Grant: Engage Grants Program

Inuktun Services, an all-Canadian company located on Vancouver Island, British Columbia, is a complete solution provider for industrial service robotics. They build a wide range of tethered robots that can be remotely controlled via a ground control station, as well as inspection tools such as high-resolution Pan-Tilt-Zoom (PTZ) cameras. The multi-tasking required in operating these robots can distract the user when navigating them through difficult passageways. Furthermore, given the importance of the time-to-operate, especially in search-and-rescue missions, quick retrieval of the robot and/or changing its course of motion to find open and safe passageways can be a very challenging task. The objective of this research project is to develop a robust algorithm for simultaneously localizing a robot, equipped with onboard imaging sensors, and mapping the environment with applications in: (1) automated travel through a pre-specified trajectory, and (2) automated retrieval of the robot at the end of its mission. In technical terms, this is categorized as a Vision-based Simultaneous Localization and Mapping (VSLAM) problem. Deployment of the forward- and down-looking imaging sensors for VSLAM will be investigated. The research targets the sensor fusion for precise VSLAM when using redundant but non-orthogonal information obtained via multiple imaging sensors and also the implementation of a graphical user interface into the final product. Two major foreseeable challenges in using VSLAM are: (1) real-time operation, and (2) dealing with environments where visual features remain very similar throughout the motion. Incrementally building a persistent and sparse map of the environment would burden the CPU exponentially over time. However, without having a persistent map, the motion estimation of the robot tends to drift over time. Real-time operation of VSLAM in moderately-sized environments is the problem to be investigated by the applicants and their team.