A Low-cost Compliant 7-DOF Robotic Manipulator
Morgan Quigley, Alan Asbeck, and Andrew Y. Ng
Abstract—We present the design of a new low-cost series-
elastic robotic arm. The arm is unique in that it achieves
reasonable performance for the envisioned tasks (backlash-free,
sub-3mm repeatability, moves at 1.5m/s, 2kg payload) but with
a significantly lower parts cost than comparable manipulators.
The paper explores the design decisions and tradeoffs made
in achieving this combination of price and performance. A
new, human-safe design is also described: the arm uses stepper
motors with a series-elastic transmission for the proximal four
degrees of freedom (DOF), and non-series-elastic robotics servos
for the distal three DOF. Tradeoffs of the design are discussed,
especially in the areas of human safety and control bandwidth.
The arm is used to demonstrate pancake cooking (pouring
batter, flipping pancakes), using the intrinsic compliance of the
arm to aid in interaction with objects.
Many robotic manipulators are very expensive, due to
high-precision actuators and custom machining of compo-
nents. We propose that robotic manipulation research can
advance more rapidly if robotic arms of reasonable perfor-
mance were greatly reduced in price. Increased affordability
can lead to wider adoption, which in turn can lead to
faster progress—a trend seen in numerous other fields .
However, drastic cost reduction will require design tradeoffs
There are numerous dimensions over which robotic arms
can be evaluated, such as backlash, payload, speed, band-
width, repeatability, compliance, human safety, and cost, to
name a few. In robotics research, some of these dimensions
are more important than others: for grasping and object ma-
nipulation, high repeatability and low backlash are important.
Payload must be sufficient to lift the objects under study.
Human-safety is critical if the manipulator is to be used in
close proximity to people or in classroom settings.
Some areas of robotics research require high-bandwidth,
high-speed manipulators. However, in many research set-
tings, speed and bandwidth may be less important. For
example, in object manipulation, service robotics, or other
tasks making use of complex vision processing and motion
planning, large amounts of time are typically required for
computation. This results in the actual robot motion requiring
a small percentage of the total task time. Additionally, in
many laboratory settings, manipulator motions are often
deliberately slowed to give the programmers time to respond
to accidental collisions or unintended motions.
In this paper, we present a robotic arm with similar
performance on many measures to high-end research robotic
Morgan Quigley, Alan Asbeck, and Andrew Y. Ng are with the Depart-
ment of Computer Science, Stanford University, Stanford, CA 94305, USA
mquigley, aasbeck, firstname.lastname@example.org
A spatula was used as the end effector in the demonstration application
described in this paper. For ease of prototyping, lasercut plywood was used
as the primary structural material.
The low-cost compliant manipulator described in this paper.
arms but at a drastically lower single-unit parts cost of $4135.
A shipped product must include overhead, additional
design expenditures, testing costs, packaging, and possibly
technical support, making a direct comparison with the parts
cost of a research prototype rather difficult. However, we
document the parts cost of our manipulator in order to give
a rough idea of the possible cost reduction as compared to
current commercially-available manipulators.
Our experiments demonstrate that millimeter-scale re-
peatability can be achieved with low-cost fabrication tech-
nologies, without requiring the 3-d machining processes
typically used to construct robotic manipulators.
A set of requirements were chosen to ensure the arm would
be useful for manipulation research:
• Human-scale workspace
• 7 Degrees of freedom (DOF)
• Payload of at least 2 kg (4.4 lb.)
– Compliant or easily backdrivable
– Flying mass under 4 kg
• Repeatability under 3 mm
• Maximum speed of at least 1.0 m/s
• Zero backlash
To meet these requirements at the lowest possible cost,
a new arm design was developed. The arm uses low-cost
stepper motors in conjunction with timing belt and cable
drives to achieve backlash-free performance, trading off the
cost of expensive, compact gearheads with an increased arm
volume. To achieve human safety, a series-elastic design was
used, in combination with minimizing the flying mass of the
arm by keeping the motors close to ground. The resulting
prototype is shown in figure 1.
A brief outline of this paper is as follows. Section II gives
an overview of other robotic arms used in robotics research.
