Motion planning is an essential part in robotics domain; it is responsible for guiding the robot motion toward the goal. It generates a path from one location to another one, while avoiding the obstacles in the way. The planning modules could be configured to check the optimality, completeness, power saving, shortness of path, minimal number of turn, or the turn sharpness, etc., in addition to path safety. In this paper the cell decomposition approximation planar is used to find a safe path; the quad-tree approximation algorithm divides the workspace into manageable free areas, and builds a graph of adjacency between them. New methods are proposed to keep the robot far away from the obstacles boundaries by a minimum safe distance. These methods manipulate the weights of adjacency graph's edges. They utilize and reflect the size of free cells when planning a path. These approaches give a lower weight to the connection between big free cells, and a higher weight to the connections between the smaller cells. The planner after that searches for the lowest cost path based on these weights. The safe path in this work is the path which keeps the robot far away from obstacles by specified minimum safety distance and it bias the robot's motion to follow the bigger areas in the workspace. The shortest path is not considered. However a tradeoff between the real path cost and the safe path cost is considered when choosing the weight values.