Experimental Dataset for Functional Offloading in Robot–Edge Architectures over Private 5G Networks

This dataset contains the experimental results obtained for the temporal analysis of functional offloading in Robot–Edge architectures over private 5G networks. The experiments were conducted using a simulated Robotnik Summit XL mobile robot running a ROS 1 navigation stack under three deployment scenarios: centralized execution (experiment_centralized_qlearning_complete.csv), distributed execution with an emulated 5G network (experiment_distributed_qlearning_emulation_complete.csv), and distributed execution over a real private 5G network (experiment_distributed_qlearning_upv_complete.csv). The dataset consists of three CSV files, one for each deployment scenario. Each row corresponds to a navigation goal and includes the timestamps of the main events along the navigation pipeline: goal reception (t_goal), velocity command generation (t_cmd), command arbitration (t_cmd_out), odometry update (t_odom), and localization update (t_amcl). These timestamps enable the computation of end-to-end and subsystem-specific latencies. The scripts used to generate and analyze the dataset, together with additional documentation, are available in the associated GitHub repository.