Here are the center plate files as they are.
If anyone want to have a look iām open to changes of course.
SESV2-Hexapod-Center-0.1.zip (130.5 KB)
1 Like
The LSS Adapter has 10K pull-ups on the FT232 IC Rx and XBee Rx on the 5V and 3.3V side of the BSS138 mosfet
Looking good!
I hope you all are doing fine on these Covid-19 days. Iām currently working at home and theyāve just closed all schools and kindergardensā¦
Like @dialfonzo posted they are not that different from the Phoenix @xan (Iāve just replaced one defect servo, guess which one). I think our first goal is to make the hex work properly and have some options for different remote control systems. Then play with different designs. I do have some ideas myself, involving two extra servos and reduction gears.
@kurte a buzzer is really convenient, nice work! Your correct, Iām using FP math on the T3.6.
Iām not sure if we will ever agree on one (remote) control only. Like mentioned above there are reasons for why standard RC remotes are used for combat robots, reliability and accuracy. Then again that can be obtained through other DIY systems. The ArbotiX controller Kurt mentioned is very simple and might be good enough for many. Using high quality gimbals is a must though. Personally I really dislike the old PS2 controller when it comes to precision (data read from the gimbals). The RB-Dfr-223 sure looks like an interesting option, for some.
Personally I prefer getting data back from the robot too, like battery status and commando confirmation.
The software Iām using on this remote are based on the software used on the ArbotiX. Iām sure it could be improved a lot. The display Iām using (a very cheap Nokia display also has some limits when it comes to update speed). Currently Iām sending data at around 100Hz. The main loop on the robot also runs around that speed. So itās pretty smooth. I could post a video explaining and demonstrating how I operate it, if interested.
3 Likes
Hi @zenta and all - Hope you are all doing well. We are hanging in there. We saw some of this coming and stocked up and pretty much self isolating. Will be interesting to play with the new stuff, but it will probably be several days after I receive it that I will feel safe enough to open box and then probably again try to disinfect everything before touching anything without glovesā¦ Yesterday I did pick up my first board from OSHPark that has connectors for LSS and Dynamixel servosā¦ They are probably pretty safe now as with gloves, I opened envelope and took the three cards out and put into zip lock and through away the rest, and then added rubbing alcohol and let it sit in it for probably about an hourā¦
In the next few days will see how much of the board I have parts forā¦
Remotes: As you mentioned it may be difficult to come up with a single control that makes everyone happy. Yours looks great. I still have a similar version around here somewhere, although not nearly as refined as yours. For me ones like the Arbotix one was sufficient for most things I did as I am not coordinated enough to handle that many controls anyway 
Wondering about your bi-directional communications with yours. I know that earlier I had issues with XBees where they would not consistently and reliably change communication directions and at times there could be long gaps of time where nothing is receivedā¦ I believe over at Trossen, that @jwatte also had issues with XBess as well and converted over to different wireless setup, maybe something like two RF95s or??? Where he had two pairs on both the remote and the robot, one for each direction? Again not sure if this would be a direction RobotShop might want to go?
As mentioned the RB-Dfr-ā¦ may well for most. Over the next couple of days, I may try to play with the Teensy and Bluetooth stuff again and verify if my PS4 is still working. If not I have several different PS3sā¦ While waiting I will probably try some of it out using the Dynamixels.
Again hoping everyone is doing well and staying safe!
2 Likes
@kurte - Nice PCB and i would not worry too much about the virus after your dip in alcohol.
Iām looking to cut the center plates very soon, to get the initial hardware out of the way.
I really like the legs that look more āinsect likeā but they leg design that i came up with is using all of the standard SES brackets which is a big plus from a company standpoint. (sorryā¦)
Everyone is open to do it the way they like even more if the code adapt easily to different shapes (dimensions).
Features:
- BotBoarduino mounting pattern
- Raspberry Pi / LSS-Adapter mounting pattern
- Slot for hook & loop
Here are 3 stance (low / mid / high)
Here some ādimensionsā that might help
If you donāt have any changes to ask, iām going to produce those for you guys.
