Robocart: Autonomous Ground Vehicle Electromechanical Foundations Design Written by Prateek Sahay (RBE/ME) Advised by Professor Alexander Wyglinski (ECE), Advisor-of-Record Professor Taskin Padir (RBE/ECE), Co-advisor August 2014–April 2015 A Major Qualifying Project Submitted to the Faculty of Worcester Polytechnic Institute in Partial Fulfillment of the Requirements for the Degree of Bachelor of Science. This report represents the work of WPI undergraduate students submitted to the faculty as evidence of completion of a degree requirement. WPI routinely publishes these reports on its website without editorial or peer review. For more information about the projects program at WPI, please see http://www.wpi.edu/academics/ugradstudies/project-learning.html.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Professor Alexander Wyglinski (ECE), Advisor-of-RecordProfessor Taskin Padir (RBE/ECE), Co-advisor
August 2014–April 2015
A Major Qualifying Project Submitted to the Faculty of Worcester Polytechnic Institute inPartial Fulfillment of the Requirements for the Degree of Bachelor of Science.
This report represents the work of WPI undergraduate students submitted to the faculty asevidence of completion of a degree requirement. WPI routinely publishes these reports on its
website without editorial or peer review. For more information about the projects program atWPI, please see http://www.wpi.edu/academics/ugradstudies/project-learning.html.
1 Overview of the different systems involved in making the golf cart autonomous. 52 CAD prototypes of the three mechanical subsystems in Robocart. . . . . . . . 63 Final mounted designs of the three mechanical subsystems in Robocart. . . 8
1.1 IHS’s prediction of the evolution of self-driving cars. . . . . . . . . . . . . . . 10
2.1 The three main mechanical components with respect to the overall system. 152.2 The rack and pinion steering mechanism in the golf cart. . . . . . . . . . . . . 222.3 The pair of universal joints in the golf cart steering mechanism. . . . . . . . 232.4 A SolidWorks Photoview rendering of the golf cart CAD model. . . . . . . . . 25
3.1 The physical golf cart we worked on. . . . . . . . . . . . . . . . . . . . . . . . . 273.2 Last year’s Raspberry Pi mount design. . . . . . . . . . . . . . . . . . . . . . . . 283.3 The new design drafted in CAD. . . . . . . . . . . . . . . . . . . . . . . . . . . . 293.4 Last year’s design was a linear actuator which pushed against one of the
wheels. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303.5 The steering column is perpendicular to the hood support, not the dashboard. 323.6 Simplified diagram of the normal functioning of the brakes. . . . . . . . . . . 343.7 Before and after comparison of Raspberry Pi mount designs. . . . . . . . . . 353.8 Simplified diagram of the revamped brakes design. . . . . . . . . . . . . . . . 363.9 The Bosch van door motor chosen for the automated braking system. . . . . 373.10 CAD model of the first design of the automated braking system. . . . . . . . 383.11 CAD model of the second design of the automated braking system. . . . . . 393.12 CAD model of the third design of the automated braking system. . . . . . . . 40
4.1 The fourth iteration of the Raspberry Pi mount design with locking bolts. . 424.2 The foot was identified as a break point and fixed for the final design. . . . 434.3 The final Raspberry Pi mount design attached to the golf cart. . . . . . . . . 434.4 The steering column was not perpendicular to the front dashboard of the cart. 444.5 Motor chart for the motor used in controlling the steering mechanism. . . . 454.6 The 10-tooth sprocket was found to not be an exact fit for the motor shaft. 464.7 An error in the CNC machine resulted in a badly cut plate. . . . . . . . . . . . 474.8 Drilling the hinge hole pattern in the correctly cut steel plate. . . . . . . . . . 484.9 Using washers would result in torques and deformations on the frame. . . . 494.10 An angle grinder was used to grind down the lip of the frame of the golf cart. 494.11 The steering column mounting tube had to be cut on a horizontal bandsaw. 50
3
4.12 The steel plate had to be machined in a CNC mill to avoid the steering column. 514.13 The clearance between the sprockets was small so extra measures were taken. 514.14 A plate was laser-cut from acrylic to hold the two sprockets apart. . . . . . . 524.15 Final assembly. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 524.16 The potentiometer assembly. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 544.17 Completely assembled steering system. . . . . . . . . . . . . . . . . . . . . . . . 554.18 Brakes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564.19 The completed brake design mounted to the underside of the golf cart. . . . 57
5.1 Free body diagram of the automated steering mechanism. . . . . . . . . . . . 615.2 Bond graph of the automated steering mechanism. . . . . . . . . . . . . . . . 625.3 Free body diagram of the automated braking mechanism. . . . . . . . . . . . 635.4 Bond graph of the automated braking mechanism. . . . . . . . . . . . . . . . . 63
4
Abstract
Networked autonomous vehicles have long been a dream, promising traffic-less cities andsafer roads. This MQP laid the foundations for a vision-based autonomous ground vehi-cle. The team this year outfitted a 1995 golf cart with sensors and motors to automatethe steering and brakes to lay a strong foundation for future teams. Additional work wasdone mounting of two stereoscopic cameras and a display for computer feedback as wellas foundational work on the controls algorithms.
