• aditya mehrotra.

chip updates: small tests today [updates]

Yes so today, I don't know how many tests we're actually going to run. The reason is I want to get started on another, project for the summer. Today for chip we're going to get the IMU sorted out. And then we're going to do some research on that plastic recycling program we were thinking of - if we do a lot of research and then we get it to Dr. Culpepper at MIT, maybe we have a chance of implementing it at school. But for now, let's think about chip's IMU problem.


So yesterday we said we were going to do two things with chip. The first, is if the platform starts to fall it will return to its last stable position, the second is if the platform falls past a recoverable point it will sit down. These tests have been hard on the robot as it involves him assuming positions he was not designed for and literally, attempting to fall. So we're going to finish up those tests this morning, and then from either later today or tomorrow, get to walking.


Essentially, all we're doing is testing the code from yesterday.


So in the following video, what we're going to do is send the platform to an angle of 0.08 and see if it recovers itself.



Then what we'll do is send the platform to a much steeper angle and see if it recovers itself or it simply sits down.



Both of these tests seemed to work extremely well. The threshold for recovery is 0.075 on the IMU and the threshold for sit is 0.11 on the IMU but if that becomes problematic we can change either of them.


It's important to note this whole algorithm is running ALL THE TIME so in any mode this recovery will occur (which means we might want to disable it for certain modes like dance mode). But no matter, that's something to deal with later as we start teaching the platform to walk and things. Right now I want to stop doing any more tests because the platform is standing and sitting perfectly so the next time we see any of this it's going to be teaching it to walk. After we teach it to walk we can teach it to use its front arms as manipulators using CMA and such.


We're going to start with all of these functions in a very very static way (no dynamics just yet), and once we've figure out how to, you know, make it walk statically, we'll work on dynamics. But It think it's also important to list what we've done so far:

 
 Chip Abilities So-Far:
 (1) hardware setup including ROS, communication from jetson to 
     RIO
 (2) PID tuning and calibration of all the motors including over-
     current limiting
 (3) Development of ROS node infrastructure/software 
     architecture.
 (4) Using webviz, rosbridge, and other utilities to setup a web-
     based debugging interface for the platform.
 (5) Generating appropriate stand/sit trajectories, and using the 
     IMU for stabilization and sitting when falling.
 

The rest of the time was spent purely on mechanical and electrical upgrades we won't talk about as we want to assume they're inherent features of the platform at the moment. Now we're going to start getting to the following abilities.


Walking forward and backward in a straight line.
Maintaining a heading of travel.
Turning the platform.
Walking side to side (strafe-ing). 
Using the front legs as manipulators. 

We are some-what more experienced in things like weight transfer at the moment, so we expect that we'll be walking in a line by the end of the week. At least, that's the goal.


#updates #omgrobots #chip #chip_in_for_good #yay #wow #imu_happenings

0 views0 comments
© copyright 2019 | aditya mehrotra