ivc design brief submission

31
TARRo2 Design Brief April 2016

Upload: jonivan-artates

Post on 23-Jan-2018

61 views

Category:

Documents


1 download

TRANSCRIPT

TARRo2 Design Brief

April 2016

TARRo2 DESIGN BRIEF 1

TARRo2 Design Brief Submitted to: UCI Rescue Robotics Competition

Jonivan Artates

Xochitl Alvarado

Jose Antelo

Joe Wijoyo

Amal Edick

Brian Mauricio

Haowen Wong

Wyeth Binder

Naziha Kibria

Sina Habibizad

ASEC RAG

Professor Jack Appleman

Irvine Valley College

April 2016

TARRo2 DESIGN BRIEF 2

TABLE OF CONTENTS Pg #

EXECUTIVE SUMMARY ……………………………………….................................. 3

I . CONFIGURATION DESIGN ……………………………………………………….. 4

1: Assessment…………………………………………………………………….... 5

2: First Responder Interface……………………………………………………….. 7

3: Navigation………………………………………………………………………. 8

4: Mobility…………………………………………………………………………. 9

5: Power & Safety…………………………………………………………………. 10

6: Structure……………………………………………………………………….... 11

II. TEAM & FACULTY INFORMATION ……………………………………………. 12

III. PROJECT ORGANIZATION & OUTREACH ……………….…………………. 13

1: Timeline………………………………………………........................................ 13

2: Marketing………………………………………………………………....….. 14

IV. NEXT STEPS & FUTURE OUTLOOK ………………………………………. 16

1: Through Phase 2 Completion………………………………………..………….. 16

2: Year 3 & Beyond………………………………………………………....……... 16

V. PERFORMANCE VERIFICATION & CHALLENGES …………………………. 17

1: Specifications……………………………………………………....…………..... 17

2: Assessment Challenges………………………………….………………………. 17

3: Navigational Challenges ……………………………………………….………...….... 17

4: Mobility Challenges…………………………………………………….……….. 18

VI. CONCLUSION ……………………………………………………………………... 19

VII. APPENDICES ……………………………………………………………………... 20

TARRo2 DESIGN BRIEF 3

EXECUTIVE SUMMARY This Design Brief describes Phase II of the Triage Assistance Rescue Robot (TARRo2) project during the 2015­16 academic year. The purpose of this report is to provide the UCI Rescue Robotics panel with sufficient information to assess the direction, implementation and effectiveness of the TARRo2 system design developed by the Irvine Valley College (IVC) Robotics Activity Group (RAG). The primary functional objective is to develop a device that will autonomously locate and assess as many victims as possible on a field with dimensions 10,000 feet 2 within a 15 minute period. During the UCI competition, orange­colored buckets simulate stationary human victims, and QR codes on each bucket contain information regarding the victim’s status. During 2015­16 year, RAG has made considerable progress in advancing the functional capabilities of TARRo2. The following six improvements have been implemented to enhance the victim locational and assessment abilities of TARRo2:

1. A laser ranging sensor (lidar) and inertial measurement unit (IMU) provide distance­to­victim data accurate to within 1 inch, enabling more accurate calculation of relative victim position.

2. An LED array provides 300 lumens of light to assist cameras in detecting victims and in reading QR codes.

3. To allow for a 180° field of view without interfering with the device’s direction of travel, sensors are mounted atop a rotating turret.

4. TARRo2 data will be transmitted via WiFi to the First Responder Interface (FRI). 5. The First Responder Interface (FRI) will have a graphical user interface (GUI) that

livestreams data and video. 6. TARRo2 data is stored both onboard the device and in a database on FRI, benefiting first

responders and future implementation efforts of a multi­robot system. To improve the maneuverability and search efficiency of TARRo2, four improvements have been made:

1. A Navigation System complemented with GPS and ultrasonic obstacle detection improves the capability to make real­time changes to the TARRo2 search strategy. The first responder operator only needs to outline a desired search field, and TARRo2 will then autonomously navigate through a set of generated waypoints within the field.

2. The device is supported by four wheels – two drive wheels and two casters. This design improves control during turning and reduces axial loading.

3. Increased wheel size (6” vs. 4” in TARRo1) and more powerful motors (170lb. in. vs. 37.5lb. in.) assures the platform is able to move over uneven surfaces more reliably than TARRo1.

4. A novel tire and wheel design prototyped using 3D printing allows the device to maintain traction on a wider variety of surfaces.

