I have an athlete's movement data [3 columns Position (x,y) and orientation (θ)]. Using particle filter I want to predict Position, Velocity, and Acceleration. I also need the state transition function and measurement likelihood function. Something similar to the example "Track a Car-Like Robot Using Particle Filter" but input is only (x, y, θ). Thanks!
General speaking, filtering based method such as partical or ukf, ekf. You should to divided into 2 parts: prediction, correction.
So you should use some kind of motion model for predicting: xyz,theta next= f(xyz.._ previous). Then use another sensor to correct by update partical weights.
Regards,
You can find an implementation of Particule Filters to mobile robot motion ( very close to your problem) , in "Probabilistic Robotics" book by Sebastien Thurn.
- Badges
- Science topic
- Similar topics
- Psychology