Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Bringing Thermal Cameras into Robotics
To expand the frontiers of computer vision by using thermographic cameras and investigate their application in robotics
Keywords: Alternative sensing, field robotics, computer vision, deep-learning
Thermographic cameras can capture detailed images regardless of ambient lighting conditions.They use an infrared (IR) sensing technology to map heat variations within the sensor’s range and field-of-view, providing movement detection and hot-spot mapping even in total darkness. Visible range covers wavelengths of approximately 400 – 700 nanometres (nm) in length. However, thermographic cameras generally sample thermal radiation from within the longwave infrared range(approximately 7,000 – 14,000 nm) with a great potential in robotics.
Thermography images are useful to identify week points on the power line, along the cable and on the isolators or containers. However, current lightweight thermal cameras are unexplored, with limited in pixel resolution (32x32 pixels) unable to deliver exceptional sensitivity, resolution and image quality for meaningful applications.
This work aims to expand the frontiers of computer vision by using thermographic cameras and investigate their application in robotics i.e. perception, state estimation and path planning. The project will combine traditional computer vision techniques together with deep-learning approaches to bring thermography images into the field of robotics.
Requirements: Background in computer vision and machine learning - Deep learning experience preferable – Excellent programming experience in C++ and Python
Thermographic cameras can capture detailed images regardless of ambient lighting conditions.They use an infrared (IR) sensing technology to map heat variations within the sensor’s range and field-of-view, providing movement detection and hot-spot mapping even in total darkness. Visible range covers wavelengths of approximately 400 – 700 nanometres (nm) in length. However, thermographic cameras generally sample thermal radiation from within the longwave infrared range(approximately 7,000 – 14,000 nm) with a great potential in robotics.
Thermography images are useful to identify week points on the power line, along the cable and on the isolators or containers. However, current lightweight thermal cameras are unexplored, with limited in pixel resolution (32x32 pixels) unable to deliver exceptional sensitivity, resolution and image quality for meaningful applications.
This work aims to expand the frontiers of computer vision by using thermographic cameras and investigate their application in robotics i.e. perception, state estimation and path planning. The project will combine traditional computer vision techniques together with deep-learning approaches to bring thermography images into the field of robotics.
Requirements: Background in computer vision and machine learning - Deep learning experience preferable – Excellent programming experience in C++ and Python
Perception, state estimation or path planning using thermographic cameras.
Perception, state estimation or path planning using thermographic cameras.
Javier Hidalgo-Carrió (jhidalgocarrio@ifi.uzh.ch) and Giovanni Cioffi (cioffi@ifi.uzh.ch)
Javier Hidalgo-Carrió (jhidalgocarrio@ifi.uzh.ch) and Giovanni Cioffi (cioffi@ifi.uzh.ch)