From cadf6d8a7096511735334f3f4ef2c3450a5dbbea Mon Sep 17 00:00:00 2001 From: Patrick Musau Date: Wed, 5 May 2021 17:07:55 -0500 Subject: [PATCH] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index e2f917a..36620d9 100644 --- a/README.md +++ b/README.md @@ -23,7 +23,7 @@ Rapidly exploring random trees (RRTs) are a family of algorithms designed to eff RRT algorithms are great at finding paths within the robot's configuration space. A lot of these approaches assume that two nearby points (or configruations) can be connected by a "straight line". For robots that can turn in place (holonomic robots), this is a valid assumption (ignoring the size of the robot for now). In this case the restrictions are mostly on the set of allowable configurations within the configuration space. However for robots that have wheels (This is us!!), this assumption doesn't quite cut it. Most wheels are are not designed to slide sideways. This imposes constraints on the vehicle referred to as non-holonomic constraints. Interestingly enough this actually a velocity constraint and I'll leave it to Dr. [Lavale](http://lavalle.pl/planning/book.pdf) to explain that (page 370). Essentially however what this means is that we have minimum turning radius and in order for our paths to be feasible we need to take this constraint into account. RRT approaches that handle such constraints typically make some assumptions about the speed and angle of the vehilce as it moves through the configuration space and the most popular variations use a tricylcle model, a simple car, a reeds-shepp car, or a dubins model. The below implementation makes use of a reeds sheep model to generate plans and it was designed for car like robots. The reed sheep curves can be generated by a set of motion primtives (turning left, moving straight, turining right) in the forward and backward directions. A more in depth description can be found [here](http://www.cds.caltech.edu/~murray/courses/me132-sp11/me132b_lec06.pdf). -Anyway enough of that here is the ROS package. The rsband_local_planner combines an elastic band planner, a reeds shepp planner and a fuzzy logic based path tracking controller, to achieve reactive local planning for Car-Like robots with Ackermann or 4-Wheel-Steering. Originally proposed by [George Kouros](https://github.com/gkouros/rsband_local_planner). Here are two gifs of how this planner works: +Anyway enough of that, here is the ROS package. The rsband_local_planner combines an elastic band planner, a reeds shepp planner and a fuzzy logic based path tracking controller, to achieve reactive local planning for Car-Like robots with Ackermann or 4-Wheel-Steering. Originally proposed by [George Kouros](https://github.com/gkouros/rsband_local_planner). Here are two gifs of how this planner works: ![Short Term Planning](images/short_term_planning.gif)