by with walking alternate legs [eg. FR & RL or FL & RR]
Nope, now it’s hard program by my programmer, so it just walkes 2 complete cycle for now. I’m sorting out the gait algo and he’ll build it into a program (C language: PIC18F252)
My other programmer is still sorting out the main chip(PIC18F4550) to send signal to the gait chip, and still fiddling with some other issues with regards to project scope/objective. So i’m working on the gait/ legs of the bot so once things are comfirm, my legs are already done. And hopefully needs little trimming
Hi Zack,
Walking? That would probably be difficult; the “side shuffle” (both legs on one side) is apparently only applicable on TROTTING robots, like the “Big Dog”. I can see where your “diagonal legs” (sequential) approach is inherently more stable, and may not require such a severe CG shift (a simpler shift, if any).
Actually, it’s not the size of a single foot, rather, the area encompassed by the feet on the ground. That’s closer to your body size.
Perhaps you can lift diagonal PAIRS of legs. Still, it’d have to be done quickly!
I like the PIC18F252/452 parts, the “PIC BOOK” uP board I’m using (multiple) was designed for that part. I’ve since upgraded to the PIC18F4620, with twice the EEPROM. I intend to connect multiple PICs together, probably via I2C as Pete has done (see “New Controller Board” under “General Electronics”.
lynxmotion.net/phpbb/viewtop … highlight=
I’m also using a PICDEM USB board, with the 18F4550; my code will compile and run on either. Not as much EEPROM, but with the USB it has possibilities!
Alan KM6VV
Dear Zack,
I certainly agree that a BEAM robot wouldn’t be appropriate for your purposes, however there is an animated gif. on that site if you go
Solarbotics.net > Beastiary > Walkers > 2 Motor Walker. that explains the shifting of the center of gravity over a sort of fulcrum created by diagonal pairs of legs.
Werner
Hi Zack,
Check out this video on a German University site:
ag47.informatik.uni-bremen.de/do … ramies.wmv
Also these papers:
ag47.informatik.uni-bremen.de/do … et-all.pdf
oricomtech.com/projects/leg-walk.htm
EDIT:
LM quad manual, explains quad gait:
lynxmotion.com/images/data/qp2-v1.pdf
Alan KM6VV
I’m not using the reverse knee design for my bot, so this leg design is out, but I’m taking note of the gait. Looks interesting
Cool! This would help alot! Thanks
I might just remove 1 servo and stick to a 2DOF leg, then I’ll lighten up the bot and probably increase my run time
Funny, I don’t see this leg in the website… Is this kit available? or is this an old kit?
Looks quite siminlar to this kit: http://www.lynxmotion.com/images/temp/2dofhex1.jpg
but the servo mounting look weird 
how is the 2nd servo linked to the tibia?
Finally exams are over… am continuing on my quad.
Made some major changes to the legs: redesign the legs and added push buttons for ground sensing
Here are some pics I took while in the testing stage:
http://i31.photobucket.com/albums/c391/ZeRoC00l/FYP/DSC01458_1.jpghttp://i31.photobucket.com/albums/c391/ZeRoC00l/FYP/DSC01459_1.jpg
Am thinking of getting a SC-32 to control the gait, as my PIC18F252 is giving me a hell of a time.
This is my quad as of NOW 
Not much to look at but i’m improving as we speak.
http://i31.photobucket.com/albums/c391/ZeRoC00l/FYP/DSC01462_1.jpg
Wonder if anyone can shed some light on stopping the gait motion when a foot sensor senses or obstacle or doesn’t sense a ground ahead?
So the robot stops the foot at that specific height, and continues with the gait Or backs away from the edge
I am now polling for the switch every XX counts down, it not only takes a considerable amount of time but it’s not very accurate. (Toppled a few times)
Any kind soul can offer their advice/SCC-32 codes and I’ll convert to C to test on my quad.
Here’s an example I found 
:
youtube.com/watch?v=Pdj6R0UTB8Y
A LM HEX with FSR pads, and it’s partially what I have in mind, and combined with the vidz below.
i31.photobucket.com/albums/c391/ZeRoC00l/FYP/th_foot_sense.jpg(found it somewhere but forgotten, it’s not mine. Credit goes to the owner)
Sorry I missed this post…
The yellow one on the right is the EH2-KT. The one on the left is a prototype I am working on to replace the EH2-KT. Here is more information…
Pasted from another thread.
I have been experimenting with the tibia on this prototype 2DOF hexapod. There is a spring that allows the tip if the tibia to float while the hex is walking. The middle leg takes the brunt of the floating when the hex is walking forward, but the fore and aft tibias do more floating when the hex is turning in place.
The tibia needs to be made from aluminum, and I need to decide if the femur will be two plates or just the one. Trying to keep the costs down.
The chassis is now aluminum…
Here is an image of the bot.
Here is a closeup of the leg.
Here is the chassis with the optional side panels.
The side pannels make a big difference in the looks. It looks 10x better in my opinion. 8)
Very interesting design, I like the servo not taking all the weight.
