dear all, i'm doing arm navigation for a 7DOF robot arm plus a 3DOF base. For the planning group containing only the 7DOF arm, both ompl and trajectory filter server work well. For the planning group containing the whole 10DOF, only the ompl works. the trajectory filter server always fails with error std::bad_alloc. It will also consume 100% CPU and make the whole computer hault for a while. when i check this in detail, i find that for the 10DOF case, when the input trajectory contains less than 8 points, the trajectory filter server works. when the points number exceeds 10, it always fail. has anybody tried the filter server for high DOF robots?

p.s. for the 7DOF case, even the input trajectory contains more than 100 points, the filter server works well. so i think my computer is not a problem (intel i5 CPU, 4G memory)

i'm using Fuerte on Ubuntu12.04.

For the trajectory filter server, i'm using unnormalizetrajectory+cubicsplineshortcutter_smoother