Sebum Chun

Improving mobile robot navigation performance using vision based SLAM and distributed filters

In this paper, we suggest a vision-based SLAM (simultaneous localization and mapping) method to improve navigation performance of mobile robot, which is used 2 encoders to calculate its position. If mobile robot is in building, tunnel or under ground facility, it is difficult to obtain navigation information from GPS only navigation system, because there are not enough visible GPS satellites. To overcome this limitation, DR (dead reckoning) system is required. However, as DR operation time goes by, the navigation error is increased because of accumulation of sensor error and noise. There are variety kinds of methods to reduce these errors. In this paper, we use a vision sensor and particle filter. Some clear points on vision sensor image are selected and tracked for error compensation. That is called a SLAM (simultaneous localization and mapping) method. In this paper, distributed particle filter is used to cope with nonlinear observation model and to deal with changing the number of measurements. Computer simulations are conducted to demonstrate the performance of suggested filter.

Adaptive GPS/INS integration for relative navigation

Abstract Relative navigation based on GPS receivers and inertial measurement units is required in many applications including formation flying, collision avoidance, cooperative positioning, and accident monitoring. Since sensors are mounted on different vehicles which are moving independently, sensor errors are more variable in relative navigation than in single-vehicle navigation due to different vehicle dynamics and signal environments. In order to improve the robustness against sensor error variability in relative navigation, we present an efficient adaptive GPS/INS integration method. In the proposed method, the covariances of GPS and inertial measurements are estimated separately by the innovations of two fundamentally different filters. One is the position-domain carrier-smoothed-code filter and the other is the velocity-aided Kalman filter. By the proposed two-filter adaptive estimation method, the covariance estimation of the two sensors can be isolated effectively since each filter estimates its own measurement noise. Simulation and experimental results demonstrate that the proposed method improves relative navigation accuracy by appropriate noise covariance estimation.

INS/Vision Sensor Integrated System for Precise Relative Position Estimation using Landmark

Despite precise relative positioning performance of a GPS based navigation system, it may require the reference station in close boundaries and can be affected by satellite observation environments. Thus, this paper presents an INS and vision sensor integrated system, which uses a combination of feature points and the landmark whose position is already known in order to overcome the limitation of GPS-only system. The proposed system has the advantage of implementing three-dimensional navigation using the geometry of landmark images. The simulation verifies the performance of integrated navigation system, by which the performance enhancement in estimating relative position of the vehicle is proved effectively.

A robust indoor positioning system using two-stage EKF SLAM for first responders in an emergency environment

A positioning method for a first responder in an indoor emergency environment is needed to strengthen a safety for rescue operation. In Korea, it is focused as one of the national agenda. Though many researches are done in this field, but they still have limitations of system preloading and a poor positioning accuracy depending on an arbitrary situation. This paper presents a robust indoor positioning system for the first responder in the scene of fire. The proposed positioning system consists of a monitoring station outside a building, portable access points (APs), and first responders. In the first place, the first responder measure distances to other first responders and portable APs in the hearable range, and calculates its relative position utilizing the simultaneous localization and mapping (SLAM) method implemented with the Extended Kalman Filter (EKF). In this process, the position information stored in the APs and first responders can serve as a map for the others. Then the initial position guess and distance measurements are delivered to the monitoring station via ad-hoc network by portable APs and first responders to correct position estimates of those in the whole system. The monitoring station also uses EKF SLAM method with all available position estimates and distance measurements. In this paper, the system configuration and positioning method are presented, and the simulation on the positioning of the portable APs and the first responders with respect to the monitoring station is carried out to certify the feasibility of the proposed system.

GPS Carrier Multipath Estimation While Attitude Determination

Incorrect ambiguity integer of GPS make a large error on attitude determination. In this paper one method is suggested for estimating the multipath of GPS carrier measurement while attitude determination. The multi-antenna system consists of 4 antennas have the same clock error help to make attitude determination effectively. If the distance between antennas is a half wavelength, it is not necessary to search the ambiguity integer and the multipath of GPS carrier measurement can be estimated. The results of the simulation are shown and analyzed.

Development of a UAV Flight Control System Using a Low Cost GPS/IMU

UAVs(Unmanned Aerial Vehicles) have many applications in military and commercial areas. The flight control system of UAVs is more important than manned aircraft`s because the mission of UAVs must be operated without a human pilot. But very heavy and expensive navigation system makes it difficult to develop UAV flight control system. In this research, GPS/IMU integrated navigation filter was developed for light weight/low cost flight control system of small UAVs. With this navigation filter, full flight control system which has real time operating capability has been developed. The performance of the flight control system is basically checked by HILSIM (Hardware In the Loop SIMulation). Finally, the flight control system is verified by showing performance test result under real flight environment.

Estimation of Angular Acceleration By a Monocular Vision Sensor

Recently, monitoring of two-body ground vehicles carrying extremely hazardous materials has been considered as one of the most important national issues. This issue induces large cost in terms of national economy and social benefit. To monitor and counteract accidents promptly, an efficient methodology is required. For accident monitoring, GPS can be utilized in most cases. However, it is widely known that GPS cannot provide sufficient continuity in urban cannons and tunnels. To complement the weakness of GPS, this paper proposes an accident monitoring method based on a monocular vision sensor. The proposed method estimates angular acceleration from a sequence of image frames captured by a monocular vision sensor. The possibility of using angular acceleration is investigated to determine the occurrence of accidents such as jackknifing and rollover. By an experiment based on actual measurements, the feasibility of the proposed method is evaluated.

Indoor Location Tracking for First Responders using Data Network

In case Wi-Fi network based First responder`s position tracking system is used, range measurement must be generated from RSSI finger print database. However, it is impossible to build up finger print database and to perform rescue operation at same time in the scene of rescue. In this paper, improvised Wi-Fi network without finger print database and pedestrian dead reckoning based first responders tracking system is proposed.

Obstacle Avoidance of GNSS Based AGVs Using Avoidance Vector

The Global Navigation Satellite System(GNSS) is being utilized in numerous applications. The research for autonomous guided vehicles(AGVs) using precise positioning of GNSS is in progress. GNSS based AGVs is useful for setting driving path. This AGV system is more efficient than the previous one. Escipecially, the obstacle is positioned the driving path. Previcious AGVs which follow marker or wires laid out on the road have to stop the front of obstacle. But GNSS based AGVS can continuously drive using obstacle avoidance. In this paper, we developed collision avoidance system for GNSS based AGV using laser scanner and collision avoidance path setting algorithm. And we analyzed the developed system.

Observability Analysis of a Vision-INS Integrated Navigation System Using Landmark

A GNSS/INS integration system can not provide navigation solutions if there are no available satellites. To overcome this problem, a vision sensor is integrated with this system. Since generally a vision aided integration system uses only feature point to compute navigation solutions, it has a problem in observability. In this case, additional landmarks, which is priory known points, can improve the observability. In this paper, the observability is evaluated using TOM/SOM matrix and Eigenvalues. There are always the observability problems in the feature-point-only case, but the landmark-use case is fully observable after the update time. Consequently the landmarks ensure full observability, so the system performance can be improved.