Section III gives an overview of the design of the arm, and
discusses tradeoffs with its unique actuation scheme. Section
IV discusses the series compliance scheme, and sections V,
VI, and VII discuss the sensing, performance, and control,
respectively. Section VIII discusses application of the robotic
arm to a pancake-making task, followed by a conclusion.
II. RELATED WORK
A. Robotics research arms
There are a number of robotic arms used in robotics re-
search today, many with unique features and design criteria.
In this section, we discuss some recent widely-used and/or
influential robotic arms.
The Barrett WAM ,  is a cable-driven robot known
for its high backdrivability and smooth, fast operation. It has
high speed (3 m/s) operation and 2 mm repeatability.
The Meka A2 arm  is series-elastic, intended for human
interaction; other, custom-made robots with series-elastic
arms include Cog, Domo, Obrero, Twendy-One, and the
Agile Arm , , , , . The Meka arm and Twendy-
One use harmonic drive gearheads, while Cog uses plane-
tary gearboxes and Domo, Obrero, and the Agile Arm use
ballscrews; the robots all use different mechanisms for their
series elasticity. These arms have lower control bandwidth
(less than 5 Hz) due to series compliance, yet that has
not appeared to restrict their use in manipulation research.
Several human-safe arms have been developed at Stanford
using a macro-mini actuation approach, combining a series-
elastic actuator with a small motor to increase bandwidth
The PR2 robot ,  has a unique system that uses a
passive gravity compensation mechanism, so the arms float
in any configuration. Because the large mass of the arm is
already supported, relatively small motors are used to move
the arms and support payloads. These small motors provide
human safety, as they can be backdriven easily due to their
low gear ratios.
The DLR-LWR III arm , Schunk Lightweight Arm
, and Robonaut  all use motors directly mounted
to each joint, with harmonic drive gearheads to provide
fast motion with zero backlash. These arms have somewhat
higher payloads than the other arms discussed in this section,
ranging from 3-14 kg. They are not designed for human
safety, having relatively large flying masses (close to 14
kg for the DLR-LWR), although demonstrations with the
DLR-LWR III have been performed that incorporate a distal
force/torque sensor that uses the arm’s high bandwidth to
quickly stop when collisions are detected.
Of the robotic arms discussed previously, those that are
commercially available are all relatively expensive, with end-
user purchase prices well above $100,000 USD. However,
there are a few examples of low-cost robotic manipulators
used in research. The arms on the Dynamaid robot 
are constructed from Robotis Dynamixel robotics servos,
which are light and compact. The robot has a human-scale
workspace, but a lower payload (1 kg) than the class of arms
discussed previously. Its total cost is at least $3500 USD,
which is the price of just the Dynamixel servos. In videos
of it in operation, it appears to be slightly underdamped.
The KUKA youBot arm is a new 5-DOF arm for robotics
research . It has a comparatively small work envelope of
just over 0.5 m3, repeatability of 0.1 mm, and payload of
0.5 kg. It has custom, compact motors and gearheads, and is
sold for 14,000 Euro at time of writing.
B. Robot arms using stepper motors
Many robot arms have been made using stepper motors.
Pierrot and Dombre ,  discuss how stepper mo-
tors contribute to the human-safety of the Hippocrate and
Dermarob medical robots, because the steppers will remain
stationary in the event of electronics failure, as compared to
conventional motors which may continue rotating. Further-
more, they are operated relatively close to their maximum
torque, as compared to conventional motors which may
have a much higher stall torque than the torque used for
ST Robotics offers a number of stepper-driven robotic
arms, which have sub-mm repeatability . However, these
are not designed for human safety. These are also rela-
tively low-cost, for example the R17 arm (5-DOF, 0.75m
workspace, 2 kg payload) is listed for $10,950 USD. Several
other small, non-compliant robots were made in the 1980s-
1990s used for teaching were also driven by stepper motors
. For example, the Armdroid robotic arm is 5-DOF and
has 0.6m reach; it uses steppers with timing belts for gear
reduction, then cables to connect to the rest of the arm .
