IndoorMCD: A Benchmark for Low-Cost Multi-Camera SLAM in Indoor Environments

Abstract

Navigating mobile robots within home environments is essential for future applications, e.g. in household or within the field of elderly care. Therefore, these systems, equipped with multiple sensors, have to deal with changing environments. This work presents the IndoorMCD dataset that allows for benchmarking SLAM algorithms within static and changing indoor environments of various difficulties. The dataset provides synchronized and calibrated RGB-D images from a low-cost multi-camera setup, as well as additional IMU data. Further, highly accurate ground truth movement data is provided. It is the first dataset that provides static and changing environments for a multi-camera setup. Evaluations with state-of-the-art SLAM algorithms show the dataset's applicability and also present limitations of current approaches. The dataset is made available in a structured format and a utility library with example scripts is provided.