Please Note: Code it already written, it just needs few tweaks in 12 hour deadline or so. I need help improving a Python MuJoCo pick-and-place pipeline using a UR5e + Robotiq 2F-85 gripper (from mujoco_menagerie). The task: pick a red box and place it in a green bin while avoiding obstacles. The stack includes MuJoCo, ur-analytic-ik (analytical IK), and OMPL (currently RRTConnect) for collision-free motion planning, plus a joint-space PID torque controller and a state-machine execution policy. Your job: tune PID gains (including “loaded” gains when holding the box), improve OMPL parameters (setRange, checking resolution, goal threshold/tolerance, interpolation) and retry/replan logic, and fix any collision/exclusion edge cases (gripper↔box, box↔table/bin, etc.) to increase repeatability (target ~75%+ success over multiple runs). Deliver updated code/notebook + recommended parameter sets.