III. OVERALL DESIGN
The arm has an approximately spherical shoulder and an
approximately spherical wrist, connected by an elbow. The
joint limits and topology were designed to enable the robot to
perform manipulation tasks while being mounted near table-
height, as opposed to anthropomorphic arms, which must
hang down from the shoulder and require the base of the
arm to be mounted some distance above the workspace.
The shoulder-lift joint has nearly 180 degrees of motion,
allowing the arm to reach objects on the floor and also
work comfortably on tabletops. A summary of the measured
properties and performance of the arm is shown in table I.
MEASURED PROPERTIES OF THE ARM
1.0m to wrist
Fig. 2.Actuation scheme for each of the proximal four DOF.
A. Actuation scheme
Figure 2 shows the actuation scheme for the proximal four
DOF. These joints are driven by stepper motors, with speed
reduction accomplished by timing belts and cable circuits,
followed by a series-elastic coupling. Using only timing belts
and cable circuits in the drivetrain results in low friction,
minimal stiction, and zero backlash. This enables the arm
to make small incremental motions (less than 0.5mm), and
there is no gearing to damage under applied external forces.
Combined with stepper motors, which have high torque
at low speeds, this leads to a low-cost but relatively high
performance actuation scheme. A downside to this scheme
is that the reduction mechanisms occupy a relatively large
volume, making the proximal portion of the arm somewhat
Using a two-stage reduction of timing belt followed by
cable circuit accomplishes not only a larger gear reduction
than a single stage, but also enables the motors to be located
closer to ground. The motors for the two most proximal DOF
are grounded, and the motors for the elbow and upperarm roll
joints are located one DOF away from ground. By placing the
relatively heavy stepper motors close to ground, the flying
mass of the arm is greatly reduced: below the second (lift)
joint, the arm is 2.0 kg. For comparison, a typical adult
human arm is about 3.4 kg .
The two-stage reduction scheme leads to coupling between
the motions of joints 1 and 2, and joints 2, 3, and 4. However,
this coupling is exactly linear and can easily be estimated as a
feedforward term in software. The routes of the timing belts
and cables can be seen in figure 3. After the timing belts
and cable circuits, the proximal four DOF have series elastic
couplings between the cable capstan and the output link,
discussed in section IV. These are used to provide intrinsic
compliance to the arm, as well as providing force sensing
The distal three DOF are driven by Dynamixel robotics
RX-64 servos. These joints do not have compliance aside
from limiting the torques. However, the compliance of the
proximal four DOF allows the end effector to be displaced
in Cartesian space in three dimensions, barring kinematic
singularities where only two dimensions will be compliant.
B. Tradeoffs of using stepper motors
Using stepper motors as actuators has a number of ad-
vantages. Stepper motors excel at providing large torques
at low speeds, which is the target regime of the arm. They
require a relatively low gear reduction, which can be accom-
plished with timing belts and cable drives. In the prototype
shoulder roll, and elbow joints. All belt routes rotate about the shoulder lift
joint. The elbow cables twist about the shoulder roll axis inside a hollow
shaft. Best viewed in color.
Cable routes (solid) and belt routes (dashed) for the shoulder lift,
Fig. 4.Compact servos are used to actuate the distal three joints.
manipulator discussed in this paper, the effective reductions
were 6, 10, 13, and 13, respectively, for the first four joints.
DC motors, for comparison, generally require a significantly
larger gear reduction through a gearbox that would be either
susceptible to backlash or moderately expensive.
The stepper motors also act as electromagnetic clutches,
improving safety if large forces are accidentally applied at
the output. If a force is applied that causes a stepper to
exceed its holding torque, the stepper motor will slip and the
arm will move some distance until the force is low enough
that the stepper can re-engage. The stepper holding torque is
approximately 60% more than the maximum moving torque
(and hence the maximum payload of the arm), large enough
to avoid needlessly slipping but small enough to make the
However, there are a few downsides of the steppers acting
as an electromagnetic clutch. First, if a stepper motor slips,
the arm may need to be re-calibrated. The arm uses joint-
angle encoders for state estimation, so closed-loop position
control can still be done even after a slip, but force sensing
will be miscalibrated (see section V). Second, the arm may
move suddenly after a stepper motor slip. The arm only
slips if relatively large amounts of force are applied, and
after a slip the steppers initially provide little resistance.