To accommodate the increased size of RAG and the fact that several development teams needed t o simultaneously implement and test various systems, TARRo’s Power and Structure Systems changed in two main ways:

1. Additional sensor systems, communication devices, and larger motors required that the power storage and distribution system be redesigned to include enhanced current flow protection, better wiring harnesses, and improved safety measures.

2. The structure consists of three platforms that can be quickly assembled and disassembled, allowing the team to independently test systems during integration.

TARRo2 DESIGN BRIEF 4

I. CONFIGURATION DESIGN The 2016 UCI Rescue Robotics Competition challenges teams to develop autonomous navigation devices; each device must be capable of locating multiple victims in a field and assessing each victim’s survival status. To simulate a real world scenario, orange buckets are used as victims, and QR codes on each bucket provide status data. Performance is judged by the amount of information returned at the end of a 15 minute survey on a grass and dirt field with an area of 10,000 feet 2 . TARRo2 autonomously provides critical and time sensitive victim information to first responders during search and rescue operations. To accomplish this, TARRo2 has a structured hierarchy of systems, each of which maximizes the effectiveness of the overall device. The hierarchy of systems is: I. Assessment ­ Sensors which locate victims and interpret status data. II. First Responder Interface (FRI) ­ Communication between TARRo2 and first responders. III. Navigation ­ Algorithms and computers that generate and communicate search paths to

Mobility System. IV. Mobility ­ Mechanical and electrical drive components and suspension assembly. V. Power & Safety ­ Batteries and electronics to store and safely distribute power to each

system. VI. Structure ­ Supporting platforms designed for parallel system development and

packaging.

Figure 1.0: System Hierarchy

See Appendix B for a comprehensive system hierarchy structure.

TARRo2 DESIGN BRIEF 5

1: Assessment System The Assessment System autonomously detects and assesses victim status; it consists of four sensors and supporting mechanical and computational components. Compared to TARRo1, the TARRo2 Assessment System’s additional components allow for greater accuracy, increased data acquisition speed, and assessment scanning motions to be independent of the device’s travel direction. The principle functional objectives for the Assessment System are:

Detect the ‘distinguishing characteristics’ (visual spectrum) of a victim from a distance of at least 20 feet in a dark environment.

Detect the distance and direction between TARRo2 and victim to within 6 inches and 3°. Detect ‘health status’ information of a victim in a dark environment from a distance of 5

feet. Make status detections with 90% reliability within 5 seconds of coming within range of

the victim. With regards to the UCI competition: the victim’s distinguishing characteristic is the bucket’s orange color, and health status refers to data embedded in a 4” QR code. To meet these functional objectives, TARRo2 incorporates the following design components:

A color identification camera using computer vision identifies blocks of orange as victims, then returns block size and relative horizontal position to a microcontroller.

A laser ranging sensor (lidar) measures distances between TARRo2 and victims. Lidar data is filtered through the microcontroller to eliminate statistical outliers.

Two QR reading cameras mounted 7.75” apart, provide a broad perspective of the victim, and increase the likelihood of quickly obtaining information as TARRo2 approaches curved surfaces of the victims.

LED headlights assist cameras with 300 lumens in dim and shadowed environments. Mounting the sensors on a servo driven turret enables 180° field of view that allows

scanning independent of TARRo2 direction of travel. Combining GPS coordinates and headings generated from the Navigation System with lidar range data allows TARRo2 to calculate lat/long coordinates for the victim location through the following equations:

V lat. = T lat. + R/k * cos( + ) V long. = T long. + R/k * sin( + )

V, T are the coordinates of the Victim and TARRo2 R is the distance between TARRo2 and Victim the IMU heading of TARRo2 the relative turret angle k a scaling factor from inch to decimal degrees.

TARRo2 DESIGN BRIEF 6

At this time, the data from the prototype indicate that the following performance specifications will be achieved:

Victim characteristics range (orange blob detection) 25ft. max

Distance to victim accuracy < 2.5 in. typical

Status detection range with headlights only. (QR code) 5ft. max

Status capture and decode rate 30hz. typical

Future performance specifications will be improved through use of higher resolution camera sensors and better optics. A scanning lidar unit will also provide TARRo2 with 360° field of view and simultaneous “scan and assess” functionality. Forward looking infrared (FLIR) sensors will provide real world victim identification and key biometric information.

