Flexible Vision-based control of a 6DOF quadrotor Chayatat Ratanasawanya May 18, 2011
Jan 18, 2018
Flexible Vision-based controlof a 6DOF quadrotor
Chayatat RatanasawanyaMay 18, 2011
2
OverviewRecallsProgress & AchievementResults
3
Recall…Develop a flexible human/machine control
system to hover an UAV (6-DOF helicopter) carrying a VDO camera beside an object of interest; e.g. a window.
Method: Human control – Joystick Machine control – Visual-servoing
4
Recall…Quanser provided height(Y), X&Z position,
and yaw controllersMap image info to positional infoPOSIT algorithmTarget object
5
SuccessfullyHeight is always control by sonar feedbackFlexible human/machine control
QballPID
IMURoll,
PitchOptitrack
Magnetometer
X*, Z*
Yaw*
Single Camera
In-flight modifiable
LQR
Roll*, Pitch*
6
Reference case: OptitrackVDO
7
Reference case: Optitrack0.02
-0.13
0.29
-0.01
15.5⁰
-5.5⁰
8
Result: Flexible controlVDO
9
Result: Flexible control0.7
0.3
2.3
2.0
9⁰
-5⁰
10
Result: Multiple targetsVDO
11
Result: Multiple targets0.7
0.3
2.3
2.0
9⁰
-5⁰
12
ConclusionImage information is mapped to positional
control inputs via POSIT algorithmResult is position-based visual servoingError is defined in object frame; NOT world
frameAchieved human-machine control flexibility Able to change desired X, Z, Yaw positions in
flightAble to hover in front of multiple targets (one
at a time). Model of each target must be known.
13
Questions?