Robot People Tracking with UKF

[mkhali5]

What's the degree of nonlinearity of your system model operator and observation operator?

[belush]

There are more or less standard cartesian-polar conversions, including therefore trigonometric functions, squares, and so on.

Please see one of the recent papers in my homepage for more information.

[mkhali5]

What's the order of your state and observation vectors?

[belush]

5 for the state

2 for the legs observation with the laser

3 for the face observation with the camera

[mkhali5]

did you try using EKF, EnKF, PF?

[belush]

I tried EKF, UKF and SIR PF. A comparison can be seen on this paper (an extended version is under review for a journal):

N. Bellotto and H. Hu, People Tracking with a Mobile Robot: a Comparison of Kalman and Particle Filters, Proceedings of the 13th IASTED International Conference on Robotics and Applications (RA 2007), pp. 388-393, 29-31 August 2007, Würzburg, Germany

[belush]

journal version:

N. Bellotto and H. Hu, Computationally Efficient Solutions for Tracking People with a Mobile Robot: an Experimental Evaluation of Bayesian Filters, Autonomous Robots, Vol. 28, No. 4, pp. 425-438, 2010.

[decepto5]

Great work guys