0

I have a start and end point and several obstacles (more or less depending on the situation). I also have a robot-like system that moves in a dynamic environnment. My job is to implement an algorithm that will allow my system to arrive to the end-point avoiding the obstacles. Has anyone worked on path/motion planning before and could give me some advice ? So far I've done a little bit of 'research' and I found that basically there are two groups of algorithms: serach-based and sampling-based. Apprently the sampling-based ones are more suitable for environments having multiple obstacles and they are also mostly used in robotics. Now testing they give a very strange trajectory, having many turns which is not at all what I want. On the other hand, search-based algorithms since they work with a grid, when we have more obstacles take a lot of memory. So, here is where I'm stuck. I was thinking of implementing RRT*, but I really don't like the path found even when I have many iterations.

I looked up algorithms like A*,D*,DLite,RRT,RRT and PRM.

Unicorn
  • 29
  • 1
  • 1
  • 3

1 Answers1

1

As you have mentioned, sampling-based methods are a lot more efficient in problems with very high-dimensional state space and you don't have to worry about grid size. The observation of strange and unsmooth paths returned by a sampling-based planner is common. Usually, people perform post-processing to fix this. You can perform shortcutting (see paper: https://ieeexplore.ieee.org/document/5509683), or you can throw the found path into an optimization-based solver (e.g. trajectory optimization) as a good initial guess that takes into account path smoothness as well.

Steven Zhu
  • 11
  • 1