Assessing the duration of intralogistics forklift operations via machine learning

In the ongoing transition to Logistics 4.0, humans and technologies are increasingly interacting in operations systems. An example is the use of wireless devices for operation logs in the warehouse management systems. The performance and quality of such systems depend on human behaviors to a noticeable extent. Valid data from compliant operations provide great value for follow-up researches, while in actual operations deviant behaviors occur and contaminate the data. Quantitative studies on whether and to which extent operators do infringe or fulfill the organizational norms in the execution of digital workflows are limited. To close this gap, in this work we conduct a data-driven assessment of the duration of forklift operations in a German warehouse owned by a grocery retailing chain. By predicting the execution time of forklift picking tasks with given features, we study the proportion of standardized operations performed by different operators based on predicting accuracy. This could serve as an effective indicator before further developing systems using machine learning technologies for better distribution of the tasks based on the relevance of the operators' skills and the tasks' characteristics.