Towards Improving SLAM Algorithm Development using Augmented Reality

Simultaneous Localisation and Mapping (SLAM) is a popular map building approach in autonomous mobile robotics. Because users demand faster and more effective algorithms, SLAM remains an active area of research. However, the increasing complexity of applications, such as the environments the algorithm is applied to, makes it difficult to debug, evaluate and optimise such algorithms. Our preliminary research indicates that the algorithm development can be improved by using Augmented Reality (AR) systems, which visualise the robot’s internal program state and related information specifically in the context of testing and debugging SLAM algorithms. Using inherent SLAM uncertainties and error-sources identified in literature, we developed requirements which an AR system must fulfil in order to optimise the testing, debugging and design of SLAM algorithms.