The moving arm may collide with other objects or people;
this is mitigated by making the arm as light as possible.
Adding backshaft encoders to the stepper motors would
enable tracking of the motor position even during rotor slips,
and enable faster stoppage of a slipping motor. Whether or
not the additional cost is justified depends on the task and the
anticipated frequency of unintended high-speed collisions.
As envisioned, stepper slips occur only as a final layer
of safety, and thus are not anticipated to be a frequent
C. Hybrid SEA/non-SEA actuation scheme
The actuation scheme of the proposed manipulator uses
series-elastic actuators (SEA) in the proximal 4 DOF, but
non-series-elastic actuation for the distal 3 DOF. The band-
width of the distal 3 DOF is somewhat higher than the
bandwidth of the proximal 4 DOF, permitting a restricted set
of higher-frequency motions. This is similar to that described
in , which employs a macro-mini actuation scheme for
the most proximal DOF and conventional actuators for the
more distal DOF.
In our scheme, the lower three DOF still get most of
the benefits of the series-elastic upper arm, including the
ability to control forces by modulating a position. The main
downside of this as compared to a full series-elastic scheme
is that the gears in the distal DOF are more affected by shock
loads, since (in the worst case) the mass of the entire arm is
past the series compliance.
D. Arm inertia and series elastic stiffness
One important tradeoff with a series-elastic robot arm is
the arm inertia and series elastic stiffness. Consider a one-
DOF arm with moment of inertia I [kg m2] driven by a rotary
joint with torsional stiffness kθ[N m/radian]. The arm will
oscillate at its natural frequency, which is f0=
If the arm has a low inertia or the series elastic coupling is
stiff, the motor driving the arm may not have enough torque
or bandwidth to compensate for this oscillation. Pratt and
Williamson suggest increasing the arm’s inertia to eliminate
this effect ; other options are to reduce the spring
constant; include damping in the series-elastic coupling; or
increase bandwidth by decreasing the motor gear reduction,
at the cost of a lower payload. For human-safe robotic arms
with low inertia, this issue can be significant.
In our arm, considering the elbow joint, the natural fre-
quency is around f0= 5.1 Hz, with kθ= 86 N m/radian
and I = 0.083 kg m2. This is close to the bandwidth of the
motors with our current gear reduction.
E. Low-cost manufacturing
Several methods were used to achieve a low-cost design.
The total cost of all of the stepper motors was $700. An
COST BREAKDOWN OF THE ARM
alternative with comparable speed/torque performance is
to use DC brushed motors with planetary gear reduction.
Although they are available for a comparable price, their
inexpensive gearheads exhibit more than 1 degree of back-
lash. High-performance gearheads or brushless motors would
increase the cost by a factor of at least two. For example,
a single zero-backlash harmonic drive actuator costs over
$1000 USD, and a brushless planetary gearmotor of sufficient
torque and 0.75-degree backlash costs $500 USD.
Lasercutting 5-ply plywood was used for most of the
structure in the current prototype. The lasercutter used (Beam
Dynamics OmniBeam 500, 500 Watts) can produce toler-
ances of 0.025mm, and excellent results were also achieved
with an Epilog Legend Helix 24 (45 Watt) laser cutter.
Dovetailing of the wood pieces was done, enabling them
to be press-fit together, and flanged bearings and shafts
were also press-fit into holes. It is unknown how the wood
structure will respond to large temperature and humidity
variations, but in a typical lab environment these are held
relatively constant. Wood is an excellent material for rapid
prototyping, and is rigid enough to meet the repeatability
design requirements. In the future, we intend to make the
complete structure out of folded sheet metal for a more
durable structure. The lower arm of the robot was made of
folded sheet metal as a first step in this direction. Folded
metal structures cannot be made to the precision of custom-
machined parts, but calibration techniques can be used to
compensate for manufacturing errors.
The other technique used to keep costs low was to avoid
all custom machining except for the lasercut structure; all
other parts were off-the-shelf. A breakdown of the parts cost
for the robot is shown in table II. Not included in this list
are the costs of laser cutter time and assembly time; laser
cutting would take 2.5 hours and assembly would take around
15 hours for additional copies of the arm.
