Research

Current research:

• Motion planning - we are very interesting in motion planning for aerial and ground mobile robots and manipulators executing diverse tasks that include several constraints (obstacles, wind, kinematic and dynamic). Our current contributions include continuous vector fields for several applications (landing, manipulation, robot cooperation), vector field learning (for individual robots and swarms of robots), optimal sampling-based planning approaches (new sampling strategies, parallelization, constraint integration, applications), and optimization based planning (GAs, traditional optimization).
• Drone-based inspection - we are applying commercial and in-house-built drones for inspection of mines and dams. In our research, we create autonomous drones that use several sensors (i.e., thermal and color cameras, lidars) to help the inspectors to detect hazards in several structures of mines and dams (i.e., pillars, embankments, spillways, etc). Our goal is to simplify and speed up the inspection process to make is more frequent and precise. 
• Space exploration - we are currently developing algorithms to control groups of aerial robots that will explore the atmosphere of Venus. The aerobots are required to fly for several days in constrained altitudes to survive the extreme conditions of the planet.
• Agricultural and forestry robots - our goal is to introduce intelligent robotic technologies in the general areas of agriculture and forestry by establishing effective strategies for designing, programming, and deploying such robotic systems. 
• Parallel and Cloud computation for long term robotics - we explore on-board parallel computation using CUDA to speed up motion planning and the large amount of memory and parallelism of the cloud, in our case AWS, to create large and detailed maps.
• Cooperation between ground and aerial vehicles in mapping and exploration missions - aerial vehicles can expand the field-of-view of a ground robot. On the other hand, a ground robot can supply power and computational resources to the aerial vehicle. We explore these benefits to form a team that is able to explore unknown environments, including mines and caves.

• Tether-powered drones - although very intelligent, current drones have a very limited endurance. In this research we allow the vehicles to fly 24/7 buy keeping them plugged to the power outlet. Sophisticated motion planners and control algorithms will then guarantee that the drone's power cable will not tangle on objects in the environment.  

• Landing drones on moving platforms - future applications will require that the drones land on moving platforms. A delivery drones can, for example, get a ride on the top of a truck to save energy while moving between two cities. In this research we investigate motion planners and perception strategies for precise landing of drones in unstructured and instrumented platforms.

For past research, please, visit my YouTube channel: