An adaptive filtering approach to target tracking

Madyastha, Venkatesh; Caliset, A.J.

doi:10.1109/acc.2005.1470139

Cited by 17 publications

(25 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The time axis for these plots are selected based on the obstacle visibility in figure 5(b), where the solid lines are the obstacle visibility with EKF estimates and the dashed lines represent the same with the case of EKF+NN. Hence the errors for obstacle 4, 5 and 6 are plotted in the time interval of [10,25] secs, [12,24] secs and [12,30] secs respectively. As expected, the performance of the adaptive EKF, i.e.…”

Section: Initial Parametersmentioning

confidence: 99%

Neural Network based Adaptive Estimation and Guidance: Application to 2D Obstacle Avoidance

Madyastha¹,

Calise

2011

AIAA Guidance, Navigation, and Control Conference

View full text Add to dashboard Cite

In this paper, an adaptive estimation method is presented for use in the control of a class of multivariable systems with unmodeled dynamics coupled to an otherwise known model for the process. The adaptive estimator takes the form of augmenting an extended Kalman filter designed for the system without unmodeled dynamics. The design of the adaptive element employs a linearly parameterized neural network. The states of the adaptive estimator are used to form the feedback controller. The neural network weights are adjusted on line using a linear observer for the nominal system's error dynamics. Boundedness of the error signals is proven using Lyapunov's direct method. To show the efficacy of the approach, the problem of obstacle avoidance in two dimensions is considered, where the goal is for an unmanned vehicle to traverse a nominal path while avoiding obstacles in the course of travel.

show abstract

Section: Initial Parametersmentioning

confidence: 99%

Neural Network based Adaptive Estimation and Guidance: Application to 2D Obstacle Avoidance

Madyastha¹,

Calise

2011

AIAA Guidance, Navigation, and Control Conference

View full text Add to dashboard Cite

show abstract

“…Next, a tracking control law will be designed with the dynamic inversion by stabilizing the augmented system. At first, a linear control law can be designed to stabilize the linear augmented system in which v is regarded as the control variable, v c , as follows, v c = Fx, (17) where F is the feedback gain matrix chosen so that A + BF is asymptotically stable. Then, let v = v c and a gg be regarded as the expected acceleration, a ggc .…”

Section: Design Of Tracking/ Following Control Lawmentioning

confidence: 99%

“…In order to implement the RPT control laws to complete the tracking/ following missions, it needs to estimate the velocity and acceleration of the target as there is only the relative displacement between the target and tracker measurable. 1,8,11,17,18,25 Actually, it is a challenge to estimate the velocity and acceleration of a target based on the measured relative displacement. There were a lot of researchers dedicating their effort to this problem.…”

Section: Introductionmentioning

confidence: 99%

Vision based Target Tracking/ Following and Estimation of Target Motion

Peng

Zhao

Lin

et al. 2013

AIAA Guidance, Navigation, and Control (GNC) Conference

View full text Add to dashboard Cite

The vision based target tracking/ following is presented. There is no interior communication between the target and tracker and their relative displacement is measured with the 3D visual sensors. The tracking/ following control law is designed in the half North-EastDown (NED) frame with the dynamic inversion and the robust perfect tracking (RPT)techniques. The half NED frame is formed by moving the origin of the NED frame to the mass center of the target and then rotating the X axis along the Z axis to be aligned with the flight direction of the target. The RPT technique is one of the better techniques to design the tracking control laws for the linear plants. It involves with the high-order derivatives of the reference to improve the tracking/ following performance. In order to implement the RPT control law, a novel algorithm is developed to estimate the velocity and acceleration of the target with the measured relative displacement between the target and tracker. The algorithm is developed based on the extended Kalman filtering and under assumption of the quasi steady states in which the derivatives of the target ground speed, flight path and directional angles are constant. The simulation results demonstrate that the resulting the closed-loop system is capable of tracking/ following the 3D maneuvering target in the scheduled relative distance and orientation with respect to the target.Keywords: estimation of target motion, extended Kalman filtering, filtering and estimations, vision based tracking, vision based applications. NomenclatureB bc Transformation matrix from the body frame to the camera frame B gb Transformation matrix from the NED frame to the body frame B gc Transformation matrix form the NED frame to the camera frame B ψst Transformation matrix form the NED frame to the half NED frame C a Camera intrinsic parameter matrix F Feedback gain matrix H Matrix of the CCD camera model I Identity matrix with appropriate dimensions P Homogenous coordinates of the points in the image frame, pixel P P= ( P x P y 1 ) P i Homogenous coordinates of the i-th point in the image frame, pixel P g Coordinates of the points in the NED frame, m P gmc Coordinates of the tracker mass center in the NED frame, m P gt Coordinates of the points on the target in the NED frame, m P gti Coordinates of the i-th point on the target in the NED frame, m ΔP gti Coordinates of the i-th point on the target in the NED frame with respect to the tracker, m ΔP tracked Coordinates of the tracked point associated with the target in the NED frame with respect to the tracker, m T v Sampling period of the visual measurement, s ΔT h Demanded change of the thrust, N V a Air speed, m/s V g Ground speed of the tracker, m/s V gt Ground speed of the target, m/s V gg Velocity of the tracker with respect to ground in the NED frame, m/s V ggt Velocity of the target with respect to ground in the NED frame, m/s a gg Acceleration of the tracker in the NED frame, m/s 2 a ggc Expected acceleration of the tracker in the NED frame, m/s 2 a ggt Acceleration of...

