Usage of drone is increasing exponentially. Our vision is to use them to delivery rare medical supply, moreover
we aim at safe flight endowing drones with artificial vision in order to avoid people and obstacles which are not visible giving GPS coordinates.
This is a spinn-off of the African Institute for Mathematical Sciences, and it can also be integrated into existing supply chain system in particular for the last mile delivery.
Delivery of medicines to disconnected areas is our current focus but we are open minded to other opportunities.
We believe that this novel transportation mean can make the world a better place.
In collaboration with 
we follow Thesis projects and Master research projects in Ghanian Universities.
This thesis aims at applying some already known theory of face detection and face tracking to drones. Face detection is a machine learning field strongly based on statistics and digital image processing. While tracking mainly relies on system theories approach like the Kalman Filter and the particle filter. The Kalman filter model assumes the true state x at time k is evolved from the state at (k−1) according to x_k= F_k * x_{k-1} + B_k * u_k +w_k where F_k is the state transition model which is applied to the previous state x_k−1, B_k is the control-input model which is applied to the control vector u_k, w_k is the process noise which is assumed to be drawn from a zero mean multivariate normal distribution with covariance fixed. In the essay, it is expected to review these concepts and to use some Python code based on the OpenCV library to test them in the real world them by using the Parrot Drone available at AIMS. Already existing code will be given. This is a 50% theoretical and 50% applied work. Knowledge of the OpenCV library is a big plus. For a review on tracking the student is addressed to Kalman, R. E. (1960). "A New Approach to Linear Filtering and Prediction Problems". Journal of Basic Engineering 82: 35 .
Obstacle avoidance is desirable for lightweight micro aerial vehicles and is a challenging problem since the payload constraints only permit monocular cameras and obstacles cannot be directly observed. Depth can however be inferred based on various cues in the image. Prior work has examined optical flow, and perspective cues, however these methods cannot handle frontal obstacles well. Possible solutions are given by optical flow or template matching techniques. In the essay, it is expected to review these concepts and to use some Python code based on the OpenCV library to test in the real world them by using the Parrot Drone available at AIMS. Already existing code will be given. This is a 50% theoretical and 50% applied work. Knowledge of the OpenCV library is a big plus. For a review on tracking the student is addressed to Andrew Burton and John Radford (1978). “Thinking in Perspective: Critical Essays in the Study of Thought Processes”. and First Results in Detecting and Avoiding Frontal Obstacles from a Monocular Camera for Micro Unmanned Aerial Vehicles Tomoyuki Mori and Sebastian Scherer .
This project is revealing exiting. It is not just research, it is not just business, it is an adventure.
I am a biomedical engineer interested in technology which can have impact in our society. I am an experienced researcher and a lecturer at the African Institute for Mathematical Sciences in Ghana
I am a mathematician from Uganda, graduated from the African Institute for Mathematical Sciences in Ghana.
I am a software engineer, supporting the overall development of the project.
I am a postdoctoral researcher at the Department of Advanced Robotics/Humanoids & Human Centered Mechatronics at the Istituto Italiano di Tecnologia (IIT), working on the WALK-MAN project.
We are currently developing a system for automatically detect injured people lying in the ground. Being the control of the current drone used in the experiments in Python, and the deformable part model using the C++ library OpenCV. Download here.
It is possible to interface the code with the AR Parrot drone API, in C++ and Python (using the Boost Libaray)
More specifically, OpenCV library make available a wonderful, easy to use and optimized implementation of Deformable Part Model Cascade Detector ( Object Detection with Discriminatively Trained Part Based Models) with the DPMDetector class, but this is only available in C++. We used cvPy for easily converting OpenCV Mat objects to Python and therefore reusing OpenCV C++ code in Python (DPM in our case). cvPy have been tested under Manjaro 15.09 with OpenCV 3.0 and Python 2.7. cvPy requires Numpy and Boost.Python. Using the included Dockerfile you can automatically build an image with the required packages using ubuntu:14.04 as base image. You can download our implementation of cvPy from Github.
If you have ideas, you want to collaborate or you are interested
in the advertisement oppotunities, drop us few lines.
We are located in the highway between Accra and Cape Coast, near Anomabo, Ghana. GPS coordinates: 5.171861, -1.139812.
+39 333 745 1463
me_AT_alessandrocrimi.com
Capecoast, Ghana / Genoa, Italy
P. O. Box, DL 676 Cape Coast