Figure 2.0: Assessment Sensors

TARRo2 DESIGN BRIEF 7

2: First Responder Interface The First Responder Interface (FRI) System enables communication and remote control between first responder operators and TARRo2. Compared to TARRo1, the TARRo2 FRI components have enhanced control, increased communication range, and expanded the situational awareness of operators. The principle functional objectives for FRI are:

Maintain a wireless connection up to 250 feet in open field, provided line of sight. Permit operators to generate and transmit autonomous search areas, remote control motor

signals, or stop operation commands. Stream live data and video to graphical user interface with less than one second latency. Record and store victim information onboard, pushing to FRI when WiFi is present.

To satisfy these functional objectives, TARRo2 utilizes the following design components:

2.4 GHz WiFi connection with 10 dBm antennas on robot and operator interface. 49 MHz analog remote control kill switch. Telemetry data and first person video data stream over WiFi. Onboard storage device with supporting sorting algorithm and database with shared

directory on operator interface. At this time, data from the prototype indicates that the following performance specifications will be achieved:

2.4 GHz Remote motor control range > 400 feet

2.4 GHz WiFi data range > 400 feet

49 MHz kill switch range 30 to 50 feet

Telemetry/Video latency 0.3 seconds typical

Directory sharing method SSH, auto­reconnect

Future performance specification will be improved through use of 5GHz WiFi frequency for increased bandwidth and reduced signal interference. Further development of the GUI will consolidate controls and information display.

TARRo2 DESIGN BRIEF 8

3: Navigation The Navigation System generates movement paths, obstacle avoidance strategies, and produces mobility control signals. Compared to TARRo1, the TARRo2 Navigation System’s components have led to smarter path planning, improved obstacle detection, and faster search capability. The principle functional objectives of the Navigation System are:

Generate a consecutive set of waypoint goals which fill at least 75% of a desired search area.

Detect obstacles > 3 inches and execute corrective avoidance maneuvers. Approach victims to within 5 feet to obtain status information, but avoid coming closer

than is necessary. Slow and stop motion (gracefully degrade) during loss of control signal.

To satisfy these functional objectives, TARRo2 utilizes the following design components:

Mission planning software, GPS, and inertial measurement unit (IMU) based navigation as implemented in an ArduPilot 2.6.

Dual ultrasonic sensors determine obstacle presence and position. Proportion/Integral/Derivative (PID) feedback loop controller to control the approach to

victims. Automatic signal loss detection and control override.

At this time prototype performance measurement and testing is ongoing. Search efficiency will improve through implementing simultaneous localization and mapping (SLAM) functions with flood fill algorithms.

TARRo2 DESIGN BRIEF 9

4: Mobility The Mobility System, which moves TARRo2 about the field, consists of the mechanical and

electrical drive components. Compared to TARRo1, the TARRo2 Mobility System’s additional

components have led to improved reliability and increased terrain capability.

The principle functional objectives of the Mobility Subsystem are:

Allow locomotion in tall grass, loose dirt and inclinations up to 20%.

Traverse obstacles, made of various materials, up to 3 inches in height.

Maintain a speed of 3 miles per hour on a level surface with a total weight of 50 pounds.

Execute turns with a radius less than the width of the robot while bearing weight.

To meet these functional objectives, TARRo2 incorporates the following design components:

Wide wheelbase (14” by 22”) improves weight distribution.

Large and open tread tire design maintains traction and clears loose debris.

Rocker suspension system for caster wheels assures 4 wheel support and enhances

traction and large obstacle traversal.

High torque motors (170 lb.in.) with large wheel diameters (6”) increase weight capacity.

Two drive wheels and two casters reduce axial loading while turning.

At this time, data from the prototype indicates that the following performance specifications will

be achieved:

Wheel Base 15.4 in. Payload Capacity 25 lbs.

Track 11.9 in. Top Speed 4 mph

Turning rad. 0.0 in. Obstacle Height 4 in. max

Weight 23 lbs. Max incline 38°

Max Torque 170 in. lbs. Max decline 37°

Max Power 0.107 hp Max roll angle 24°

Future performance specifications will improve as suspension systems are implemented on both

axles. Independent driving and turning of each wheel will increase terrain capability. Direct drive

brushless motors will improve efficiency and reduce weight and noise.

Figure 3.0: Suspension Articulation

TARRo2 DESIGN BRIEF 10

5: Power & Safety The Power and Safety System stores and safely distributes electrical power to other TARRo2

systems and safeguards. Compared to TARRo1, the TARRo2 Power and Safety System’s

additional components have led to improved safety, longer run time, and increased electrical

reliability.

The principle functional objectives of the Power & Safety System are:

Distribute regulated voltage at 7.2 and 5.0 volts to components of all systems.

