This review article aims to address common research questions in passive polarized vision for robotics. What kind of polarization sensing can we embed into robots? Can we find our geolocation and true north heading by detecting light scattering from the sky as animals do? How should polarization images be related to the physical properties of reflecting surfaces in the context of scene understanding? This review article is divided into three main sections to address these questions, as well as to assist roboticists in identifying future directions in passive polarized vision for robotics. After an introduction, three key interconnected areas will be covered in the following sections: embedded polarization imaging; polarized vision for robotics navigation; and polarized vision for scene understanding. We will then discuss how polarized vision, a type of vision commonly used in the animal kingdom, should be implemented in robotics; this type of vision has not yet been exploited in robotics service. Passive polarized vision could be a supplemental perceptive modality of localization techniques to complement and reinforce more conventional ones.

Autonomous navigation requires robust strategies, particularly in GPS-denied environments. Of great interest for geolocation or to estimate North is the measurement and processing of polarized skylight patterns. In this study, we fully describe an open source and easily upgradable simulator, named OpenSky, which can simulate the measurement of sky polarization properties, i.e., the light intensity, the angle of polarization and the degree of linear polarization of light seen through a polarimetric camera. OpenSky is an open source simulator available on open repositories (Github and Open Science Framework). It structured around 5 blocks, each of which can easily be improved, replaced or completed by users. These blocks individually simulate sky polarization properties, skylight intensity, optical conjugation, polarizing filters and sensors. The high fidelity of OpenSky was assessed by comparing the simulated camera captures to experimental raw images, from which both the angle and the degree of polarization were extracted by image processing. OpenSky will be useful for developing novel celestialbased sensors and estimating their potential relevance to scientific communities (fundamental or applied sciences). OpenSky could also become a powerful tool to train or validate deep neural networks. One major interest is the ability of the OpenSky simulator to generate many sky conditions at various positions on Earth, which can be difficult and costly to obtain in real world.

Global navigation satellite systems (GNSS) provide robust and accurate navigation information. However, GNSS can be spoofed, potentially leading to dramatic errors in localization estimation. Inspired by nature, research has been undertaken into skylight polarization as a reliable alternative to the existing navigation systems. Recent studies have used skylight polarization to estimate the observer’s position. Nevertheless, such methods generally require a precise knowledge of time, date, and solar ephemeris. Alternatively, time invariances of the skylight polarization pattern, obtained from at least three polarization snapshots, were used to estimate the position of the north celestial pole (NCP) in the sky, without additional information, providing the true north bearing and the latitude of the observer. Here, we explain how these skylight polarization time invariances are related to the solar declination, and how this relation can be used to find the position of the NCP. Based on this property, we introduce a method to find the position of the NCP requiring only two skylight polarization observations over time and a prior knowledge about the solar declination. We tested our method on real skylight polarization images acquired by a polarimetric camera during a 1 month observation. A comparison between estimated and ground truth coordinates validates the theoretical model. Finally, our method can also be used in reverse to find the solar declination from the skylight polarization time invariances and the position of the NCP, demonstrating how skylight polarization can be used as a celestial calendar.