In recent years, aerial robotic inspection has been increasingly deployed in safety-critical infrastructure. Autonomous robotic inspectors can protect human workers from difficult-to-access and harsh areas. Moreover, they can provide cheaper and more accurate inspections. The autonomy challenge requires onboard localization, reactive mission planning, and remote real-time access to onboard sensory data. In the context of the EU H2020 project AutoASSESS, an autonomy stack has been developed for the fully actuated aerial inspector developed at RAM, including the requested features assuming LiDAR equipment.
This capita selecta will extend the existing stack by introducing the camera as an additional perception modality (1) in the simulator, (2) in the user interface, and (3) in the localization pipeline. This will lead to (1) enhanced remote control via visual input, (2) improved localization performance (within the project scope), and (3) visual inspection (as a possible extension).