Safe storage of energy for at least 60 minutes of nominal operation, and capability to

quickly restore after depletion.

Protect components and environment in cases of short circuit, overvolt, undervolt, stall

condition, and impending danger.

Immobilize TARRo2 remotely from at least 50 feet.

For the UCI competition: robot must be powered entirely by electrical power.

To meet these functional objectives, TARRo2 incorporates the following design components:

Regulation and smoothing circuits provide consistent voltage despite varying current

demands.

Quick disconnect wiring harness distributes power to individual components.

Large capacity NiMH battery packs reduce battery swap downtime.

Fuse block, voltage detection, bump stop, and remote kill protect TARRo2 from electrical

danger.

At this time, the data from the prototype indicate that the following performance specifications

will be achieved:

Energy 72 watt hours

Remote Kill Distance 53 ­ 28 ft. (environment dependent)

Energy Restoration Method Battery swap

Battery Specification 6 cell, 7.2v pack. 2 parallel packs

Run Time 60 minute (typical, Appendix C)

Future performance specifications will improve by implementing LiPo cells and onboard

charging circuits. Cell balancing and thermal regulation will prolong battery life and extend

operational runtime. Conformal coating of electrical circuits will provide additional

environmental protection.

TARRo2 DESIGN BRIEF 11

6: Structure The Structure System physically supports TARRo2’s onboard components. Compared to

TARRo1, the TARRo2’s additional assessment sensors have led to the introduction of additional

structural components. Overall TARRo2 now has increased component capacity, ease of

serviceability, and faster system development cycles.

The principle functional objectives are:

Provide location and mounting provisions for electrical components, wire harnesses, and

mobility hardware for each of their necessary orientations.

Facilitate simultaneous development and testing of TARRo2 systems.

Support weight of structural components and additional payload while maintaining the

center of gravity within 2” behind drive wheel axis.

For the UCI competition: weight limit is 25 pounds, robot size limit is 24 inches wide by 24

inches long, and operates on a relatively flat grassy terrain.

To meet these functional objectives, TARRo2 incorporates the following design components

(Appendix F):

Laser cut platforms enable component organization, wiring routes and mounting

provisions.

A modular design with three standardized platforms permits rapid disassembly and

independent development of component systems.

Weight biased component placement and CAD simulated platform loading.

At this time, the data from the prototype indicate that the following performance specifications

will be achieved:

Length 22.4 in.

Width 14.1 in.

Height 13.4 in.

Future performance specification will improve through better manufacturing methods and

materials. A casing will add aesthetic appeal while protecting the internal hardware of TARRo2

from environmental factors.

TARRo2 DESIGN BRIEF 12

II. TEAM & FACULTY INFORMATION

The Irvine Valley College (IVC) Robotics Activities Group (RAG) is a subgroup of the Applied Science and Engineering Club (ASEC), a noncredit extracurricular program at IVC. Lead by Jonivan Artates (mechanical engineering, year 2), the 2016 RAG team is made up of:

Jose Antelo, mechanical engineering, year 2 Xochitl Alvarado, biomedical engineering, year 2 Joe Wijoyo, computer science, year 2 Brian Mauricio, electrical engineering, year 1 Amal Eldick, chemical engineering, year 1 Haowen Wong, electrical engineering, year 1 Naziha Kibria, mathematics, year 2 Wyeth Binder, civil engineering, year 2 Sina Habibizad, biomedical engineering, year 1

The RAG members were divided into six different system teams, each in charge of different aspects of the TARRo2 development process. The Structure and Integration team consists of Jonivan and Naziha. Jonivan and Joe make up the Control and Communication team, Wyeth and Joe the Assessment team, and the Navigation team consists of Jose and Amal. Xochitl and Sina are in charge of Mobility, and Power is handled by Brian and Haowen. ASEC RAG is led by faculty mentor Professor Jack Appleman. Other IVC faculty involved in the development of TARRo2 are Professors Brett McKim, Iknur Erbas­White, Matt Wolken, Brian Monacelli, Alec Sim, and Zahra Noroozi. The team receives funding and support from Dean Corrine Doughty, Dean Lianna Zhao, and Merry Kim.

TARRo2 DESIGN BRIEF 13

III. PROJECT ORGANIZATION & OUTREACH The 2015­2016 academic year marks the second year of development on the TARRo project,

phase two of a three year plan. The goals of this endeavor were to provide hands­on experience

and application of classroom knowledge to a real world project. At the end of each academic

year, the team will compete with their robot in the ground rover division of the Rescue Robotics

Competition hosted by the University of California, Irvine (UCI).