List of Acronyms
ALV Autonomous Land Project
CAD Computer-aided design
CAM Computer-aided manufacturing
CMU Carnegie Mellon University
CNC Computer Numerical Control
DARPA Defense Advanced Research Projects Agency
DRC DARPA Robotics Challenge
EBS Electronic Braking System
ECU Electronic control unit
GPS Global Positioning System
LIDAR Light image detection and ranging
MQP Major Qualifying Project
RALPH Rapidly Adapting Lateral Position Handler
ROS Robot Operating System
WPI Worcester Polytechnic Institute
2
Acknowledgments
My heartfelt thanks go out to everyone who made this project possible. To those who helped
me in times of need as I tackled a project like none other before, Matthew Forman, Kevin
Arruda, and Professor Stafford: your generosity and patience were a godsend. Words are
beyond me as I attempt to thank Joe St. Germain from the RBE lab and Bob Boisse from the
ECE shop for endlessly lending me their tools and expertise. Additional thanks to Tracey
Coetzee in the RBE department for all her help with budgeting and preparations for Project
Presentation Day. Lastly, I’d like to thank Professor Wyglinski for his unwavering guidance
and enthusiasm, and for his dream, without any of which I could never have grown as
much as I did this year. I wish him the best of luck with this ambitious project.
3
Executive Summary
Motivation
Autonomous vehicles have long been a dream. The motivations behind them are clear; the
promise of saving 1.2 million lives a year [1] and solving traffic congestion problems [2]
has struck a chord with scientists, engineers, and programmers around the world.
With the price of sensors and computing steadily decreasing year after year, affordable
self-driving automobiles are starting to become a reality. This MQP aims to imagine a
divergent take on autonomous vehicle technology by challenging modern vision algorithms
combined with affordable sensing technology.
Sensors are a big part of autonomous navigation. Although sensors like LIDAR have
dominated recent ventures, their price, bulk, and vulnerable location atop autonomous
vehicles could hinder their uptake in the future. As a result, this project sought instead to
pursue vision-based object-recognition and navigation.
Proposed Approach
For this project, a 1995 electric golf cart was also acquired for prototyping with. Significant
upgrades had to be made to its design in order for it to become autonomous. Because of
the large scale of this project, the project was broken down into separate subsystems, each
of which was worked on separately to be combined sometime in the future. The different
4
Figure 1: Overview of the different systems involved in making the golf cart autonomous.
systems requiring modifications are shown in Figure 1.
The scope of this paper involves the three mechanical subsystems—the stereoscopic
cameras, the automated steering mechanism, and the automated braking mechanism. Each
of these subsystems posed their own unique challenges and were dealt with very differently.
Raspberry Pi cameras were used for the steroscopic cameras, since they were cheaply
available and readily able to be connected to a wireless network through Raspberry Pi
computers. A mounting solution was required for the Raspberry Pi cameras and computers
which needed to fulfill the following requirements:
• Hold two Raspberry Pis
• Hold two Raspberry Pi camera modules
• Ensure the two cameras face the same direction
• Allow the the Raspberry Pi’s and cameras to be easily removable (to allow the Rasp-
berry Pis to be debugged more easily off the cart)
5
(a) Raspberry Pi mount. (b) Automated steering. (c) Automated braking.
Figure 2: CAD prototypes of the three mechanical subsystems in Robocart.
