IEEE Robotics & Automation Magazine - December 2018 - 60

Eng

age

ct
onta

C

Switch
Implement.

θb

Else

µc

ct
onta

Commands

Control Algorithms

Output

zsp := zb(te)

Contact
Controller

Input

Free-Flight
Controller

respect to yt b . We assume a constant distance between W b
and W e, annotated by L m .
Considering our application scenario, four phases for the
system can be identified: free flight, engage, contact, and disengage. The engage phase is the period during which the
approach is initiated up until the system is in stable contact.
The disengage phase occurs from the moment the separation
is initiated until the system has recovered its position setpoint, placed at a fixed distance C from the surface. At the
start of the engage phase, we require that the end effector be
near the surface (<30 cm).
Figure 5 illustrates the switching control strategy used durFigure 4. The end-effector prototype. Except for the metal
ing the different phases. Two separate controllers are used: a
components, motor, and flexible element, all parts were
free-flight controller and a contact controller. Depending on
produced using rapid-prototyping techniques.
the phase, different inputs are given to these controllers, and
different controller outputs are used. The same applies for the
Control Strategy
manipulator set-point. Both controllers receive the multiroWe define the frames in our system as illustrated in Figure 2. tor's state information: position p b, velocity v b, orientation
Frames W w, W b, and W e represent the inertial (world) frame, R wb , and angular rates ~ b .
the multirotor's body-fixed frame, and the end-effector frame,
As Figure 5 shows, the free-flight controller is used with
respectively. Frame W b has its origin in the center of mass of the manual set-points in the free-flight phase; during the disthe multirotor, with xt b aligned with the forward direction engage phase, the controller has a constant set-point C from
and zt b with the thrust vector. Frame W e is oriented as W w the surface, determined at the start of the disengage phase. In
and has its origin in the elastic component connecting the the engage phase, the implementation of the contact controlend effector to the manipulator. The roll, pitch, and yaw ler as presented in [2] is used, with the three desired body
angles of the multirotor around (xt b, yt b, zt b) are denoted by angles as reference inputs and the height set-point given by
(z b, i b, } b), respectively. We assume that the origin of frame z sp = z b (t e), where t e is the time the engage mode was
W e, p e always lies in the (xt b, zt b) plane. This implies that the entered. During the contact phase, the modified implementapitch angle and thrust of the multirotor define the magnitude tion of the contact controller, as described in the following, is
of the contact force. The manipulator angle n is given by the used so that the controller depends only on angular set-points
angle between the axis xt b and the vector vp be expressed in and angular state measurements. This controller is positionW b, and positive rotation is defined counterclockwise with independent, which allows the multirotor to track the ground
locomotion without active coordination between the set-points of the end
effector and the multirotor.
Free
Because a vertical surface is asFlig
psp := (xsp, ysp, zsp)
ht
sumed, the manipulator is given the
φcmd, θcmd,
psp := (xb(td) - C,
set-point n sp = i b in the free-flight,
age
ψcmd, Fu
yb(td), zb(td))
eng
s
i
engage and disengage phases, so that
D
Free
ψsp
the end effector is always in front of
Fligh
t
an
W b . The manipulator is assumed suffiDise d
ngag
ciently fast to accurately track i b . In
e
UAV State:
Mixer
pb, vb, Rb, ωb
the contact phase, the manipulator setge ct
ga
point is set to a specified value
En onta
dC
n sp = n c, which follows from the conan
ditions required for equilibrium.
φsp, θsp, ψsp
Motor

Manipulator
cmd (µsp)

C

Figure 5. A schematic illustrating the switching control strategy. Depending on the phase,
different inputs and outputs are selected.

60

*

IEEE ROBOTICS & AUTOMATION MAGAZINE

*

December 2018

Free-Flight Controller
The free-flight controller in the experiments described here consisted
of a generic cascaded position controller. It consists of four different
stages, where each subsequent stage
takes the output of the previous stage
IEEE Robotics & Automation Magazine - December 2018

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2018