IV. SERIES COMPLIANCE
The robot uses a compliant coupling in the proximal four
joints. This provides increased human safety, allows the arm
to be compliant even though the stepper motors are not
backdrivable, and is used for force sensing as the deflection
across the compliance is measured.
A diagram of the compliant coupling is shown in figure 5.
Its operation is similar to the elastic couplings described in
, , . At the joint, a capstan used in the cable
circuit (labeled 1 in figure 5) is suspended via bearings on
the same shaft as the output link (2). The capstan is then
no external forces. Right, an applied force causes rotation.
Diagram of the series compliance. Left, compliant coupling with
polyurethane in the series compliance. The joint was quasi-statically moved
through 70% of its normal operating range.
Stiffness of the elbow. Some hysteresis is exhibited due to the
connected to the output link through the compliant element.
Two plates connected to the output link extend through the
middle of the capstan, which has two holes cut through the
middle of it. Each hole contains a polyurethane tube (3),
which is compressed between the plate from the output link
and the side of the hole in the capstan. In figure 5(right),
the capstan (4) is held stationary while an external force
(F) is applied. This causes one polyurethane tube (5) to
compress while the other (6) expands. The polyurethane
tubes are initially pre-compressed to slightly more than half
of their maximum possible compression so that they will
always remain in compression as the output link moves with
respect to the capstan.
Polyurethane was used to provide some mechanical damp-
ing of the joint, which gives the arm some hysteresis but
helps eliminate oscillations. However, springs can readily be
used in their place. Tubes were used instead of rods or balls
to give the output links around 4 degrees of compliance in
each direction, which requires several millimeters of travel.
Figure 6 shows the stiffness and hysteresis of the compliant
coupling in the elbow joint.
As previously discussed, the first four joints of the manip-
ulator are actuated by relatively large stepper motors embed-
ded in the base and shoulder. The intrinsic stability of stepper
motors forms a key aspect of the sensing strategy: assuming
the stepper motors do not slip, the series of step motions
the motors undergo can be precisely integrated to give the
input displacement. Joint angles are measured directly using
optical encoders. The deflection of the compliant element can
thus be measured as the difference of the (post-reduction)
motor position and the joint angle, thus permitting force
Integration of the motor step counts occurs on embedded
microcontrollers in the first two links of the manipulator. This
integration commences at power-up, and thus the motor step
integration is best seen as a relative position estimate. To
estimate the position offsets, enabling comparison with the
(indexed) absolute joint-angle encoders, the manipulator is
driven to the index pulses and held stationary. The stepper
count when the manipulator is stationary at all encoder index
pulses can be taken as a static offset to permit force-sensing
calibration, barring hysteresis or plastic deformation of the
The distal three joints are actuated by Robotis Dynamixel
RX-64 servos, which feature internal potentiometers with a
usable range of 300 degrees. The potentiometer voltage is
internally sampled by the servo.
To simplify the manipulator wiring, the stepper-motor
drivers and servos share a common RS-485 bus. Sensors are
sampled and actuators are commanded at 100 Hertz.
In the future, initial static pose estimation will be provided
by accelerometers , enabling generation of safe trajecto-
ries to reach the encoder index pulses.
The arm’s performance on several metrics was measured.
Closed-loop repeatability was tested by moving the arm
alternately between a home position and eight locations
distributed around the workspace. The repeatability at the
home position is shown in figure 7, where the position of the
arm is plotted after it returned from each alternate location,
as measured by an optical tracking system.
The encoders can register changes of 0.036 degrees, which
corresponds to 0.64mm at the base joint with the arm fully
extended. The stepper motor at the base joint can command
changes of 0.52mm at the end effector. Moving down the
arm, each subsequent motor can command sequentially finer
motions due to increased effective gear ratios and shorter
distances to the end effector.
Payload was measured by adding weights until the step-
pers slipped when slowly moving through the worst-case
configuration. Maximum velocity was measured by com-
manding the fully-extended arm to move upwards at the
maximum rate of the stepper controllers while observing the
end-effector velocity with an optical tracking system. These
experiments demonstrated a maximum payload of 2.0 kg and
a maximum velocity of 1.5 m/s.