show abstract

“…The estimates of range, rangerate and LOS rates have to be generated by an estimator, usually an EKF. Various implementations of the EKF are possible, employing different states 6,8 , and could be augmented with an adaptive element for robust estimation in the presence of unmodeled dynamics and disturbances 9 . The Image Processing block provides the position and size of the leader aircraft in the camera images.…”

Section: Figure 1 Closed-loop Uav Systemmentioning

confidence: 99%

6-DOF Nonlinear Simulation of Vision-Based Formation Flight

Sattigeri

Calise

Kim

et al. 2005

AIAA Guidance, Navigation, and Control Conference and Exhibit

View full text Add to dashboard Cite

This paper presents an adaptive guidance and control law algorithm for implementation on a pair of Unmanned Aerial Vehicles (UAVs) in a 6 DOF leader-follower formation flight simulation. The objective of the simulation study is to prepare for a flight test involving a pair of UAVs in formation flight where the follower aircraft will be equipped with an onboard camera to estimate the relative distance and orientation to the leader aircraft. The follower guidance law is an adaptive acceleration based guidance law designed for the purpose of tracking a maneuvering leader aircraft. We also discuss the limitations of a preceding version of the guidance algorithm shown in a previous paper. Finally, we discuss the design of an adaptive controller (autopilot) to track the commands from the guidance algorithm. Simulation results for different leader maneuvers are presented and analyzed. NomenclatureX = coordinate along inertial X-axis Y = coordinate along inertial Y-axis Z = coordinate along inertial Z-axis X & = velocity component along inertial X-axis Y & = velocity component along inertial Y-axis Z & = velocity component along inertial Z-axis a = body-axes acceleration vector f = body-axes specific force vector g = acceleration due to gravity p = roll-rate q = pitch-rate r = yaw-rate φ = roll attitude θ = pitch attitude ψ = heading angle α = angle of attack, subtended angle β = sideslip angle δ = actuator signal a τ = actuator model time constant d τ = actuator model time delay wo τ = washout filter time constant p K = proportional control gain I K = integral control gain d K = derivative control gain R = range between leader and follower X λ = relative azimuth angle between leader and follower XY λ = relative azimuth angle between leader and follower P = inertial position V = speed γ = flight path angle los û = unit vector from the follower to the leader aircraft ν = pseudo-control signal ad ν = neural network output h ν = pseudo-control hedging signal ∆ = inversion error, or modeling error due to approximate dynamic inversion n ω = natural frequency ζ = damping ratio ( ) ⋅ sat = linear saturation operator • = dot-product operator Proj = projection operator ( ) ⋅ ⋅, Subscripts e = elevator a = aileron r = rudder T = throttle cmd = command com = command x = projection along body X-axis y = projection along body Y-axis z = projection along body Z-axis L = Leader F = Follower wo = washout i = waypoint index min = minimum c = command filter, reference model

show abstract

An adaptive filtering approach to target tracking

Cited by 17 publications

References 39 publications

Neural Network based Adaptive Estimation and Guidance: Application to 2D Obstacle Avoidance

Neural Network based Adaptive Estimation and Guidance: Application to 2D Obstacle Avoidance

Vision based Target Tracking/ Following and Estimation of Target Motion

6-DOF Nonlinear Simulation of Vision-Based Formation Flight

Contact Info

Product

Resources

About