Purdue University researchers are working on an innovative method called HADAR (Heat-Assisted Detection and Ranging) that aims to revolutionize machine vision and perception in robotics. HADAR combines thermal physics, infrared imaging, and machine learning to enable robots and autonomous systems to perceive texture, depth, and physical attributes of objects and scenes, even in challenging lighting conditions. The new method can discern texture and depth and understand the physical characteristics of individuals and their surroundings.
The traditional methods used in machine vision, such as LiDAR, radar, and sonar, have limitations that increase as they scale up, including signal interference and potential risks to eye safety. While video cameras working on sunlight or other sources of illumination are advantageous, they face difficulties in low-light conditions like nighttime, fog, or rain.
Thermal imaging is a passive sensing method that can detect heat radiation from objects in a scene, enabling vision in darkness and inclement weather. However, it suffers from the “ghosting effect,” where thermal images lack texture and features, hindering machine perception.
HADAR overcomes these challenges by recovering textures from the cluttered heat signal and accurately disentangling temperature, emissivity, and texture (TeX) of all objects in a scene. The method achieves a level of perception similar to daylight, allowing robots and autonomous systems to see through pitch darkness as if it were daytime.
The researchers are working on improving the size and data collection speed of HADAR for practical applications in self-driving cars, robots, agriculture, defense, geosciences, healthcare, and wildlife monitoring. The initial applications focus on automated vehicles and robots operating in complex environments, with potential for broader use in various industries.
The work has been featured in the peer-reviewed journal Nature, and the Purdue Innovates Office of Technology Commercialization has applied for a patent on the intellectual property. The researchers have received funding from DARPA to support their research, and they are actively seeking industry partners to further develop their innovations.
Table of Contents
Frequently Asked Questions (FAQs) about Machine Vision
What is HADAR?
HADAR stands for Heat-Assisted Detection and Ranging, a patent-pending method developed by researchers at Purdue University. It revolutionizes machine vision and perception in robotics by combining thermal physics, infrared imaging, and machine learning.
How does HADAR work?
HADAR leverages thermal physics and infrared imaging to detect and analyze heat radiation from objects in a scene. Machine learning algorithms are then used to recover textures and accurately perceive temperature, emissivity, and texture (TeX) of all objects, even in challenging lighting conditions like pitch darkness.
What are the benefits of HADAR?
HADAR enables robots and autonomous systems to see in low-light conditions as if it were daylight, overcoming the limitations of traditional methods. It enhances machine perception, allowing for better discernment of texture, depth, and physical attributes of scenes and objects.
What are the potential applications of HADAR?
The initial applications of HADAR focus on automated vehicles and robots operating in complex environments. However, the technology could be further developed for various industries, including agriculture, defense, geosciences, healthcare, and wildlife monitoring.
How fast is HADAR in processing information?
The current sensor used in HADAR takes around one second to create one image. To apply it to self-driving cars or robots, efforts are being made to improve the size and data collection speed to achieve a frame rate of around 30 to 60-hertz per second.
What makes HADAR different from traditional methods like LiDAR and thermal imaging?
HADAR is fully passive and physics-aware, combining thermal physics, infrared imaging, and machine learning. It overcomes the “ghosting effect” observed in traditional thermal imaging, where textures and features are lost. Moreover, it can perceive scenes in low-light conditions like pitch darkness, providing advantages over traditional active sensors like LiDAR, radar, and sonar.
More about Machine Vision
- Purdue University: Link
- Nature Journal: Link
- HADAR Research Paper in Nature: Link
- Purdue Innovates Office of Technology Commercialization: Link
- DARPA: Link
