using two of the cycloidal gears on the hips, and two on
the knees,” noted Sahin.
RIT is looking for sponsorship from the Harmonic Drive
gear manufacturer. The school already has a sponsorship
from a cycloidal gear maker. “However, we have purchased
sample gears from each type of company and will be
testing them soon,” commented Sahin. RIT students are
testing the gears on one leg, knowing that what will work
on one leg, will work on the other.
All four of the previous TigerBot designs range in size
from 2. 5 to above four feet, each with functional walking
gaits that RIT developed using inverse kinematics
algorithms. Using these algorithms enables RIT students to
build increasingly larger robots without changing more than
a few inputs to the calculations. By simply changing the size
of the links in the algorithm, the algorithm can produce
calculations for correct joint angles for the given humanoid
robot motion the roboticists desire. “TigerBot II had an ROS
(Robot Operating System) based inverse kinematics
algorithm with successful operation,” said Sahin.
RIT students searched for servo products and found the
ClearPath: an all-in-one product that incorporates a three-phase permanent magnet brushless servo motor; a high
resolution optical encoder; a DSP-based, all-digital vector
servo drive; and a motion controller. “The size,
performance, and overall power capacity of these servos
were very attractive as space is at a premium inside the
robot. With ClearPath, we don’t have a servo drive to install
and there’s no motor cable between the drive and motor in
this servo product,” Sahin explained.
The all-in-one servo for OEM applications offers the
highest commercially available power density in the
fractional horsepower class. “It uses a servo compensator
with advanced feed forward gains and numerous
proprietary heuristics, and adaptive gains as well as
nonlinear extensions. Its three-phase vector torque control
provides fast torque response, independent of rotational
speeds,” commented Sahin.
The folks at Teknic responded to RIT’s request for
corporate sponsorship and because they are a local
company, they were able to send out two engineers to visit
the campus to do design reviews and trainings. “They also
recommended a couple of vendors for gearbox
technology,” said Sahin.
RIT expects both the hardware and the software in the
new human scale TigerBot to turn out uniquely different
from those of the previous TigerBot iterations. The TigerBot
will have higher quality motors, stronger gearboxes, and
minimal backlash — no backlash will really spread to the rest
of the robotic system thanks to using these motors and
gearboxes. The TigerBot already has an ROS software
implementation for everything, including the test joints. So,
To have tour guide capabilities, the RIT TigerBot will
eventually have to be able to walk safely and smoothly
throughout the halls for the area covered. “It will also have
to have smooth interaction with visitors when they ask a
question,” remarked Sahin.
“We are in the middle phase of the multi-year TigerBot
development project. We have designed humanoids with 22
joints and have explored several manufacturing techniques
and motor-to-gear pairs,” commented Sahin. Now comes
the third and final phase where the robot grows to human
scale, and proposes new opportunities and new challenges
for expansion as a platform.
“I believe once the physical implementation of the
robot is completed, we will be able to complete the
algorithms for safe navigation and human interaction (using
recordings of the lab info, voice recognition, and speakers
for one word interactions),” said Sahin. RIT must explore,
test, and achieve the right balance of structural materials,
gears, and motors for a stable predictable robot that will be
safe in a human environment. For that, more work must be
done in the balancing, walking, navigation, and guide
aspects of the robot.
TigerBot will ultimately house and use a number of
different advanced sensor technologies for navigating in
and interacting with its environment.
“We are exploring sensors including accelerometers,
gyros, force sensors, current sensors, IR distance sensors,
and ultrasonic distance sensors. We are considering
Microsoft Kinect for 3D mapping and localization,”
disclosed Sahin. The robot will also use a touch screen
interface, voice recognition technology, and voice-to-text
The inverse kinematics algorithm uses an ROS library
which RIT has modified to suit the TigerBot’s dimensions.
RIT replicated the robot in a simulation tool with a physical
world engine that is entirely compatible with ROS.
Further, RIT developed an algorithm for TigerBot II that
balances the robot using accelerometers, gyros, and current
sensors for the servomotors. “We will be expanding upon
this successful approach,” concluded Sahin.
Named after the RIT Tigers men’s ice hockey team at
the Institute, TigerBot will instill pride in the school with its
unique capabilities and offerings. SV
12 SERVO 07.2015