• Protect the cameras without obstructing the view of the cameras
• Allow access to all necessary ports of the Raspberry Pi
Next, the steering system needed to undergo some major changes to allow the computer
to control it. The requirements for the system were listed as follows:
• Supply at minimum 9.19 ft-lbs of torque to the steering column
• Allow the wheels to be turned from fully left to fully right in about 1 second
• Must be back-drivable
Finally, the brakes on the golf cart also needed to be automated in order for the computer
to control them. The requirements of the new system were:
• Fully engage in under 0.5 seconds
• Supply a force of 100 lbs to the brake cables
• Must be back-drivable
• Negative braking
Part of the approach involved drafting CAD prototypes for each of these three systems
to ensure the different parts would fit together and function without issues. Different iter-
ations of the designs for each system are shown in Figure 2.
6
Discussion and Results
The Raspberry Pi mounting system took a unique approach to the design. Once the de-
sign was drafted in CAD, the parts were individually laser-cut from acrylic and assembled
together. A spring-loaded mechanism ensured the electronic components would stay to-
gether, but also be easy to remove to be worked on elsewhere. All of the listed requirements
were successfully met.
The steering system was more difficult to implement. A motor was added to the dash-
board area of the golf cart with a chain-and-sprocket system to supply the steering column
with sufficient torque and speed to turn it from fully left to fully right in 0.9 seconds. A
rotational sensor was added with its own sprocket for position feedback data. The torques
were also determined such that a person of average strength could overpower the motor,
thereby making the system back-drivable. All of the listed requirements of the system were
successfully met.
Finally, the braking system was tackled. Since it had been decided the system should be
back-drivable, a new mechanism was developed for this system whereby the motor could
actuate the brakes, but a human passenger could also apply the brakes at any time without
having to fight the motor at all. The brakes could be fully actuated within 0.2 seconds.
Unfortunately, it seemed, even after much design work, that it was impossible to meet
the requirement of implementing negative braking—the only viable way of including this
feature was by converting the brakes to a hydraulic system, which was deemed too costly
for this MQP. Thus, this requirement had to be dropped. However, out of the remaining
listed requirements for this system, all of them were successfully met.
Conclusions and Recommendations
This project is a very ambitious one, and it is regretful that more of the subsystems could
not be completed this year, and that the team never got to see the golf cart run. Before
7
(a) Raspberry Pi mount. (b) Automated steering. (c) Automated braking.
Figure 3: Final mounted designs of the three mechanical subsystems in Robocart.
Robocart can begin to drive autonomously, however, there is still plenty of work to be done
on the various subsystems. For one thing, more sensors are required for the computer to
be able to intelligently allow Robocart to navigate autonomously. Not only would some
cheap LIDAR or ultrasonic distance-sensing be useful, but also possibly encoders on the
wheels, and limit switches on the steering and braking systems for safety reasons. There
still remains plenty of work to be done on the software side of the robot, as well. Faster and
more accurate algorithms for object-recognition and navigation will continue to evolve, and
Robocart should be allowed to evolve with them. Some mechanical improvements would
also make Robocart function better, such as replacing the brake system with a hydraulic one.
A hydraulic system would also fulfill the requirement of implementing negative braking,
which was unfortunately unable to be done this year due to its sheer cost.
All-in-all, however, the mechanical upgrades done to Robocart this year can be consid-
ered a success. I’m very grateful for the help I received not only from Professor Wyglinski,
but also from other professors around campus, staff at Washburn, and lab managers. Fi-
nally, I’m also very grateful for the collaborative and nurturing environment at WPI, without
which this project would not have been possible.
8
Chapter 1
Introduction to Autonomous Vehicles
Autonomous vehicles have been in development for over 65 years—in fact the first cruise-
control systems were introduced in 1948 [3]. Multiple car manufacturers estimate the first
commercial driverless cars to be released by 2020 [3][2]. The promise of saving 1.2 mil-
lion lives a year [1] and solving traffic congestion problems [2] has struck a chord with
scientists, engineers, and programmers around the world. Thanks to quantum leaps made
in computing technologies in the past 30 years—cheap sensing, reliable object recognition,
and real-time, portable, large-scale data analysis—automated vehicles are becoming a real-
ity. Inspired by ongoing research today from around the world, this MQP aims to imagine a
divergent take on autonomous vehicle technology by challenging modern vision algorithms
combined with affordable sensing technology.
