Open-source implementation of IGP2, an integrated goal prediction and planning system for autonomous driving originally developed at Five. Implemented and validated prediction module over 2 datasets of real-world trajectories, and the prediction + planning modules in the CARLA autonomous driving simulator.

Development of indoor/outdoor capable multirotor craft with 1-1.5kg payload. Initial offering aimed to assist soldiers in urban areas. Technology was featured in The Times and Forbes.People & technical lead of a research and development team of 5 control, software, simulation and perception engineers. My key responsibilities and achievements include:- Implementation of Agile and SCRUM methodologies to carry out project management.- Completion of a comprehensive feasibility study on state-of-the-art sensing solutions for autonomous vehicles. Selection of the most adapted sensor pack for our applications.- Completion of a comprehensive a feasibility study on indoor flight stability. Assessment of the main criteria affecting indoor flight stability and choice of the propulsion configuration to achieve required performance while satisfying power and endurance requirements. Designed, implemented and completed performance evaluations of multiple technologies, as well as control and perception algorithms geared to improve flight stability.- Supervision of the development of a virtual environment allowing the full aircraft software stack to be run alongside a simulation of the craft in the world, using the Gazebo simulator. This included:• A physically accurate simulation of craft and its propulsion system. The sensors were modelled with representative noise profiles.• Simulation can be run on a host machine either in SITL (Software In The Loop, with hardware emulated) or HITL (Hardware In The Loop)- Supervison the development of a sandbox Simulink model of aircraft control systems and powerplant, allowing for the rapid prototyping of new control systems algorithms and strategies.

- Developed a low latency, high accuracy “follow me” autonomous behavior for a multirotor in the ROS and PX4 ecosystems, employing sensor fusion between an onboard camera and a 2D lidar. Solution delivered to the customer and used in military exercises.- Determined aircraft characteristics (drag coefficient, endurance, max speed …) from flight testing data and modelling methods.- Developed automatic, real-time dampening of IMU data and restart of VIO positioning algorithm under shocks, in order to prevent divergence of position states in the craft’s estimator. - Set up of multirotor autopilot and companion computer software, tuned control systems and estimator. Troubleshooted hardware and software malfunction from flight data and hardware inspection. - Live prototype demonstrations with clients and stakeholders, technical presentations of aircraft’s key capabilities.

- Designed implementation framework and software to implement a multi-agent path planning method to Kheperas and Kilobots robots, combining Model Predictive Control with Differential Game Theory.- Created an optimisation method to select metaparameters that would systematically lead to trajectories avoiding collision with obstacles or other agents, and preventing agent deadlock.- Validated the implementation on multiple real life scenarios, and demonstrated the optimiser's ability to improve on the theory by creating new types of trajectories achievable by the agents.- A paper presenting some of the thesis’ key findings was published in a reputed scientific journal.

Leading a team of 21 fellow students in designing a fixed wing autonomous UAV, with the goal of building capability for Imperial College to participate in UAV design competitions. Onboard camera employed computer vision to detect and identify drop-off locations. A global and local planner were designed to autonomously control the craft to fly towards the drop off locations and deliver a payload. Obtained financial support from the Department of Aeronautics (£1600), and a dedicated manufacturing and testing space. Received Exceptional Achievement Award from Department for building the project and team from scratch, and completing the journey from the white board sketches to the maiden flights. ICAV is now a yearly project held at the Aeronautics Department, with a new craft being designed each year. The award-winning team participates every year to the IMechE UAS challenge.

- Investigated pooling issues in an aerosol based medical device. Designed and tested a solution involving capillarity based evacuation channels. Conducted FEA analysis to validate structural integrity of system. Design completely eliminated pooling issues and was approved for production.- Responsible for maintaining a highly dynamical schedule for medical devices delivery to a medical testing laboratory. Liaised with said laboratory to schedule a testing plan that would be the most efficient and likely to meet our client's requirements.- Improved the efficiency, capability and robustness of algorithms used to quality control medical devices parts before assembly.