MSLS (Mars Science Laboratory Simulation) final progress

[Lexem100]

you misspelt friend

[mpeniak]

@Lexem100 I've noticed since it's been here for 2 years already :D

[asparuh13]

nice!

[yebbey]

How did you test the robustness? Did you e.g. randomly generate navigable terrains and see how many of them the rover was able to navigate through? Really nice in any case!

[mpeniak]

Hi,

Thanks for your comment. It was my BSc project so I did not get much time for testing :-)

However, the rover was evolved in many different terrain configurations and and in most cases the final controller was quite reliable. I am currently doing some more research on this as well as on using active vision with neural network systems.

[Teravus]

Nice work! Did you still set the velocities to set the wheel rotations?

Did you end up with a PID controller?

[mpeniak]

Thanks...

the output from the neural network is directly used to change velocities of the joints(wheels rotation rate and steering angle). No PID controller used though.

[kjlg74]

Fabulous!

Seeing it detect and avoid that hole in the ground on its own was great. Man, that thing has a lot of sensors, and its clearly making good use of them. Very cool!

Btw, what's the music? I love it.

[mpeniak]

thanks...the music is from Daft Punk and song is called Harder, Better, Faster, Stronger