The path the industry will take to evolve autonomous vehicles is generally agreed upon
by large automobile manufacturers and market analysis groups alike [2][4]. As shown in
Figure 1.1, this consists of beginning with features like adaptive cruise control, automated
emergency braking, and park assist technologies, which exist today, and passing through
intermediary features such as highway and traffic jam assistive features before arriving at
fully autonomous vehicles.
Several other labs have demonstrated the viability of autonomous cars in general, such
9
Figure 1.1: IHS’s prediction of the evolution of self-driving cars.
as those of Google and the Autonomous Systems Laboratory at the University of California,
Santa Cruz, as well as several major car manufacturers including Toyota, Nissan, Cadil-
lac, and Audi [2], but many of these systems rely on expensive, bulky, and ungainly roof-
mounted LIDAR detectors [5]. A unique part of the mission of this MQP is to achieve
the same or better results using vision only—clearly it is possible because humans do this
already.
Autonomous vehicle research has exploded in past decades, due to increased fascination
with driverless vehicles and the impact they can have on society. Cars today come with
options for adaptive cruise control, lane detection, and automated parallel parking—all
features that rely on sensors and computing. Further advanced autonomous vehicles blend
human-control with autonomous systems, and as result, have the ability to activate brakes
in emergencies or alert drivers of dangers [5]. Tracing the origins and historical discoveries
of autonomous vehicle technologies leads us to the basis for the Robocart MQP.
10
1.1 Major Milestones in Autonomous Vehicle History
Early on, fully-autonomous vehicles, (ones that did not rely on devices embedded into
roads), were few and far between [5] before Martin Marietta, in conjunction with some
research facilities and funded by DARPA, introduced the Autonomous Land Vehicle Project
in 1985 [5][6]. Martin Marietta’s ALV used computer vision and laser scanning for sensing
and six server racks for path correcting. It successfully traveled a half mile on an empty
road in 1985, but was notoriously fickle and easily tricked by shadows and small variations
in lighting [7]. In the same time period, Ernst Dickmanns in Munich introduced saccadic
vision and Kalman probabilistic filters for use in autonomous vehicles [5].
A decade later, in 1995, Carnegie Mellon developed the Rapidly Adapting Lateral Po-
sition Handler (RALPH) which used computer vision to determine the location of the
road ahead to autonomously steer a car as two researchers controlled the throttle and
brakes [5][8]. Dean Pomerleau was able to "teach" an artificial neural network to drive
the car (it learned to use the grass as boundaries) and was able to successfully drive on a
highway at 55mph [9]. Researchers from Carnegie Mellon were able to use this software
to drive an autonomous car from Pittsburgh, PA to San Diego, CA for over 98% of the jour-
ney [5][8], a project called computer vision and No Hands Across America. By this point,
autonomous path planning was somewhat solved to a degree, but there were still many
issues to be resolved before a car could actually drive itself.
DARPA’s Grand Challenge in 2005 challenged universities to make driverless cars to
traverse a 132 mile-long off-road driving course in the Mojave Desert. The competition
was actually the second of its kind—the first DARPA Grand Challenge in 2004 had been
quite a disaster [9]. The competitors again took a wide number of approaches, utilizing
combinations of GPS, radar, LIDAR, computer vision, sonar, and machine learning to navi-
gate a trafficless desert course at speeds up to 25 mph [5]. The winning autonomous car,
Stanley of Stanford University, used machine learning to distinguish errant sensor readings
from the bumping around of the car and differing light conditions, and accounted for them
11
using probability distributions. Essentially, it was able to reason about the accuracy of its
readings and make fewer errors—only about 1 in 50,000 [9].
Four years later, in 2010, the Vislab from the University of Parma in Italy constructed
a fully-electric autonomous vehicle that embarked upon and completed the VisLab Inter-
continental Autonomous Challenge: an 8,000 mile road trip from Parma to Shanghai [1].
Throughout the journey, the vehicle encountered a variety of traffic, road, and weather con-
ditions [5]. Unlike cars from the DARPA Grand challenge, the Vislab vehicle largely relied
on image processing for local mapping. Other sensors onboard included laser-scanning and
GPS, but the lasers were mainly used for detecting terrain [1]. Vislab proved the reliability
and viability of vision algorithms rather than the use of complex sensors.
