Combined coverage path planning for autonomous cleaning robots in unstructured environments

Abstract

Complete coverage path planning is a key problem for autonomous cleaning robots, which concerns not only the cleaning efficiency but also the adaptability to unstructured environments. But the diversity of environments and limited perception ability of the robot make the problem still unsolved. In this paper, a novel strategy of combined coverage path planning is proposed, which combines the random path planning and local complete coverage path planning. The random planning lets the robot run straight until an obstacle is encountered. After turning a random angle, the robot continues the straight run. This mode is easy to implement and can provide the robot with the flexibility to environments. And local complete coverage path planning works out a comb-like path depending on dead reckoning. The comb-like path can cover every part in a relative small area. All these functions are just based on general hardware: ultrasonic sensors, infrared sensors, incremental encoders, DC motors, vacuum, etc. Finally the experiment shows that this strategy can work efficiently and robustly in common family environments.