Grant: Discovery Grants Program - Individual

Autonomous robots and drones are now developed for providing services for many industrial sectors and applications including goods delivery, an inspection of complex infrastructure, search and rescue and aerial photography.  Small drones have issues carrying long-range sensors and are more suitable for indoor settings or short-range outdoor missions. However, drone technology is also expanding the scale of its operation using vertical take-off and landing (VTOL) type helicopters. Application of them requires those vehicles to be equipped with multiple sensors having long-range sensing capabilities and they are typically employed for long-range outdoor missions. On the other hand, a micro aerial vehicle (MAV) can maneuver in highly cluttered and complex environments and can operate with or without GPS. One of the main factors for safe operation is sensor reliability. Loss of GPS signal, featureless conditions, and operating under no lights, can lead to failure of the navigation system; hence reliable operation requires a multi-sensory based navigation approach. In this DG cycle, we propose three distinct research objectives and themes to facilitate the operation of MAV under different environmental conditions. The first research objective develops Ultra-Wide band (UWB) sensor-based ranging and self-localization of robots operating in complex in-door environments. One of the key advantages is that range-based estimations can provide drift-free localization as opposed to other vision or Lidar-based methods. In the second objective, this research proposes multi-sensory based navigation solutions for the long-range operation of VTOL vehicles. First, this research develops reliable aerial data while using a visual, Lidar, odometer and mapping (VLOAM) sensor on an outdoor VTOL vehicle. VLOAM provides a data processing pipeline for online navigation and builds a map of the traversed environment, utilizing data from a 3D Lidar scanner, a camera, and an inertial measurement unit (IMU). Visual inertial navigation systems (VINS) are enhanced while incorporating AI-based feature extraction solutions for long-range operations. In the third objective, this research proposes robust trajectory control of MAV using nonlinear model predictive control (NMPC). Multi-robot operation requires the robots to navigate under constraints and NMPC has the unique ability to control while accommodating constraints. The issue of computational complexity is removed through novel NMPC control structures and we will design fast NMPC for aerial vehicles to operate on faster trajectories. The program expects to provide 5 doctoral and 4 masters level HQP training. Many industries including aerospace and ocean industries in Canada are now attempting to use autonomous vehicles for many sub-sea and arctic missions. Robotics and its related technology are now demanded by many industries, that includes aerospace, ocean and oil and gas industries in Canada.