• aditya mehrotra.

chip updates: let's try making this thing take a few steps, OMG IT WALKS [updates]

So today the goal is, now that we have solved weight transfer, to make this thing walk. We're also kinda testing battery power for the first time because I wanted to sit outside so chip is outside on full battery power so we will see how this goes. So far no changes in performance. I will say friction is A LOT higher outside which is probably good for us. After we do that we're going to get ride of the IK node, move all the IK into master and publish CMDS from master, and somehow we're also going to need to update trajectory with any IMU corrections and things, so this will take some time after walking. So let's get started with WALKING let's put this thing in WALK MODE WOOO!

We're going to do this ONE LEG AT A TIME. So we don't get overwhelmed in testing.

So here's test number one: I think in this test we lifted the leg a little too less. So let's try lifting it more.

Here's test number two: I think that's a step!

Here's test number three: we increased the step height and increased the gate.

And here's the final test: that's a step!

Here's a funny. In the final test there's a very very interested bee trying to eat chip's leg. I don't know what about chip is flowery but I don't think he'd taste very good. I was getting scared because the bee was getting so close to the gearbox I thought it'd be pulled in and squashed :(

So here's the good news! We can achieve a STEP. Meaning once we achieve all four steps, we can evaluate the performance of chip and see if we can get him to walk forward for a little bit. Once we get that done. I'm planning to save this version of the code and upload it to GitHub. The we're going to make a new catkin_ws and start a new version of the code that merges IK and the IMU stuff into MASTER. One main code file so we can command the COMMANDS not just xyz foot positions you know.

But overall pretty good success. The next step is getting all four legs stepping forward in the FL, BR, FR, BL configuration!

So the other thing we need to think about is how hard the motors tend to hit the deck which means we really need to put some covering on them to protect them we're going to need to do that soon but not just yet. In the meantime, let's edit the code for all four steps.

step with me: two-four legs

Yessss here we go. We're going to create the next four steps the same way we did this above step.

Okay so now we have two legs in the step-with-me defaults code. Let's see how two steps works (it did so we moved to three and four with some small changes).

RESULT! We have the robot walking four steps forward! One for each leg! YAY! So there are some instabilities of course, at the end the front-left leg seems to be in a slightly behind position than it should be and I think that's for a few reasons.

(1) The front left leg has that weird thing going on with the upper link where the metal is bent and we need to replace the gearbox. So that could be one reason.

(2) The second reason could be zeroing, we haven't gotten a really good method of zeroing the whole platform. We'll look into fixing that at some point.

So what are the next STEPS, haha get it? The next steps are going to be to get the platform to strafe and spin using this very weird walk/balance technique. Right that's the next two steps and when that happens we can go full remote control with the platform around the yard. When we're done with that we're going to freeze and cleanup the current code. And once we've done that, we're going to upload that version to GitHub and that'll be one version!

After that we'll delete the code like we said before start merging things and allow for everything to be done in Master because of the complexity we need. That's when we'll start cleaning up code and making the platform better.

Another thing we noticed. The sit/stand of the platform works well inside because the feet can slide over the floor and the walk works terribly inside because the feet can slide over the floor. Outside, however, the platform has a lot more grip resulting in the ability to walk with this gate and move a little more dynamically without slipping. But that also means at some point we'll have to change the sit/stand because in some of the videos you see I have to pick up and place the platform back down to get the feet in the right position when the robot sits back down. But that's for the next version of the code.

Overall, very good success today :)

#updates #omgrobots #walking_finally

1 view0 comments
© copyright 2019 | aditya mehrotra