Abstract
Systematic area coverage is a methodical approach to cover an entire space or area during a cleaning, inspection, or surveillance operation. Especially in floor cleaning, area coverage is considered a key performance indicator. However, traditional systematic area coverage approaches often focus on performing complete coverage with a model-based trajectory generation. They do not take the potential risks and hazards that may be present in the cleaning environment into account. As a result, floorcleaning robots are often challenged by the hazardous components in their environment which lead them to fail and consequently prevent them from performing their best. This situation has led to continuous research involvement in floor-cleaning robots. However, mostly, they are focused on improving the perception capabilities of the robots by incorporating more sensors which is an expensive solution for most cleaning robots. Risk-aware Coverage Path Planning (CCP), on the other hand, involves analyzing the cleaning space for potential hazards and creating a cleaning plan that minimizes the risk of accidents or damages. This research also brings up the Failure Mode and Effect Analysis (FMEA) as a tool that provides awareness of hazards present in the environment as an additional layer to the CPP process. Moreover, this research has used path transform and reinforcement learning methods as foundations to address offline and online path planning respectively. Here, a quantitative measurement defined for the risk level at a selected point in the environment is used as the key parameter to assess the safety of the path suggested by the algorithm. In order to validate and evaluate the proposed method, a modified version of the hTetro cleaning robot has been used. Thus, the results presented have proven the effectiveness of the proposed method.