search

carlosluis / multiagent_planning

Implementation of several multiagent trajectory generation algorithms
GNU General Public License v3.0
91 stars 23 forks source link

question about the sequential convex programming implementation #6

Open RandyChen233 opened 1 year ago

RandyChen233 commented 1 year ago

Hi,

The code is very helpful, and I'm just curious if it will be a more fair comparison if we run the CUP-SCP and dec-SCP both in a receding-horizon fashion to compare against the DMPC algorithm. If I understand correctly, the SCP algorithms are currently run in a fixed-length horizon

carlosluis commented 1 year ago

Hi,

Sorry for the late reply.

From the top of my head I don't know how to make cup-SCP and dec-SCP receding-horizon. Wouldn't that just end up being something akin DMPC? Since this is used for offline trajectory generation only, the receding horizon part in DMPC is just a more efficient way to do computations, rather than being a method for online adaptation of the trajectory.

As for fair comparisons, I still believe comparing these three methods is fair: your are solving the same problem and you want to compare different optimization strategies to solve it. If you have any new algorithm in mind to solve the point-to-point problem, then that would also be fair to compare against.

Hope this helps!