Revolutionary Technology Enables Robots to See in Darkness
A research group under the leadership of Professor Zubin Jacob from the University of Perdue has developed a revolutionary technology known as Hadar, which allows robots to recognize texture and depth even in complete darkness. The breakthrough thermal detection and determination of the range system enables machines to have vision capabilities in low-light conditions.
According to a scientific article published in Nature magazine, it is projected that by 2030, one in ten cars will be autonomous and 20 million robots will assist with both household tasks and work-related activities. The effectiveness of such systems heavily relies on the quality of computer vision.
Traditional technologies like lidars or radars struggle to operate in low-light conditions, while ordinary video cameras are ineffective in the dark. Although thermal imagers can function at night, they only provide blurry non-textured images.
Hadar addresses these challenges by combining information theory, thermal physics, and machine learning. This innovative mechanism maximizes the extraction of data from thermal radiation, allowing for the restoration of sharpness and texture of objects.
The multi-spectral infrared camera plays a crucial role in Hadar. It captures images across various ranges of infrared radiation and decomposes the heat signal into temperature, radiation, and texture components. Machine learning algorithms are then applied to restore the image from a noisy signal.
Field tests conducted in nighttime natural conditions demonstrated Hadar’s ability to recognize fine details that are imperceptible to the human eye, such as ripples on water, wrinkles on tree bark, and grass structures. The image quality achieved was comparable to daytime footage, even in adverse weather conditions like fog and rain.
The development of Hadar opens up new possibilities for future car designs and robotics, particularly in the areas of navigation and image recognition. It can also be utilized in agriculture to monitor crops at night and in medicine to enhance thermal imaging capabilities.
Furthermore, Hadar shows promise in security and video surveillance systems and has clear potential in the military sector. Additionally, it could pave the way for the development of a new generation of consumer night vision tools.
However, there are still challenges to be addressed. One such challenge is to make the system more compact. Currently, the Hadar sensor is large and heavy due to the requirement of multiple colors of invisible infrared radiation for the algorithms to function effectively. The researchers acknowledge the need to reduce the size and cost of the system while also increasing the speed of