The invention discloses a strip-shaped robot path planning method based on a self-learning ant colony algorithm. The strip-shaped robot path planning method based on the self-learning ant colony algorithm is characterized by comprising the following steps: step 1, environment modeling; step 2, initializing stage; step 3, initial searching; step 4, overall updating of grid map pheromone; step 5, self-learning searching; and step 6, outputting of a planning path. The strip-shaped robot path planning method based on the self-learning ant colony algorithm is greatly improved for an ant colony algorithm calculating process, a self-learning strategy is introduced, grid-process environment modeling is treated specially, by the used grid method, the ant colony algorithm deals with a strip-shaped robot path planning process under the condition that barrier cells do not need to be expanded, a new shortest path calculating method is provided, thought of machine learning is fused in the ant colonyalgorithm, the efficiency of path planning of the ant colony algorithm is improved by effective combination of methods such as pheromone, heuristic information, positive feedback and greedy search, and a strip-shaped robot can pass through a narrow channel according to the outline of the strip-shaped robot so as to implement the shortest path planning.