Key points are not available for this paper at this time.
Abstract Traffic accidents claim the lives of millions of people each year. In addition to preventing these accidents, technology in autonomous vehicles has the potential to spark a technological renaissance. Thermal imaging camera is well known for its capability of providing information about pedestrians and animals in low visibility situations, while RGB cameras are disabled with insufficient visible light. Fusing a thermal imaging camera into a sensor system for an autonomous vehicle will help avoid running over pedestrians and animals on the road under extreme weather and environmental conditions, such as total darkness, rain, fog, and dust. Ensuring that a low-cost but effective sensor is used to be able to detect warm-body obstacles in the path of the vehicle. This engineering solution will provide proper device initialization and utilization in autonomous vehicle (AV) applications. The project's objective is to work and develop a Light Detection and Ranging (LiDAR) and a FLIR thermal imaging system to measure the distance of live bodies on the road. The geometric intrinsic and extrinsic calibration of thermal imaging cameras is challenging since the regular black-white checkerboard method used for RGB cameras is not visible by the thermal camera. A new geometric calibration method of a thermal imaging camera will be discussed, tested and validated in this study. Physical prototypes of thermoelectric cooled and heated pads will be utilized to better calibrate the FLIR/LiDAR sensor system. Calibration algorithms will be developed in Python. The depth measurement of a pedestrian by the thermal camera and the LiDAR will be conducted. Challenges and design constraints will also be discussed.
Salim et al. (Thu,) studied this question.