We provide a new lifelong robotic vision dataset (¡°OpenLORIS-Object¡±) collected via RGB-D cameras mounted on mobile robots. The dataset embeds the challenges faced by a robot in the real-life application and provides new benchmarks for validating lifelong object recognition algorithms.

Data Collection

Several grounded robots mounted by depth cameras and other sensors are used for the data collection. These robots are moving in the offices, homes, and malls, where the scenes are diverse and changing all the time. In the OpenLORIS-Object dataset, we provide the RGB-D video dataset for the objects.

The robot is actively recording the videos of targeted objects under multiple illuminations, occlusions, camera-object distances/angles, and context information (clutters). We do include the common challenges that the robot is usually faced with. For example,

• Illumination. In a real-world application, the illumination can vary significantly across time, e.g., day and night differences. We repeat the data collection under weak, normal, and strong lighting conditions, respectively. The task becomes challenging with lights to be very weak.
• Occlusion. Occlusion happens when a part of an object is hidden by other objects, or only a portion of the object is visible in the field of view. Since distinctive characteristics of the object might be hidden, occlusion significantly increases the difficulty for recognition.
• Object size. Small-size or elongated objects make the task challenging, like dry batteries or glue sticks.
• Camera-object angles/distances. The angles of the cameras affect the attributes detected from the object.
• Clutter. Clutter refers to the presence of other objects in the vicinity of the considered object. The simultaneous presence of multiple objects may interfere with the classification task.