search

team-fusionx / CarND-Capstone

Team FusionX - Capstone Project for Udacity Self-Driving Car Nanodegree
MIT License
8 stars 3 forks source link

Make car respond differently on Track 1 vs Carla & Track 2 #139

Closed teeekay closed 6 years ago

teeekay commented 6 years ago

Possible solution to have a smoother ride on Site while still accelerating fast enough in the simulator on Track 1. Use lower acceleration/jerk and slower creep when not on Track 1. Had to increase dyn_tl_buffer to 3.75 because car stops sooner than planned on Track2 when using lower deceleration. It just missed reaching zone where state would get set to stopped by 5 cm, so extended by 25 cm to compensate.

edufford commented 6 years ago

Also, the tuning might need to be readjusted again after changing the DBW throttle/brake gains for getting in to the stopping zone.

teeekay commented 6 years ago

I put this up to see what you think of this approach of different acceleration/velocity behaviors for sim track 1 vs. sim track 2 / site, and I wanted to see it in combination with the other change of no double stop. So if you want to continue on this path, we can discard the other PR. I put up the other PR in isolation in case you didn't want to go down this route.

Making the changes in the launch files makes total sense. I was keeping things isolated to the smallest footprint to demonstrate this change. If we proceed, I will change it to the launch file.

I concur on the need to revisit the tuning of the buffer zones once changes are made to throttle/braking gains.

edufford commented 6 years ago

Ah, got it. Since we are pretty close to achieving the target pattern, I’m thinking that it’s best to make only one or two changes each trial to dial it in so DBW gains + PR#138 may be enough for the next submission. This branch is good to keep as a backup for after that.

edufford commented 6 years ago

Since we were able to get a successful test result using the FX-v3.2 release, no need to pursue this change. Closing the PR.