NAMOSIM: a Robot Motion Planner for Navigation Among Movable Obstacles

A NAMOSIM scenario with one robot and two obstacles is visualized in RViz.The robot's plan is shown by the blue lines.The darker shade of blue indicates the part of the path involving an obstacle transfer.The empty rectangles indicate the planned obstacle placements.The social costmap is seen in the rainbow background. SummaryNAMOSIM is a mobile robot motion planner and simulator designed for the problem of Navigation Among Movable Obstacles (NAMO).The planner simulates robots navigating in 2D polygonal environments where certain obstacles can be grasped and relocated to enable robots to reach their goals.NAMOSIM extends the classic navigation problem with a layer of interactivity, posing interesting research questions while remaining well-defined and amenable to various algorithmic approaches.NAMOSIM is intended for researchers and developers working on robot navigation in dynamic environments, particularly where physical interaction is necessary.NAMOSIM supports the development of custom NAMO algorithms using a modular agent-based architecture.It includes a baseline agent implementing Stilman's NAMO algorithm (Stilman & Kuffner, 2005) and incorporating a communication-free coordination strategy for multi-robot scenarios (Renault et al., 2024).A variety of other agent behaviors are implemented, and new