Performing reliable and computationally efficient loop closure detection in real-world environments still remains a challenging problem. In this paper, we propose a novel method for efficient loop closure detection in different times of day. An illumination invariant color transform is applied to images that are represented by a whole-image descriptor, named PALM. The efficiency of our method resides either in description of the places or in image matching in which FLANN is used for fast nearest neighbor search. With this approach, searching time is decreased about 70 times compared to standard brute-force search with no significant loss of accuracy. According to the experiments that are performed in real-world datasets, the proposed method successfully accomplishes to detect loops under varied illumination conditions with high accuracy, and it allows real-time operation for long-life localization and mapping.

As most robot navigation systems for large-scale outdoor applications have been implemented based on high-end sensors, it is still challenging to implement a low-cost autonomous groundbased vehicle. This paper presents an autonomous navigation system using only a stereo camera and a low-cost GPS receiver. The proposed method consists of Visual Odometry (VO), pose estimation, obstacle detection, local path planning and a waypoint follower. VO computes a relative pose between two pairs of stereo images. However, VO inevitably suffers from drift (error accumulation) over time. A low-cost GPS provides absolute locations that can be used to correct VO drift. We fuse data from VO and GPS to achieve more accurate localization both locally and globally, using an Extended Kalman Filter (EKF). To detect obstacles, we use a dense depth map that is generated by stereo disparity estimation and transformed into a 2D occupancy grid map. Local path planning computes temporary waypoints to avoid obstacles, and a waypoint follower navigates the robot towards the goal point. We evaluated the proposed method with a mobile robot platform in real-time experiments in an outdoor environment. Experimental results show that the mobile vision and control system is capable of traversing roads in this outdoor environment autonomously.

Haptic devices have been studied as useful tools for motor learning. In use of the devices, proprioceptors are used for movement perception. Therefore, it is important for designing the devices to characterize proprioceptor performance. This study focuses on the scheme to perceive velocity differences from beforeacceleration velocity for various accelerations. We measured the velocity JNDs (Just Noticeable Differences) in one-way elbow flexion movements to examine the following two hypotheses. (1) "Local scheme"; the magnitude of the present acceleration, which can be regarded as a local velocity difference, plays a crucial role. (2) "Global scheme"; the global velocity difference, which is defined as the difference between the present accelerating-velocity and the before-acceleration velocity, does so. For each of the schemes, the following characteristics are expected. (1) If acceleration magnitude affects the perception, humans cannot notice the velocity differences in acceleration conditions less than a threshold. It results in a tendency that velocity JNDs in small acceleration conditions would be much larger than those in large acceleration conditions. (2) If acceleration does not, but if the global velocity does, the velocity JNDs stay unchanged even though acceleration varies because humans perceive the velocity differences, just referring two different absolute velocities.

ERL Emergency is an outdoor multi-domain robotic competition inspired by the 2011 Fukushima accident. The ERL Emergency Challenge requires teams of land, underwater and flying robots to work together to survey the scene, collect environmental data, and identify critical hazards. To prepare teams for this multidisciplinary task a series of summer schools and workshops have been arranged. In this paper the challenges and hands-on results of bringing students and researchers collaborating successfully in unknown environments and in new research areas are explained. As a case study results from the euRathlon/SHERPA workshop 2015 in Oulu are given.

This paper aims at investigating the ancient Chinese textile in order to facilitate the growing trend of an interdisciplinary study between art history, industrial design and imaging science. This is an early attempt to study how decorative patterns of the textiles were created with various weaving techniques with the help of digital technology. Since the captured fabric image only reveals the floating yarn, the combination of the underneath yarns are unknown. From the mathematical point of view, the weaving technique can be regarded as a research problem of combinatorics that contains how the yarns of weft and warp intersect with each other. Hence, the analysis of the weaving pattern contains two layers: (a) detection of the floating yarn and (b) estimation of the combination of the underneath yarns. Previously, the regular bands (RB) method is a tool for regularity analysis that has been successfully applied to patterned fabric inspection. This paper achieves the first layer goal, which applies computer vision technique in the imaging science through the RB method to achieve the detection of the floating yarns of images of some ancient Chinese textiles. Ancient textile samples from Ming dynasty, China (ca. 1368-1644 CE) are utilized for the experiments in the paper.

This paper presents an accurate and robust surgical instrument recognition algorithm to be used as part of a Robotic Scrub Nurse (RSN). Surgical instruments are often cluttered, occluded and displaying specular light, which cause a challenge for conventional vision algorithms. A learning-through-interaction paradigm was proposed to tackle this challenge. The approach combines computer vision with robot manipulation to achieve active recognition. The unknown instrument is firstly segmented out as blobs and its poses estimated, then the RSN system picks it up and presents it to an optical sensor in an established pose. Lastly the unknown instrument is recognized with high confidence. Experiments were conducted to evaluate the performance of the proposed segmentation and recognition algorithms, respectively. It is found out that the proposed patch-based segmentation algorithm and the instrument recognition algorithm greatly outperform their benchmark comparisons. Such results indicate the applicability and effectiveness of our RSN system in performing accurate and robust surgical instrument recognition.

In this paper, we propose a simple but efficient video segmentation scheme for real-time video applications. First, we temporally separate video frames into scenes, comparing the chisquare distance between consecutive frames. To partition each frame into disjoint regions, then, a pixel-wise color clustering scheme is employed, which is based on K-means clustering and EM algorithm. Finally, we regularize computational complexity to apply the proposed scheme into embedded video processing system. Due to pixel-wise video segmentation with very low complexity, the proposed scheme yields a realistic framework for real-time video applications.

We present a hybrid multi-line scan approach which enables simultaneous acquisition of light field & photometric stereo data. While light fields capture mostly large-scale surface deviations and rely on visible surface structures, photometric stereo is primarily sensitive to fine surface deviations and does not rely on visible structures. The combination of both approaches yields a solid performance for a large variety of depths, ranging from macro- to microscopic scales. Contrary to traditional photometric stereo, that relies on a strobed illumination, our approach uses two constant light sources which, however, generate multiple illumination geometries in different portions of the camera's field of view. Our object is moving on a conveyor belt during the acquisition process. Due to our multi-line scan sensor the object is observed from several viewing angles. The object's movement is causing each object point to be illuminated under several illumination directions. Hence, during our acquisition process the object points are captured under all feasible viewing angles and lighting conditions. In our system, surface normals are derived making use of the Lambert's cosine law. However, due to the lack of illuminations spanning orthogonally to the transport direction, the surface normals can be inferred only in the transport direction. We present a variational approach for 3D depth reconstruction designed specifically for our hybrid setup that jointly takes into account the light field as well as photometric stereo depth cues and provides one globally consistent solution. Depth maps obtained by the proposed algorithm show both the large-scale accuracy as well as sensitivity to fine surface details.

We present a line-scan stereo system and descriptor-based dense stereo matching for high-performance vision applications. Additionally we introduce a post-processing step based on total variation (TV) regularization for robust disparity estimation. Descriptor-based matching utilizes the Stochastic Binary Local Descriptor (STABLE). The performance of STABLE was shown to be superior to other binary descriptors, both w.r.t. stereo reconstruction quality as well as runtime performance. Regularized estimation of disparity maps is suggested as a hierarchical and iterative post-processing procedure where the Pseudo-Huber-TV norm was employed. We describe the hardware setup consisting of two line-scan cameras mounted in a car trailer and observing the road surface. Presented are results of 3D road surface reconstruction which are used in applications of road infrastructure maintenance.