Alessandro Scoppio Email & Phone Number

The Hague, South Holland, Netherlands

Leading the Machine Learning team at Mainblades, and setting the vision of the company in terms of ML and Python algorithms. On top of ML engineers, I lead and oversee our internal team of data annotators.

The Hague, South Holland, Netherlands

Provide support for experimental test flights and train new pilots to use Mainblades inspection drone.

The Hague, South Holland, Netherlands

Part of the ML team at Mainblades, working on computer vision.Contributed to bringing deep learning based object detection technology to production in the context of aircraft inspections performed with a drone. Worked on data quality, developing and maintaining ML training and evaluation code as well as tooling.

The Hague, South Holland, Netherlands

Spent almost two years helping the Navigation team at Mainblades getting our inspection drone to production. Contributed on tasks such as autonomous flight, obstacle avoidance, business logic integration, UAV integration and communication with mobile devices, LiDAR sensing, odometry and mapping.

The Hague, South Holland, Netherlands

AbstractUnmanned Aerial Vehicles (UAV), commonly referred to as drones, are a class of vehicle that isgaining increasing popularity in the context of autonomous vehicles. Odometry estimation is thetask of predicting the movement occurred between two sensor observations to determine the positionand orientation of a vehicle, a notoriously challenging problem. Inertial Measurement Units (IMUs),cameras and 3D LiDAR are some of the most commonly used sensors to aid the task of… Show more AbstractUnmanned Aerial Vehicles (UAV), commonly referred to as drones, are a class of vehicle that isgaining increasing popularity in the context of autonomous vehicles. Odometry estimation is thetask of predicting the movement occurred between two sensor observations to determine the positionand orientation of a vehicle, a notoriously challenging problem. Inertial Measurement Units (IMUs),cameras and 3D LiDAR are some of the most commonly used sensors to aid the task of odometryestimation. Recently, the success and rapid development of deep learning has popularized newlearning-based paradigms to solve perception related tasks such as image recognition and objectdetection. Specifically, lot of effort has been devoted in finding effective representations to process3D data with neural networks efficiently. In this Thesis, an end-to-end learning-based odometryestimation framework is proposed, tailored to the context of autonomous visual inspections ofaircraft. The method proposed utilizes sensor specific neural architectures to extract ego-motionfeatures from IMU and LiDAR sensors, with strategies to fuse such signals and learn to regressposition and orientation of an autonomous drone. Moreover, to train aforesaid networks in asupervised manner and evaluate their performances, a simulation environment is proposed, whichis used to generate a domain-specific datasets correlated with simulated sensors observations andground truth annotations. Experiments conducted validate the proposed method and show thatit is possible to approximate a drone’s position and orientation by exploiting Recurrent NeuralNetworks (RNNs) and 2D spherical projection of 3D points clouds. An extensive evaluation ispresented, and a comparison of different sensor modalities shows that the proposed method is ableto decrease the prediction error my leveraging each fused modality. Show less

The Hague Area, Netherlands

Researching data augmentation and Neural Architecture Search techniques for improving image recognition models.

Bari Area, Italy

Developed python parser to optimize and automate financial operations

Bari Area, Italy

Managed and coordinated a team of drivers for efficient and cost effective deliveries

Bari Area, Italy

- Carried out bachelor thesis during a six-month internship at the Italian National Research Council in the robotics lab- Developed software for automatic detection of safe landing areas for drones using 3D point clouds- Individual and self-organized research work under the supervision of two researchers

Master Of Science - Ms, Artificial Intelligence

Bachelor Of Science - Bs, Computer Science And Technologies For Software Production, 110/110 Cum Laude

High School Diploma, Science, 82/100