Novel fixed-length decimal encoding mechanism is proposed, which is for techniques of path planning of mobile robot based on an improved genetic algorithm. Effective path grids are adopted to identify the robot's motion surrounding. The obstacles can be described as polygons based on the given grids. The only vertexes of every obstacles polygon have corresponding unique decimal series value by the intersection of the grids. Then the planning paths of the mobile robot are encoded into decimal encode string, which equals to the sum of the numbers of all effective vertexes of polygons in path grids. In the chromosome, the decimal value of each nonzero bit is the code of vertex about corresponding obstacle for the path planning. The sequence of vertexes in a chromosome is that in the planning path. Such an encoding manner has the excellent adaptation for fixed-length decimal encoding mechanism; it also easily overcomes the trap of obstacles in path planning algorithm. The algorithm makes the path planning more simple and effective, and speeds up convergent speed of the algorithm. Numerical simulation to the problem shows the efficiency of the proposed algorithm.