TITLE:
A Comparative Implementation Benchmark of Gazebo and Unity for xArm Lite6 Robotic Grasping Simulation
AUTHORS:
Haotian Yang, Yijia Zhang
KEYWORDS:
Robot Simulation, Gazebo, Unity, xArm Lite6, Robotic Grasping, Digital Twin, MoveIt, ROS
JOURNAL NAME:
Journal of Computer and Communications,
Vol.14 No.9,
September
30,
2026
ABSTRACT: Robot simulation platforms are widely used for motion planning, control validation, grasping-task prototyping, and digital-twin demonstration, but platform comparisons are often based on qualitative descriptors. This paper presents an implementation benchmark of Gazebo and Unity using the same xArm Lite6 six-degree-of-freedom robot and a six-stage pick-and-place task. Gazebo uses ROS Noetic, MoveIt!/OMPL, joint controllers, gazebo_ros_control, and gazebo_grasp_fix; Unity uses the URDF Importer, Articulation Body joints, staged C# targets, and trigger-based parent-child attachment. The revised evaluation reports auditable indicators rather than unsupported impressions, and it explicitly separates completed implementation evidence from prospective repeated-trial measurements. Both implementations completed all six task stages (6/6). In the implemented prototypes, Gazebo scored 1 for online planning and 1 for controller-chain execution, whereas Unity scored 0 for both because its trajectory was scripted. Both scored 1 for deterministic attachment dependence and 0 for physics-preserving grasp transport. The results therefore support Gazebo for planning/control-pipeline validation and Unity for scene-level interactive reproduction, while neither implementation supports direct claims about physical grasp robustness. The paper further analyzes Sim-to-Real failure mechanisms, specifies a controlled Gazebo ablation, and documents implementation failure conditions.