Reinforcement Lerning-Based Overhead Crane Control for Handling in Sensor-Failure Scenarios

Overhead cranes are a key component of many industrial processes, enabling the rapid and safe transportation of loads. An important aspect of their operation is reliability, including fault tolerance of components such as sensors. In the event of a sensor failure, the control system must be able to continue operation, using only the available information about the object's condition. This paper presents a control system for a bridge crane based on reinforcement learning, taking into account the object's state estimates. The system was designed to be robust to sensor failures while maintaining high control quality. In simulation studies, the effectiveness of control with different state estimation accuracy was investigated. The results showed that the system is able to ensure stable and safe operation of the crane, even in the case of errors in state estimation.