The 2015­16 academic year’s team began with three members, all of whom returned for their

second year on the project. Presentations were made in STEM classes and clubs to promote the

project to other students. Additionally, RAG demonstrations were hosted by IVC administration

to promote the endeavor to faculty of the college and district. Prospective members submitted

applications, and selection was based on academic potential, previous experience, and an

expressed passion for robotics and its applications; five additional members were brought onto

the project during the fall semester.

The first four months served as a training period which allowed members to expand their

knowledge base and skillsets. During weekly RAG meetings, members developed a foundation

in the study of robotics, studying aspects of mechanical and electrical engineering, computer

science, as well as applicable concepts in fabrication and economics. Additionally, guest

lecturers were hosted throughout the year, including faculty experts in various fields: Photonics

from Professor Brian Monecelli, Differential Global Positioning System (DGPS) from Professor

Matthew Wolken, and Rapid Prototyping from Professor Brett McKim. Through the winter,

weekly time commitments were ramped up to three meetings a week, and membership increased

to eleven.

Over winter break, the project began the iterative “Design, Build, Test” process of the TARRo2

project. Group meetings were held on a weekly basis in order to monitor progress and to manage

resource allocation. In addition, means for online communication were implemented for an open

forum style of documentation archiving. The transparency created between system development

teams allowed for reduced iteration cycles, assisting the integration processes. Additionally, the

RAG team incorporated community outreach as an integral part of development. Members

benefitted through various opportunities of interaction with industry professionals and academic

mentors, and served as IVC ambassadors to intramural and extramural robotics and STEM

events. This year, RAG has participated in or attended events such as the DARPA DRC, IEEE

chapter STEM expo, Maker­Faire mini, and the MD&M West manufacturing expo. The team has

also presented to multiple local high schools and colleges throughout the 2015­16 year.

1. Timeline During the initial training period, Management team members organized the scope, time, quality,

and budget resources in order to construct a success criteria for the TARRo2 project by the end

of the academic year. Based on this projection, tasks and deadlines were laid out in

reverse­chronological order. This was mapped using a Gantt chart, allowing a critical path, team

resources, and subsequently, task priorities to be mapped out ( Figure 4.0 ). Milestones were

placed at the projected completion date of each system, and marked the beginning of integration

with other systems. The time frames for each task were used as a reference from which managers

TARRo2 DESIGN BRIEF 14

could track development progress. Reallocation of resources permitted flexibility in meeting

milestone dates. During the development phase, emphasis was placed on documentation of

individual tasks. These files were available to the entire team, and were used to determine how

parallel developments would integrate. Clearly defined inputs, outputs, functionality, and

constraints were created for each component, system, and the project as a whole. By design,

these boundaries were established to mesh cleanly with their respective neighbors from the

physical, electrical, and programming aspects. Once development was complete, system

integration connected the various chains of inputs and outputs. The expanding systems were

frequently tested during the integration period to facilitate debugging subtle inconsistencies and

runtime errors.

Figure 4.0: Tasks and subsequent integration phases.

2. Marketing The long term vision of this project is to develop a commercially viable device capable of

assisting first responders in an emergency structural collapse disaster. In developing a plan to

market, the current state of robotics technology, industry and markets are of prime consideration.

The complex marketing strategy of TARRo addresses several factors:

1. A small niche customer base that would likely purchase response devices.

2. Large upfront costs to ownership and ongoing routine service costs.

3. Unpredictable product demand that relies on environmental disaster situations.

4. Wide range of environments and tasks that TARRo must be able to manage.

The team developed a number of commercialization strategies, seeking to maximize product

value.

Inspired by the successful commercialization of the iRobot Roomba “vacuum robot,” the RAG

team anticipates that a TARRo V1 could be implemented to provide after­hours surveillance of

large structures to supplement stationary video and fire detection systems (Appendix A). This

enables the TARRo customer base to become accustomed to TARRo V1 before an emergency

situation occurs. Utilizing the onboard assessment sensors and WiFi connectivity, TARRo V1

data regarding the building structure, total individuals present, and robot status would be

