Automatic Control of Wall-following Mobile Robots Based on Machine Learning

Mobile robots have become increasingly popular in various industries and applications due to their versatility and ability to perform tasks safely, effectively, and independently. They can operate in hazardous or inaccessible areas, thus reducing the risk of injury to human workers. Mobile robots can also work continuously without breaks, thus increasing productivity and reducing labor costs. Wall-following mobile robots (WMRs) use sensors and algorithms to detect and avoid obstacles while maintaining a constant distance from them. They are instrumental in environments with narrow passages or limited visibility, such as underground tunnels or mines. This study used a benchmark dataset collected by the SCITOS G5 robot to evaluate various Machine Learning (ML) algorithms. This research guides on selecting an ML algorithm to achieve a balance between accuracy and memory consumption. The experiments showed that eXtreme Gradient Boosting (XGB) performed the best with a macro F1-score of 0.9934 on 24 sensor readings.