This facility is a large room (~10x10x5.5 m volume) equipped with an array of Optitrack motion capture cameras and an accompanying software. The system allows to track up to 50 rigid objects with a position error less than 1 mm and with a frame rate of up to 200 fps. The cameras illuminate the scene with IR light so the use of special passive markers on the tracked objects is necessary. Most of the wall surface is covered with protective net. Initially designed for the experiments with flying robots this arena can be used in many other robotic applications. AR drone is available on site for experiments which is equipped with markers and with open source flight controller.
Pattern formation algorithms for swarms of robots can find applications in many fields from surveillance and monitoring to rescue missions in post-disaster scenarios. Existing algorithms that enable complex configurations usually require a centralized control, a communication protocol among the swarm in order to achieve consensus, or predefined instructions for individual agents. Nonetheless, trivial shapes such as flocks can be accomplished with low sensing and interaction requirements. We propose a pattern formation algorithm that enables a variety of shape configurations with a distributed, communication-free and index-free implementation with collision avoidance. Our algorithm is based on a formation definition that does not require indexing of the agents. The aim of this experiment is to test in a real scenario formation control algorithms that have been developed theoretically and tested through computer simulations or basic experiments with ground robots.
This experiments are part of a project to develop a swarm of drones capable of helping in search and rescue (SAR) operations by finding people in large water bodies (sea or lakes). Research outcomes from this project will directly contribute toward next-generation autonomous rescue operations, which have a big potential in saving human lives.
The proposed research concentrates on developing a simple and applicable method to tune controllers, e.g., PD, widely used both in commercial and open-source flight controllers (like Pixhawk, Naze32, CC3D Open Pilot) in a matter of minutes. However, due to a variety of UAV applications, it is important to perform tuning in flight conditions, without dynamics modeling of the drone, using iterative approach. Literature studies reveal no results for iterative controller autotuning performable as above, during flight. The OPTIM-TUNE will perform automated tuning of controllers, with special view to UAV control robustness against changes in propellers’ aerodynamic efficiency. OPTIM-TUNE could also be used to find the best model parameters used in the approaches adopting partial knowledge about the UAV dynamics.
The proposed research is a natural extension of the prior common actions, making it possible to verify the theses previously presented, and develop new software tools for optimal auto-tuning of controllers with special attention paid to robust, fault-tolerant, control, or different degree of effectiveness of the propellers (Tilt-hex). The robotic laboratory, to which the TNA is addressed released recent publications related to such problems, as the one concerning Tele-MAGMaS: Aerial-Ground Co-manipulator System, or about drag-optimal allocation method for variable-pitch propellers.
The project objective is verification of the proposed control algorithms on 5 nonholonomic mobile robots using OptiTrack motion capture system. Conducted experiment would allow adjustment of the control parameters to achieve better performance of the swarm movement. Said control algorithms are: virtual spring damper mesh control, distributed PD control, artificial potential control, worm creep algorithm . All algorithms would in theory allow for swarm movement in environment with and without obstacles.
The project allowed for verification of 2 control algorithms on 5 nonholonomic mobile robots using OptiTrack motion capture system : virtual spring damper mesh control with and without obsticles and swarm selforganization using worm creep algorithm. Conducted experiment allowed for adjustment of the control parameters to achieve better performance of the swarm movement.
The proposed research concentrates on the use of machine learning algorithms for the measurable improvement in quality and safety of autonomous flights of cargo drones for last-centimeter (<5 km) person-to-person delivery.
Objective 1: Conducting extensive experimental tests in the Laboratory of Intelligent Systems with the use of in-flight real-time model-free minimum-seeking auto-tuning method (based on the golden-search algorithm) to obtain sets of optimal position controller parameters for the cargo drone; in further stage, a bank of controllers should be obtained to deploy safe, precise, agile and robust flight controllers in a presence of wind gusts based on selected cost functions,
Objective 2: The proposed research at the Laboratory of Intelligent Systems (using a wind tunnel) to optimize the autonomous landing process of the cargo drone, for various wind gusts conditions, to obtain minimum landing time or minimum energy effort solutions.
The proposed research will allow to develop new solutions for process optimization of auto-landing and safe, precise, agile and robust point-to-point flights of cargo drone for last-centimeter delivery. Solutions from the areas of control theory/machine intelligence/optimization techniques should potentially result in gaining new knowledge and extend state-of-the-art.