Senior/Lead Computer Vision Engineer (Navigation)

The Role:

We are looking for an experienced Computer Vision Engineer to develop and optimize an autonomous navigation system for UAVs. You will be responsible for developing a solution based on visual-inertial odometry methods in GNSS-denied environments.

Must have skills:
● 5+ years of commercial experience in Computer Vision and Machine Learning preferably in UAVs, robotics, or autonomous systems
● Proficiency in Python
● Experience with OpenVINS
● Experience with such frameworks as TensorFlow and OpenCV, YOLO or PyTorch
● Experience with visual-inertial odometry, camera position estimation, detection and tracking of visual features of static objects
● Understanding of key performance-defining metrics of computer vision algorithms
● Familiarity with AirSim, Gazebo or other simulation software
● Experience with different footage formats: RGB, greyscale, infrared, thermal
● Experience in deploying and optimizing models for single-board computers such as Raspberry Pi, Nvidia Jetson
● Familiarity with UAVs, their typical on-board real-time CV pipelines
● Familiarity with drone flight control principles
● Experience with deep learning, neural networks and CNN
● Intermediate level of English

Nice to have skills:

● Understanding of Kalman filtering and sensor fusion techniques
● Familiarity with Linux
● Familiarity with the MAVLink protocol and ArduPilot
● Experience with Google Coral AI accelerator or other edge TPUs
● Experience with geospatial data processing and GIS tools (e.g., GDAL, Rasterio, QGIS)

Your responsibilities:

● Develop, train, and optimize computer vision models for an autonomous UAV navigation system based on visual-inertial odometry and other methods
● Collaborate with UAV and embedded engineers for deploying CV models to hardware
● Develop model performance metrics for evaluating model accuracy and inference
● Perform flight simulation, participate in field testing and flight data analysis to validate model performance
● Optimize the models for various weather and lighting conditions (including night time) as well as a range of UAV flying speeds/heights with the focus on result reproducibility
● Utilize geospatial data for real-time updating of estimated UAV position