We describe a novel approach for computing collision-free \emph{global}
trajectories for $p$ agents with specified initial and final configurations,
based on an improved version of the alternating direction method of multipliers
(ADMM). Compared with existing methods, our approach is n