Aerospace Engineering Graduate Colloquium, Prof. Manish Kumar, : Navigation of Drones in Unknown Indoor Environments – Challenges and Opportunities-

Navigation of Drones in Unknown Indoor Environments – Challenges Title of talk: Navigation of Drones in Unknown Indoor Environments – Challenges and Opportunities

Safe and reliable use of drones in indoor environments poses several challenges from technological perspectives when it comes to their beyond visual line of sight (BVLOS) operation. While autonomous navigation capabilities in drones in outdoor environment have been achieved (and available in commercially available drones), doing the same for indoor environment has been extremely elusive. The most fundamental challenge that has been preventing us from achieving this capability is carrying out localization, i.e., determining drone’s 3D location. In absence of GPS, that is widely used for outdoor environment, indoor drones need to rely on, often noisy, onboard sensors (such as vision, laser, ultrasound) to carry out localization. Additionally, drone’s operation in a complex indoor environment in proximity with humans and other obstacles leaves very little room for errors. Attempting to solve these problems using complementary and redundant sensors would result in higher payload, an unaffordable luxury for indoor drones, where payload constraints need to be tightly enforced to keep the drone size within limits. In this talk, I will present our work aimed at solving the above problems. Our overall approach combines multi-sensor fusion method for localization with navigation, path planning, and obstacle avoidance in a human-centered fashion to achieve a safe and reliable flight in indoor environments. Results from adaptive Kalman Filtering that combines data from LiDAR based 3D SLAM with IMU and another 1D LiDAR will be presented. By combining the Dynamic Window Approach with an Artificial Potential Field method, we achieve robust, collision-free navigation that supports fully autonomous operation while allowing human supervision when needed. We will share insights from two use-cases of indoor drones: telehealth applications and warehouse inventory management.