Targeted and cost-efficient solar cleaning: A novel RTVA approach to robotic path planning

The efficiency of photovoltaic systems can be significantly compromised by surface soiling, leading to substantial energy losses. While autonomous cleaning systems are increasingly deployed, most existing approaches rely on uniform surface cleaning or time-based scheduling, often ignoring the uneven distribution of contaminants. In this study, a contamination-aware robotic cleaning system is developed, integrating real-time image processing, environmental sensing, and energy-efficient path planning. The system identifies soiled regions through vision-based analysis and schedules cleaning actions according to environmental conditions. A novel routing algorithm, the Recursive Tabular Validation Algorithm, is introduced to solve the Traveling Salesman Problem in the context of partial and targeted cleaning. Unlike conventional heuristics that optimize the visit order, this algorithm prioritizes edge selection and employs a recursive table validation mechanism combined with path-intersection analysis to eliminate suboptimal routes. The prototype comprises four 25 W photovoltaic panels, and scalability simulations were conducted to evaluate algorithmic performance under larger problem sizes. Experimental results demonstrate a 19 % reduction in cleaning energy consumption, with the proposed algorithm outperforming the Nearest Neighbor and the Nearest Neighbor +2-Opt combinations by 17 % and 15 %, respectively. By prioritizing energy-aware route planning and condition-based cleaning, this study introduces a novel contribution to the field of autonomous PV maintenance systems and offers a scalable solution for enhancing the operational efficiency of solar farms.