Learning and then recognizing a route, whether travelled during the day or at night, in clear or inclement weather, and in summer or winter is a challenging task for state of the art algorithms in computer vision and robotics. In this paper, we present a new approach to visual navigation under changing conditions dubbed SeqSLAM. Instead of calculating the single location most likely given a current image, our approach calculates the best candidate matching location within every local navigation sequence. Localization is then achieved by recognizing coherent sequences of these “local best matches”. This approach removes the need for global matching performance by the vision front-end - instead it must only pick the best match within any short sequence of images. The approach is applicable over environment changes that render traditional feature-based techniques ineffective. Using two car-mounted camera datasets we demonstrate the effectiveness of the algorithm and compare it to one of the most successful feature-based SLAM algorithms, FAB-MAP. The perceptual change in the datasets is extreme; repeated traverses through environments during the day and then in the middle of the night, at times separated by months or years and in opposite seasons, and in clear weather and extremely heavy rain. While the feature-based method fails, the sequence-based algorithm is able to match trajectory segments at 100% precision with recall rates of up to 60%.

SeqSLAM : visual route-based navigation for sunny summer days and stormy winter nights

To address the problem of estimating camera trajectory and to build a structural three-dimensional (3D) map based on inertial measurements and visual observations, this paper proposes point-line visual-inertial odometry (PL-VIO), a tightly-coupled monocular visual-inertial odometry system exploiting both point and line features. Compared with point features, lines provide significantly more geometrical structure information on the environment. To obtain both computation simplicity and representational compactness of a 3D spatial line, Plucker coordinates and orthonormal representation for the line are employed. To tightly and efficiently fuse the information from inertial measurement units (IMUs) and visual sensors, we optimize the states by minimizing a cost function which combines the pre-integrated IMU error term together with the point and line re-projection error terms in a sliding window optimization framework. The experiments evaluated on public datasets demonstrate that the PL-VIO method that combines point and line features outperforms several state-of-the-art VIO systems which use point features only.

https://www.mdpi.com/1424-8220/18/4/1159/pdf

PL-VIO: Tightly-Coupled Monocular Visual-Inertial Odometry Using Point and Line Features.

THE meetings of the Institution of Electrical Engineers will be resumed in London during the second half of the current session if the present conditions continue. This applies to the ordinary meetings, informal meetings, meetings of the three technical sections and of the London Students’ Section. As regards activities in the provinces, the various committees will decide in the light of local conditions whether they are able to carry out programmes of meetings, visits and functions. The first ordinary meeting of the programme for the second half of the session will take place on January 25, at 6 p.m., when a discussion on “Fire-Fighting Equipment for Electrical Installations”, based on the E.R.A. Report on this subject, will be introduced by Messrs. H. W. Swann, J. Hacking and R. A. McMahon.

Institution of Electrical Engineers

Autonomous outdoor navigation requires reliable multisensory fusion strategies. Desert ants travel widely every day, showing unrivaled navigation performance using only a few thousand neurons. In the desert, pheromones are instantly destroyed by the extreme heat. To navigate safely in this hostile environment, desert ants assess their heading from the polarized pattern of skylight and judge the distance traveled based on both a stride-counting method and the optic flow, i.e., the rate at which the ground moves across the eye. This process is called path integration (PI). Although many methods of endowing mobile robots with outdoor localization have been developed recently, most of them are still prone to considerable drift and uncertainty. We tested several ant-inspired solutions to outdoor homing navigation problems on a legged robot using two optical sensors equipped with just 14 pixels, two of which were dedicated to an insect-inspired compass sensitive to ultraviolet light. When combined with two rotating polarized filters, this compass was equivalent to two costly arrays composed of 374 photosensors, each of which was tuned to a specific polarization angle. The other 12 pixels were dedicated to optic flow measurements. Results show that our ant-inspired methods of navigation give precise performances. The mean homing error recorded during the overall trajectory was as small as 0.67% under lighting conditions similar to those encountered by ants. These findings show that ant-inspired PI strategies can be used to complement classical techniques with a high level of robustness and efficiency.

/pdf/antbot-a-six-legged-walking-robot-able-to-home-like-desert-25c175i5om.pdf

AntBot: A six-legged walking robot able to home like desert ants in outdoor environments.

A Review of Visual-Inertial Simultaneous Localization and Mapping from Filtering-Based and Optimization-Based Perspectives

Navigation plays a vital role in our daily life. As traditional and commonly used navigation technologies, Inertial Navigation System (INS) and Global Navigation Satellite System (GNSS) can provide accurate location information, but suffer from the accumulative error of inertial sensors and cannot be used in a satellite denied environment. The remarkable navigation ability of animals shows that the pattern of the polarization sky can be used for navigation. A bio-inspired POLarization Navigation Sensor (POLNS) is constructed to detect the polarization of skylight. Contrary to the previous approach, we utilize all the outputs of POLNS to compute input polarization angle, based on Least Squares, which provides optimal angle estimation. In addition, a new sensor calibration algorithm is presented, in which the installation angle errors and sensor biases are taken into consideration. Derivation and implementation of our calibration algorithm are discussed in detail. To evaluate the performance of our algorithms, simulation and real data test are done to compare our algorithms with several exiting algorithms. Comparison results indicate that our algorithms are superior to the others and are more feasible and effective in practice.

