CHIP is a 6D pose estimation dataset for robot manipulation in real-world industrial scenarios, released in 2025 by FBK-TeV in collaboration with Ikerlan and Andreu World. Related papers include... CHIP: A multi-sensor dataset for 6D pose estimation of chairs in industrial settingsIt aims to fill the data gap in existing benchmarks, which mainly focus on household objects and laboratory settings and lack real industrial conditions.

This dataset contains 77,811 RGB-D images, covering chair models with seven different structures and materials. The images were acquired by a robotic arm in a real industrial environment using three types of RGB-D sensors: Intel RealSense D435, RealSense L515, and StereoLabs ZED. The 6D pose annotations for all chairs were automatically generated by the robot's kinematics system, with an average of approximately 11,115 high-precision annotations per chair type. The dataset also includes scenes with varying degrees of occlusion, including moderate to severe occlusion caused by the robotic arm and human operators, accompanied by fine-grained interfering objects, fully reflecting the complexity of industrial environments.