Closed zichunxx closed 1 year ago
The "independent" solver (GST_INDEP
) looks to me like a tentative effort and writing an FCL-only solver (one that doesn't depend on libccd). Our early tinkering with it left us feeling like the GST_LIBCCD
solver was more reliable. I don't know if that's definitively true, but I can say that based on that belief, all of the recent efforts to provide solver fixes have been solely focused on GST_LIBCCD
.
Feel free to try GST_INDEP
, if you find circumstances where it clearly provides superior results, please let us know.
1645774_smaple.json
1645774_smaple.json文件是我对fcl模型采样后的点云数据.
我在计算机械臂到外部环境距离的时候,使用fcl::GJKSolverType::GST_LIBCCD;发现最近点并不在建立的圆柱体表面,而当我将类型更换为fcl::GJKSolverType::GST_INDEP后,效果非常好.
这是使用fcl::GJKSolverType::GST_INDEP的结果(正确):
这是使用 fcl::GJKSolverType::GST_LIBCCD的结果(错误):
Hi!
Are there some instructions on the narrow phase solver
GST_INDEP
to know the difference betweenGST_INDEP
andGST_LIBCCD
?Help would be much appreciated!