https://www.mdpi.com/1424-8220/14/9/17068/pdf

A Novel Angle Computation and Calibration Algorithm of Bio-Inspired Sky-Light Polarization Navigation Sensor

This paper presents a method to extract heading information from the sky light polarization pattern and successfully implements the polarized light compass (PL-compass) under foliage environment with heavy occlusions for the first time. A classifier based on the primal support vector machine is trained to detect the sky area, and the random sample consensus algorithm is used to choose the inliers among the sky area. The sun vector is built upon the valid measurements. The attitude angles (Roll and Pitch) from the inertial navigation system (INS) are used to establish the Zenith direction when the platform tilts during traveling. The results show that the PL-compass can provide accurate heading angle (RMSE = 1.58°) by seeing only a tiny part of the skylight (minimum of 0.028%) in the dynamic vehicle test. The PL-compass is also integrated with the visual INS based on the indirect extended Kalman filter to improve the navigation accuracy. Each time a new measurement becomes available, the states estimated by the INS are used to check the consistency of the innovation and reject the outliers. The results show that the position errors of the integrated navigation system are greatly reduced with the help of the PL-compass. The estimated trajectory tracks the ground truth very well. The position error is 25.9 m (1.03% of the distance traveled) with heading errors less than 1° in the experiment.

Polarized Light Compass-Aided Visual-Inertial Navigation Under Foliage Environment

Skylight polarization provides a significant navigation cue for certain polarization-sensitive animals. However, the precision of the angle of polarization (AOP) of skylight for vehicle orientation is not clear. An evaluation of AOP must be performed before it is utilized. This paper reports an evaluation of AOP of skylight by measuring the skylight polarization patterns of clear and cloudy skies using a full-sky imaging polarimetry system. AOP measurements of skylight are compared with the pattern calculated by the single-scattering Rayleigh model and these differences are quantified. The relationship between the degree of polarization (DOP) and the deviation of AOP of skylight is thoroughly studied. Based on these, a solar meridian extracted method is presented. The results of experiments reveal that the DOP is a key parameter to indicate the accuracy of AOP measurements, and all the output solar meridian orientations extracted by our method in both clear and cloudy skies can achieve a high accuracy for vehicle orientation.

https://www.mdpi.com/1424-8220/15/3/5895/pdf

An Evaluation of Skylight Polarization Patterns for Navigation

Animals, such as Savannah sparrows and North American monarch butterflies, are able to obtain compass information from skylight polarization patterns to help them navigate effectively and robustly. Inspired by excellent navigation ability of animals, this paper proposes a novel image-based polarized light compass, which has the advantages of having a small size and being light weight. Firstly, the polarized light compass, which is composed of a Charge Coupled Device (CCD) camera, a pixelated polarizer array and a wide-angle lens, is introduced. Secondly, the measurement method of a skylight polarization pattern and the orientation method based on a single scattering Rayleigh model are presented. Thirdly, the error model of the sensor, mainly including the response error of CCD pixels and the installation error of the pixelated polarizer, is established. A calibration method based on iterative least squares estimation is proposed. In the outdoor environment, the skylight polarization pattern can be measured in real time by our sensor. The orientation accuracy of the sensor increases with the decrease of the solar elevation angle, and the standard deviation of orientation error is 0 . 15 ∘ at sunset. Results of outdoor experiments show that the proposed polarization navigation sensor can be used for outdoor autonomous navigation.

https://www.mdpi.com/1424-8220/17/11/2623/pdf

Design and Calibration of a Novel Bio-Inspired Pixelated Polarized Light Compass

The remarkable polarization vision of animals provides a significant inspiration for robotic navigation and visual enhancement, as polarization pattern provides additional information besides spectral signatures. A novel bio-inspired polarization camera is proposed in this paper, which can realize real-time image-based polarization measurement. The composition of the system is described and the optimal estimation of the polarization state is derived based on the least square algorithm. This paper concentrates particularly on the camera orientation algorithms and visual enhancement methods with it. To estimate the camera’s heading angle with the skylight polarization pattern, the sun vector is established as an optimization problem of finding the minimum eigenvector. The solar meridian is also estimated from the degree of polarization pattern by detecting reflectional symmetry axes. The result shows that the measured polarization patterns are very close to the theory. The maximum orientation error of the proposed method based on angle of polarization is about 0.5°. The average error is 0.012° with standard deviation of 0.28°. Thus, the novel polarization camera could be used as sun compass. When observing scenes in distance, the polarization camera is used to decouple the airlight from the object radiance, which results in much better contrast. More importantly, the polarization information is helpful for scene identification and object detection. The result also shows that the polarization camera can reasonably cope with the semi-reflection problem.

Yujie Wang

Papers

A Novel Angle Computation and Calibration Algorithm of Bio-Inspired Sky-Light Polarization Navigation Sensor

Polarized Light Compass-Aided Visual-Inertial Navigation Under Foliage Environment

An Evaluation of Skylight Polarization Patterns for Navigation

Design and Calibration of a Novel Bio-Inspired Pixelated Polarized Light Compass

Bionic Orientation and Visual Enhancement With a Novel Polarization Camera