Beginning in 2011, Google started a self-driving car project that leverages their map-
ping technology in order to navigate roads. This prompted the Nevada Department of
Motor Vehicles to issue the first driver’s license for an autonomous vehicle. Along the way,
Google has discovered more challenges involved in autonomous driving, including having
to program aggressive behavior for moving through a four-way intersection.
In 2014, Volkswagen implemented the AdaptIVe Project with the objective of creating
autonomous vehicles that can function in various levels of traffic and driving scenarios.
Specific goals include navigating a traffic jam, parking in a parking garage, and eventually
creating a robotic taxi.
1.2 Sensors in Autonomous Vehicles
Even though sensors like LIDAR are becoming more affordable and their use more widespread,
they still cost tens of thousands of dollars in today’s market. The Robocart MQP aims to di-
rect its research toward LIDAR-less navigation, citing Vislab’s Intercontinental Autonomous
Challenge and other progressions in sensing analysis algorithms as evidence. Thus, it was
decided to use standard Raspberry Pi cameras because they were cheaply available and
12
could relay information over a network easily through a Raspberry Pi module.
1.3 Report Structure
The following chapter, Chapter 2, lists some of the topics that are relevant to this project
which may not have been covered before in classes at WPI. Chapter 3 covers the thought
processes behind the designs of the Raspberry Pi mount, electronic steering, and electronic
braking systems. Chapter 4 discusses the actual implementation of these ideations: the
steps followed and the challenges encountered. Chapter 5 includes a definitive analysis of
whether or not the requirements of each system were met, as well as bond graphs for the
steering and braking mechanisms. Finally, Chapter 6 discusses some more of the work that
needs to be done, as well as recommendations and brief notes that could help teams in the
future.
13
Chapter 2
Project Fundamentals
This chapter explains the background of this project and its short history, as well as some
foundational knowledge necessary to understand much of the electromechanical aspects
of this report.
2.1 About this MQP
The MQP described in this report is the second in a series of a much larger project con-
ceived by Professor Wyglinski to wirelessly network together multiple autonomous vehi-
cles, including aerial drones as well as ground vehicle. Last year’s team was the first to
begin work on this project; they had a large team of five engineers working on both the
aerial drones as well as the ground vehicle. Although they made significant progress, they
found the project was too large for one MQP. As a direct result, this year the project was
split into one ground vehicle MQP and two aerial drone MQPs.
This year, the ground vehicle MQP originally consisted of three members—Liz Miller
(RBE), Gabe Isko (RBE), and myself (RBE/ME). Because I was the only dual-major on the
team, I wanted to try to overload in C-term and finish at the same time as Liz and Gabe.
Partway through the year, however, I realized I would not be able to finish in time, so it was
decided that the MQP would again be split into Liz and Gabe’s part and my part so that we
14
Figure 2.1: The three main mechanical components with respect to the overall system.
could submit in different terms. Thus, Liz and Gabe focused on the ROS programming and
networking side, while I focused on designing and installing the mechanisms and sensors
necessary for automating the golf cart. The different components of the overall system
focused on in this report are outlined in orange in Figure 2.1.
As a result, this report focuses mainly on the electromechanical aspects of the ground
vehicle project. For the networking aspects of the ground vehicle, please see Liz’s project
report on the WPI Electronic Projects Collection page here. The networking and ROS as-
pects of this project were pursued separately, and at some other time can be joined together
with the mechanical aspects.
2.2 Sensors
Autonomous cars are only as good as the sensors and algorithms behind them. Affordable
sensing technology has been a significant challenge for the robotics community.
Common sensing techniques include rudimentary sensors, best characterized by SONAR
emitter/detectors. SONAR rangefinders work by emitting high-frequency sound waves and
Prateek, Thank you for your phone call today. Below is a link to the datasheet along with pricing. EPC part# MA63S-??h�p://www.encoder.com/literature/datasheet-ma63s.pdf MSRP: $780.00 ea Student Discount of 30%: $532 ea WARRANTY: 3-years from date of EPC shipment. Let me know if you have any ques�ons about the configura�on or anything else. Regards, Sarah WalterTechnical Sales ManagerEncoder Products Company | www.encoder.com464276 Highway 95 South | Sagle, Idaho 83860T: 800.366.5412 Ext. 4785 | F: 208.263.0541 | E: [email protected]
DISCLAIMER: Encoder Products Company (EPC) has made our best effort in providing this cross reference. Due to themany varia�ons of encoder specifica�ons between manufactures, it is ul�mately the responsibility of the customerand/or EPC Distributor to verify that our suggested cross reference will work in the intended applica�on. Each encoderis custom built and EPC will not be responsible if our suggested cross fails to perform due to the configura�on of thecustomer’s applica�on. Once built and shipped, EPC products are not returnable.
