While households in developed countries receive skyrocketing data
rates through optical fibers and smartphones step into the 5G era,
roughly one half of the world’s population cannot connect to the
internet. Even beyond developing economies, bringing data
connectivity to areas where it cannot currently reach would
drastically benefit applications such as the internet-of-things, smart
agriculture/forestry, wildfire suppression, search-and-rescue
missions, paramedical interventions, and emergency response handling
to name a few.
To this end, the UAVs in the targeted technology are equipped with a communication module that connects to the ground users on one side and to the cellular terrestrial infrastructure on the other side. The user information may even be relayed through multiple UAVs before reaching its destination. For this technology to be viable, the UAVs must be able to navigate without human supervision to locations with favorable propagation conditions, that is, where the signals that they receive from and transmit to the ground users and cellular infrastructure are not significantly blocked by obstacles such as buildings or mountains.
The key approach in this project is to construct radio maps that describe the propagation conditions in a certain region. Using these maps, the UAVs rely on artificial intelligence algorithms to determine the appropriate locations and can even adapt to changes in the user positions and connectivity requirements as well as to coordinate with other UAVs. Our preliminary findings already showcase the ability of artificial deep neural networks to construct radio maps from a small number of measurements collected by the UAVs.
By 2030, myriads of unmanned aerial vehicles (UAVs) will pervade the skies of populated areas, serving millions of people worldwide for transportation of goods, construction, agriculture, surveillance, and search-and-rescue operations to name a few. Human life will experience a profound transformation since daily tasks such as food delivery or grocery shopping will be carried out by autonomous UAVs. But before the era of UAVs can set in, a number of technical and legal challenges need to be addressed, mainly due to the safety concerns that flying vehicles pose for citizens. Technical challenges include low-altitude flight technology, sense-and-avoid capabilities, handling lift-offs and landings in urban areas as well as the initial and final meters of the trajectory, prioritizing traffic such as public service UAVs (e.g. fire fighters or police drones), and so on.
To address these challenges, the primary objective is to develop an integrated technology of communication and tracking for low-altitude autonomous operations in populated areas. Besides contributions to UAV communications as well as tracking of low-altitude flying objects, a major novelty of the research plan is to cross-fertilize these two areas towards mutual performance enhancements. In other words, a central hypothesis is that location can improve performance metrics of spectrum-cognizant communications, whereas communications may facilitate tracking of UAVs operating in low altitudes.
Over the years, demand for intelligent robots with high mobility across different environments has risen dramatically. Intelligent collaborative robots play a increasingly significant role in challenging and high-risk operations such as inspection, surveillance, search and rescue (SAR), etc. The exceptional abilities (i.e., agility, robustness and stability) of biological snake to traverse complex uncertain terrains, and their adaptability to an astonishing variety of environments, inspire researchers to develop snake robots for use in complex, dangerous, and challenging applications. Due to the complicated interactions between the snake robot and the adjacent cluttered environment, the development of efficient locomotion is known to be challenging and it has only been achieved in controlled laboratory setups so far. Moreover, most of the previous works focus on specific scenarios with sufficient a priori knowledge of the terrain morphology and properties. Based on the current state of the art, the following knowledge gaps can be identified: i) efficient locomotion in uncertain environments is still challenging; ii) transition between different terrains (i.e., slippery mud, sandy trails, shale slopes, rocky ground, ...) has yet to be achieved; iii) flexibility and scalability to different tasks (i.e., deployment, collection, combination of locomotion and functions) is still an open problem. Furthermore, testing new control methods in a real setup environment is difficult because collisions may damage both the robot and the environment.
Based on the identified knowledge gaps, the objective of this work is threefold: