Thursday 16 December 2010

Updated Simulation

I have spent the last week updating and reconfiguring the Webots simulation of the Mantis. I have go to the point where I need to start testing some of the changes I am making to the walking algorithms and in particular the terrain adaptation. I was considering trying to get the 1/10 scale Mantis up and running again but decided to give the simulation another look to see if I could get the contact sensors running better and reduce the latency between Webots and the HexEngine. Currently the two are linked via TCP/IP and I am streaming data using UDP. I was thinking of writing a FACU emulation mode within the simulation that could be linked to the HexEngine via a 460Kbps serial link, which should reduce any latency further and guarantee packet ordering, but having fiddled around with the UPD code for a day or two while simultaneously adding feature to the model things are currently working pretty well!

New simulation model takes on a tough slope!
 I have added some very useful LED's to the model, one at each joint on the leg, one on the foot and one at the ankle. These give a great visual indication of joint limit warnings and errors, and contact detection respectively. I have also cleaned up the Webots controller code so I can switch on and off features using the keyboard instead of re-compiling.


I have two versions of the Mantis, one with terrain which has been really useful for testing my algorithms, and monitoring behaviour. Still getting contact latency between the simulation and the Engine, and the feet sometimes disappear below the terrain and get stuck! but generally the results are good, as you can see in the images, I'm pretty sure we won't have the power or grip to climb a lope as shown!

I don't thing the real Mantis could have climbed that slope!