Ruiqing Fu

Anomaly detection and localization in crowded scenes using short-term trajectories

In this paper we present a method to detect and localize abnormal events in crowded scene. Most existing methods use the patch of optical flow or human tracking based trajectory as representation for crowd motion, which inevitably suffer from noises. Instead, we propose the employment of a new and efficient feature, short-term trajectory, which represent the motion of the visible and constant part of human body that move consistently, for modeling the complicated crowded scene. To extract the short-term trajectory, 3D mean-shift is firstly used to smooth the video frames and 3D seed filling algorithm is performed. In order to detect the abnormal events, all short-term trajectories are treated as point set and mapped into the image plane to obtain probability distribution of normalcy for every pixel. A cumulative energy is calculated based on these probability distributions to identify and localize the abnormal event. Experiments are conducted on known crowd data sets, and the results show that our method can achieve high accuracy in anomaly detection as well as effectiveness in anomalies localization.

Rubbot: Rubbing on flexible loose surfaces

This paper presents a newly-designed robot named “Rubbot” dedicated to climbing on soft flexible clothes. Equipped with novel grippers which grip and rub on clothes, Rubbot is able to climb on flexible clothes and control how much fabric to grasp by feedback from infrared sensor. Rubbot also has a frame which has three passive folders which adjust the climbing posture of Rubbot. This not only makes Rubbot quite functional with clothes of different thicknesses and curved surfaces, but also makes Rubbots motion more flexible. A theory of the deformation of cloth is then presented based on an analysis of creases created while Rubbot is climbing, this leads to a more reliable method to climb flexible surfaces. Finally experiments have verified that Rubbot is effective on flexible surfaces, as it can climb on 95% of the surfaces human clothes and still perform well on non-rigidly backed cloth.

Path planning for clothes climbing robots on deformable clothes surface

This paper proposes a novel path planning method for a robot to climb on the deformable clothes surface. Based on the deformable characteristic of the clothes, the tension force of clothes is analyzed and the model of tension degree is established. A clothes climbing robot called Clothbot is composed of a two-wheeled gripper and a 2 Degrees of Freedom (DOF) tail. Based on the locomotion of this robot, the weights of tension degree and the locomotion characteristic are added into the A* algorithm. Combined with the two weights applied, the optimal path to the target for the Clothbot is obtained. The Clothbot has been developed to evaluate the algorithm. The simulation and the experiments have verified the feasibility of this method. In addition, The error state of the movement of the robot which is called side tumbling has been corrected by the motion of the 2-DOF tail.

A wall-following strategy for mobile robots based on self-convergence

This paper proposes a novel wall-following strategy for mobile robots. This strategy establishes the self-convergence mathematical model by analyzing the relationship between the motion characteristics of mobile robots and installation position of sensors. Different from conventional wall-following approaches that are mostly based on the multi-sensor information fusion technology, this strategy manages to execute the wall-following activity with only a single distance proximity switch. This approach brings several advantages over the previous ones, including the avoidance of mutual interference between sensors and reduction of high hardware cost. In this paper we present the establishment, analysis and implementation of this strategy, with showing the experimental results and giving its potential applications.

An effective approach for active tracking with a PTZ camera

The concept of active tracking is presented to simulate the characteristics of human vision in intelligent visual surveillance. The Pan/Tilt/Zoom (PTZ) camera is generally used for active tracking. In this paper, we present a novel and effective approach for active object tracking with a PTZ camera, and construct a near real-time system for indoor and outdoor scenes. The tracking algorithm of our system is based on the feature matching, with the PID control to drive the camera. The feature extracted from moving people is described as a region covariance matrix which combines the spatial and statistical properties of the targets (e.g. coordinates, color, and gradient). Results from indoor and outdoor experiments demonstrate the effectiveness and accuracy of our approach.

A Stochastic Perturbing Particle Swarm Optimization Model

The particle swarm optimization (PSO) algorithmis a generally used optimal algorithm, which exhibits good performance on optimization problems in complex search spaces. However, traditional PSO model suffers from a local minima, and lacks of effective mechanism to escape from it. This is harmful to its overall performance. This paper presents an improved PSO model called the stochastic perturbing PSO(SPPSO), which tries to overcome such premature convergence through perturbing the swarm with the perturbation and acceptance probability. The performance of the SPPSO is compared with the basic PSO (bPSO) on a set of benchmark functions. Experimental results show that, the new model not only effectively prevent the premature convergence, but also keep the rapid convergence rate like the bPSO.

A stochastic scattering particle swarm optimizer

The particle swarm optimization (PSO) algorithm is a swarm intelligence technique, which has exhibited good performance on finding optimal regions of complex search spaces. However, the basic PSO (bPSO) suffers from the premature convergence in multi-modal optimization. This is due to a decease of swarm diversity that leads to the global implosion and stagnation. It is an acceptable hypothesis that maintaining a high diversity produces a good effect on the search performance of the PSO algorithms. In this paper, we propose a novel optimizer, called the stochastic scattering particle swarm optimizer (SSPSO), which tries to overcome the premature convergence through scattering the swarm stochastically, with a new and simple diversity measure. The performance of the SSPSO is compared with the bPSO on a set of benchmark functions. Experimental results show that, the SSPSO not only prevents the premature convergence to a high degree, but also keeps a rapid convergence rate. Thus, it is clearly a better substitute for the bPSO and other repulsion-based PSO algorithms.

A fast robot path planning algorithm based on image thinning

A fast, reliable mapping and path planning method is essential to autonomous mobile robots. Popular global algorithms usually adopt occupancy gird maps (OGM) or a topological map (TM) to generate a practicable path. In this paper, we present a new path planning algorithm based on the image thinning process and the extraction of key nodes. It requires much less time and memory compared with the OGM method for generating a path, and much simpler, more robust than the TM, while preserving advantages of both. With the new algorithm, a collision-free path could be obtained in real time. Results from simulation experiments validate its efficiency and reliability.

Robust localization system for an autonomous mower

This paper presents a robust vision-based localization system for an autonomous mower, which is significant for both the meadow map building and the successful area covering. Instead of setting the monocular camera toward the scene, which suffers from the disturbance of moving objects, less mark points or variation of illumination, we equip the camera toward the ground with constant illumination compensation. To achieve the localization of the mower, point features are extracted and matched between pairs of frames. Motion is incremental obtained by calculate the rotation and translation transformation of matched feature point pairs. As the angle accumulated error has greater contribution to the location error, angular acceleration sensor is adopted to compensate the angle error especially in the steep turning case. Experiments on meadow with our mowers demonstrate the robustness of our localization system.

Non-binding lower extremity exoskeleton (NextExo) for load-bearing

In this paper, we present a novel non-binding lower extremity exoskeleton (NextExo) for bearing load, where there is no binding point between the NextExo and human. With the innovative structure, the NextExo is able to stand in balance without attaching human, and bear the weights of its own and load completely. This also avoids the damage to operator caused by long-time binding. The NextExo has eight degrees of freedom, all of which are active joints powered by hydraulic actuators. It shadows human motion by one-to-one joints mapping. The man is as the core in the system to keep the NextExo in balance. Meanwhile, the constraint based on Zero Moment Point theory is adopted. The design concept, hardware structure, control scheme and preliminary experiments of NextExo are discussed.