At Suind, we are building a comprehensive crop protection system driven by autonomous drones, computer vision and deep learning. Our goal is to help farmers across the globe to sustainably prevent crop losses, increase production and reduce input costs.

In this role I developed robust algorithms for autonomous drone navigation in a variety of environments including GPS degraded/denied environments like cluttered urban environments or completely indoor settings. Specific work areas include:1. Development of Visual-Inertial odometry and SLAM algorithms for resource constrained computing platforms commonly used on drones.2. Multi-sensor fusion, for example Vision and GPS fusion.3. Development of planning and control algorithms especially for Obstacle Avoidance in real-life settings.4. Prototyping with different embedded platforms including Intel, Nvidia, Qualcomm etc. as well as different sensor types including VI-sensors, Depth sensors, GPS receivers etc.

Projects:Precision Landing with a Cargo Drone (February-August, 2018) (Master's Thesis) (Robotics and Perception Group, ETH/UZH)Developed a system for precise and safe, vision ­augmented autonomous landing of an otherwise GPS controlled cargo drone in cluttered environments. Was able to achieve centimeter level landing accuracy in comparison to the meter level accuracy of GPS. The system was developed to autonomously track a delivery mission and seamlessly fuse GPS and VIO based navigation to provide the best overall mission accuracy. Additionally the system was able to avoid obstacles and unsafe landing locations to land on a safe spot. This project was done in collaboration with the Laboratory of Intelligent Systems at EPFL, Lausanne.Online Loop Closure Detection (September-December, 2017) (Robotics and Perception Group, ETH/UZH)Developed a solution to provide online loop closure detection for the semi­-direct visual odometry (SVO) system developed by the Robotics and Perception group under Prof. Davide Scaramuzza. Loop closure detection aids in mitigating the drift that is accumulated by an odometry system over time, making localization more robust. The solution had to be computationally optimal to run on limited onboard resources on a drone providing reasonable accuracy at the same time.

Assisted masters and bachelor students with their projects focused on Perception and Control of NAO humanoid robots from Aldebaran (now Softbank Robotics). These robots formed a part of the Team NomadZ, the robotic soccer team from ETH Zurich that participated in the "Standard Platform League" at the Robocup. Was also a member of the team at various competitions including Robocup World Cup 2017 (Nagoya, Japan), Robocup World Cup 2018 (Montreal, Canada) and the Robocup German Open 2018 (Magdeburg, Germany).

Worked as site electrical and control engineer at Brockman 4, one of the major iron ore mine sites in Western Australia. Main focus was towards improving the control loop performance and optimising the plant through put as well as providing technical support for PLC, Communication and other electrical systems on site.

Worked as a fixed plant systems engineer at Rio Tinto's remote 24X7 operations centre, providing systems support for a variety of applications like citect SCADA, wonderware SCADA, Pilz safety PLC system, PAS systems.

Worked on an all wheel driven electric bicycle. Specific tasks were:1. Design and programming of inverter circuits, with a regulated throttle input to drive bldc motor.2. Testing and failure analysis.3. First considerations regarding a battery management system.

Worked as a trainee in Areva's HV transformer design team. Worked on Earthing transformer design and NGRs. Developed an opmisation program for cutting of CRGO laminations.