search

gramaziokohler / integral_timber_joints

Robotic Assembled Timber Structures with Integral Timber Joints
MIT License
9 stars 0 forks source link

Round 2 TAMP Test for Hypar #114

Closed yck011522 closed 2 years ago

yck011522 commented 3 years ago

While we are pretty certain that Hypar is ready for timber order, we can still run a second round of planning test (without smoothing) after the geometry had been moved 100mm higher.

yck011522 commented 3 years ago

seq_n 25 is very hard to plan and so far with 2 hours of attempt, no luck. This is always getting stick at RoboticFreeMovement(#A346_M0, Free Move to bring Beam ('b25') to assemble_approach position on structure., traj 0)

Yes, it is a long beam, but this looks okay reachable (see below). Does not even look like it will need the linear buffer segment I suspect it just needed a few extra rrt iterations to find the trajectory.

Therefore I exposed the rrt_iteration options in #115 to run.py and will try to increase that from 400 to 1000 in my further studies to see if it can overcome "hard" paths.

2021-09-29 15_24_00 2021-09-29 15_24_07

yck011522 commented 3 years ago

Related to #120.

yck011522 commented 3 years ago

Planning is completed in about 12 hours (5be0173fbc84d9254843f6c8d6cf84c694564be8) after @yijiangh last update of the planner. All the motions are planned, currently waiting for #122 to be resolved so we can check the config jump and polyline collision.

yck011522 commented 3 years ago

A more vigourous over-night planning config:

p -m integral_timber_joints.planning.run --problem CantiBoxRight_process.json --design_dir 211010_CantiBox --reinit_tool --load_external_movements --write --smooth --solve_timeout 21600 --rrt_iterations 1000 --seq_n 30
yck011522 commented 2 years ago

Currently blocked by #144 and #145