⇦
elapsed time: 0 seconds
current phase: (A)


elapsed time: 1 seconds
current phase: (A)


elapsed time: 2 seconds
current phase: (A)


elapsed time: 3 seconds
current phase: (A)


elapsed time: 4 seconds
current phase: (A)


elapsed time: 5 seconds
current phase: (A)


elapsed time: 6 seconds
current phase: (B)


elapsed time: 7 seconds
current phase: (R)


elapsed time: 8 seconds
current phase: (R)


elapsed time: 9 seconds
current phase: (R)


elapsed time:10 seconds
current phase: (R)


elapsed time:11 seconds
current phase: (R)


elapsed time:12 seconds
current phase: (R)


elapsed time:13 seconds
current phase: (R)


elapsed time:14 seconds
current phase: (R)


elapsed time:15 seconds
current phase: (R)


elapsed time:16 seconds
current phase: (L)


elapsed time:17 seconds
current phase: (L)


elapsed time:18 seconds
current phase: (L)


elapsed time:19 seconds
current phase: (L)


elapsed time:20 seconds
current phase: (L)


elapsed time:21 seconds
current phase: (L)


elapsed time:22 seconds
current phase: (L)


elapsed time:23 seconds
current phase: (L)


elapsed time:24 seconds
current phase: (L)


elapsed time:25 seconds
current phase: (L)


elapsed time:26 seconds
current phase: (L)


elapsed time:27 seconds
current phase: (L)


elapsed time:28 seconds
current phase: (L)


elapsed time:29 seconds
current phase: (L)


elapsed time:30 seconds
current phase: (L)




FUSION 2022 sinulation

In this (single point object) toy example I try to mimic the motion of a sport car / F1 performing a maneuver composed by a couple of opposite hairpin corners, generated by quite signif‌icant centripetal accelerations.
The ground truth motion is split in four phases, each of them with their proper accelerations:

phaselongitudinal
acceleration		centripetal
acceleration
(A)-cceleration1.5g0.0g
(B)-rake-2.0g0.0g
(R)-ight turn-0.2g-2.0g
(L)-eft turn0.3g2.0g

Here "g" is the gravitational acceleration (≈9.81 m/s2). The simulation is in real time.
The white track represents the real object track, while the colored tracks are the output of the trackers involved (wich are fed with the very same input measurments - the yellow stars). The trackers share the same correction step (Kalman corrector) and they dif‌f‌ers only in the prediction step adopted. More precisely, the following predictors are considered:

Λ:OT2:2[Tesori et al.]
Coordinated TurnPCT[Roth et al.]
Linear Kinematic ModelNCJ[Bar-Shalom et al.]
(model family)(specif‌ic model)(ref‌erence)

⇨