F eat u r esstandard size 25 Package (2.5" x 2.5")Durable Magnetic Technology—No Gears or BatteriesServo and Flange MountingMultiturn Absolute Encoder (14 Bit/40 Bit)ssI and CANopen CommunicationsIP67 sealing availableThe Model MA63S Multiturn Absolute Accu-Coder™ is ideal for a wide variety of industrial applications that require an encoder with the capability of absolute positioning output, even in power-off scenarios. Its fully digital output and innovative use of battery-free multiturn technology make the Model MA63S exceptionally reliable. The MA63's robust and durable magnetic technology and available IP67 seal readily handle the harshest industrial environments, including those with elevated electrical noise. Available with several shaft sizes and mounting styles, the Model MA63S is easily designed into OEM and aftermarket applications.
Co M M o N A p p l i C AT i o N SRobotics, Telescopes, Antennas, Medical Scanners, Wind Turbines, Elevators, lifts, Motors, Automatic Guided Vehicles, Rotary and X/Y positioning Tables
M o D E l M A63S o R D E R i N G G u i D E
NOTES:1 Available with SSI only.2 For mating connectors, cables, and cordsets see Encoder Acces-
sories on page 102 or visit www.encoder.com. For Pin Configuration Diagrams, see page 107 or visit www.encoder.com.
3 Available with CANopen only.4 For single-turn resolution, enter '00' (SSI only).
Blue type indicates price adder options. Not all configuration combinations may be available. Contact Customer Service for details.
Mechanical
12
SINGLE TURN RESOLUTION
12 Bit (CANopen)08 to 14 Bit (SSI)
SHAFT SIZE38 3/8", 0.375"10 10 mm
INpUTVOLTAGE
5 5 VDC1
V3 10 to 30 VDC
MOUNTINGMA 2.50" FlangeMC 2.50" Servo Mount
MULTITURNRESOLUTION
18 18 Bit (CANopen)40 40 Bit (CANopen)00 to 40 Bit (SSI Only)4Price adder for Bit > 18
COMM pROTOCOLp1 CANopenp2 SSI
MODELMA63S Absolute Series
Electrical
OUTpUT CODE C1 BinaryC2 Gray1
CONNECTOR TYpE2
SMJ 5-pin M12 Side Mount3SMK 8-pin M12 Side Mount1
TEMp 0-80° C StdT5 -40-80° C
SOFTWARE REVA Revison A
SEAL IP50 StdS3 IP66S4 IP67
MA63S MA38 V3 p210 C1 SMJA
M o d e l M A 6 3 S - M u l t i t u r n A b S o l u t e e n c o d e r
Bosch Motors are used in the 1999 Toyota Sienna and the 1999 Ford Windstar. If you wish to purchase an additional Bosch motor, you must buy the entire “Power Sliding Door unit”. The Bosch motor is the right hand side motor. Great care must be taken when removing the motor from the front door unit. The retaining clips must be removed from the output shaft or damage will occur to the shaft
FISHER-PRICE MOTOR INFORMATION The following are approximate performance data for the Fisher-Price motor/gearbox sets supplied in the kits.
GLOBE MOTOR
GLOBE MOTOR AND DRIVE ASSEMBLY SPECS
Warning: The Globe Motor cannot support side loads.
No Load Speed: 75 RPM Stall Torque Clockwise:
34 Nm
Stall Torque Counter-Clockwise:
30 Nm
Stall Current: 44 Amps All specs at 12 Vdc.
Motor no-load speed 15,000 RPM Motor stall current 57 A Motor stall torque 0.380 N-m (mili-NEWTON meters)
Gearbox ratio 124:1 No-load speed w/gearbox 100 RPM (estimated)
Stall torque w/gearbox 34.7 N-m (estimated)
Motor with Drive Assembly Motor Only No Load Speed: 87 RPM ± 1 97 RPM Stall Torque: 150 In-lb 30 oz-in Stall Current: 18.5 Amps 18.5 Amps No Load Current 0.820 Amps 0.820 Amps All specs at 10 Vdc.