Will you require a basic assembly guide for this ?
All the best,
3 Likes
Your configuration will work just fine @dialfonzo 
I donāt see any need for further assembly guide your pics speak for them self. I Assume you are using an idle horn instead of four nuts when assembling the coxa using the C-bracket and the other servo bracket?
Nice finish on the OSHPark board @kurte! Good to hear you are safe (extremely safe IMHO) :-).
About the remote, I still have issues with the bidirectional communication. I thought it was bad coding from my side. Iāve not had time to figure it out though. Sometimes I loose data from the robot and I donāt get the status I want.
For the ones not familiar with it, we created a ācable clipā that can be mounted on the new generation of SES brackets.
Those are handy to run the cables as well as equipped with magnets.
As i was adding the clips to the legs, i added some clip positioning on the center frame as well.
4 Likes
@dialfonzo - My guess is I will be able to figure it out, but having drawings like the above will make it simpler!
And it looks good enough the way you have it for a good starting point.
@zenta - XBees as mentioned I ran into the issue and I know others have as well, including jwatte as you can find several threads on trossen like: http://forums.trossenrobotics.com/showthread.php?7287-Still-getting-crunchy-performance-from-Xbee&highlight=xbee
One interesting thing about that thread was he did an adapter board to mount an ESP8266 in an XBee holder. Not sure if that is the direction I would go or to go back to RC or go to something like RF24s or LRAā¦
@dialfonzo One thing to note about assembly. Personally I prefer having all femur and tibia servos facing forward. What do you think about that?
2 Likes
You caught thatā¦! Somebody is watching me.
I actually cheated by making one leg assembly and duplicated it 6 times in the final assembly. In reality I would have one side mirror the other so all output shaft facing the same side.
3 Likes
FYI to all those waiting for the hexapod prototype hardware - the body panels are in production and weāll ship them as soon as we have all parts at our warehouse in Canada. Sorry again for the delay.
3 Likes
Hi All,
Got a little issue with file size for attachments here thatās been resolved today.
Attached is the actual assembly in STEP file.
SESV2-LSS-HEX-2020-03-17.zip (8.1 MB)
Our apologies, but still waiting for the top and bottom body panels before we ship the full hardware.
2 Likes
No hurry, hope everyone is doing well there and staying safe!
1 Like
Weāre doing are best and it seems to be working for most of us!
Maybe this is what weāre missing:
I definitely want a surrogate! I am starting to get shack wacky.
(Iām a home body, I canāt imagine how my wife or other extroverts feel. Also, I want my surrogate to look like Bruce Willis too)
Iāve been working diligently on Robot IK and Dynamics within ROS2. Firstly, itās been a really tough road the last 2 months getting this dog-crap (i.e. my code) working only to discover my URDF file I exported from SolidWorks way back in December had the joint coordinate systems completely randomizedā¦XYZ unit vectors were all over the place. No wonder when I entered coordinates or orientations for end-effectors they would be all twisted and wrong. lol. I had to go back and redo the URDF but now all is right in the (cartesian) world.
So I heavily refactored my code into something clean and friendlier. I now have stuff separated into State objects and the physics-of-motion models, and visual renders that act on states. Now current state and target state are just 2 sides of the same coin and I can render states as ghost-mesh models with indicators within RViz2. RViz is actually quite flexible. Iāve figured out how to add manipulators for end-effectors or what-not and context menus. In the video I have one to control the robot standing height. My node is also doing proper gravity compensation on the compliant joints and predicting torque current draw very accurately. Robot Dynamics node can handle any number of legs so looking forward to trying out on the Hexapod as well.
Robot is still shaky since I have no PID or trajectory smoothing, so that is my next task.
5 Likes
Really liking the progress Colin! Thank you for sharing here.
On another note, the body panels have arrived, but still waiting for a few additional products before we can create the kits to ship.
2 Likes
Good to hear about the hexapod. I am coding to support both biped and hex (or quad) and my latest refactor got rid of the last 2-leg specific code, so I look forward to testing on the hex too.
1 Like