TARRo2 DESIGN BRIEF 15

collected; all information is sent to a remote interface and transmitted into a cloud storage and analysis system accessible by public first responder agencies. This feature would scale with additional robots, allowing multiple floors or areas to be covered simultaneously. The presence of TARRo security will increase the safety of TARRo­surveilled areas without unnecessarily risking the lives of security guards. In the event of a natural disaster, surviving TARRo devices could relay live streams of time­critical data, crucial for saving lives and effectively managing first responder resources. The reason for developing TARRo V1 for the the surveillance and security market is its size. This market is more mature than the disaster response robotics segment, and valued to reach $42 billion versus $1 billion respectively within the next 8 years, respectively (Appendix A). To offset the upfront costs of ownership tied to robotic systems, TARRo would be provided as a data subscription service. Data from TARRo would be wirelessly uploaded to a server over, which provides data processing specific to customer needs. Services would include crowd analytics, traffic management, and structural information in addition to first responder assistance. Additionally, TARRo’s software systems would be managed through this connection, allowing for remote troubleshooting and firmware update capabilities. To further reduce customer operating costs, the team has focused on a modular design for all systems. This allows the customer to have a system that is tailored to their needs without excessive complexity or expense that is typical of off­the­shelf robot solutions. In emergency situations, TARRo seeks to provide a comprehensive assistance system. Applications include attaching the Assessment System to an off the shelf drone for aerial search and rescue, or using the Mobility System to tow aid through rough terrain. The expectations of an effective general disaster response robot are difficult to define, and are usually redefined “in the moment.” In contrast, robot systems need functionality that is defined specifically, which is then followed by a long development time frame. TARRo minimizes this disparity through a development cycle focused on the most current standards and an expandable system design. The National Institute of Standards and Technology (NIST) has published Standard Test Methods for Response Robots (Jacoff, Appendix A) to which TARRo V1 will use as a performance baseline. Using these quantified performance merits, a customer would be able to make informed decisions in choosing a robot. Beyond these standards, additional systems will be designed to plug and play, quickly and easily adding specific functionality. This way, customers would be able to start with a version of TARRo that is sufficient to their needs, and add­on as necessary, eliminating the costly cycle of acquiring and testing multiple platforms.

TARRo2 DESIGN BRIEF 16

IV. NEXT STEPS & FUTURE OUTLOOK

1. Through Phase 2 Completion

The development of TARRo2 has required the introduction of several technologies, including the

navigational unit, lidar, and tire design. The use of a navigational unit has greatly improved path

planning, and its integration with computer vision and lidar technologies has improved the

Assessment and Mobility Systems. The shift to urethane­casted wheels and a rocker suspension

for the rear caster wheels increases TARRo2’s range of mobility, allowing it to more effectively

assess victims in a disaster. Additionally, this year’s team was restructured by treating the various

systems in a modular fashion. Streamlining development, testing, and system integration

benefited the team dynamic and increased member productivity.

2. Year 3 & Beyond

Looking ahead to TARRo3 development, the team hopes to implement several new technologies

to further improve functionality. In order to better locate and identify victims in a structural

collapse disaster, a thermal imaging sensor will be integrated, expediting the recognition process.

Known as Forward Looking Infrared (FLIR), this sensor will allow the Assessment System to

more efficiently locate human targets, which are identified by unique heat signatures. This, in

turn, will alleviate pressure on the subsequent systems, making for a more efficient product.

Another sensor that would benefit TARRo is a scanning lidar unit. As the demand for these

sensors increase, new manufacturing methods and economies­of­scale has greatly reduced prices.

Implementing this sensor would enable Simultaneous Localization and Mapping (SLAM). This

functionality would provide significant gains to the value and capability of the TARRo system.

To take advantage of SLAM technology, the team plans on introducing flood fill algorithms to its

Assessment and Navigation Systems. This would autonomously generate search waypoints

within a prescribed field.

By using such an algorithm, flood fill will allow TARRo to retain the positional coordinates of

visited locations, ensuring that while scanning for targets, the same geographic location is not

visited twice. Once again, this allows for a dramatic increase in efficiency and frees up time and

resources that can lead to TARRo assessing a wider geographic area.

Lastly, the team hopes to develop a means for a multi­robot system in the near future. Deploying

multiple mobile units or a mobile unit communicating with fixed surveillance units would

increase data throughput and create the possibility for exponential growth in every area of

functionality. An expanded FRI would allow multiple robots to communicate with one user

interface; the consolidation of multiple data inputs into a network allows for better

decision­making and machine learning over time.

TARRo2 DESIGN BRIEF 17

V. PERFORMANCE VERIFICATION & CHALLENGES

1. Specifications The current performance specifications (listed below) were determined empirically during the integration and testing phases of the TARRo2 project and will be further updated in coming weeks. Length 22.4 in. Top Speed 4 mph Energy 72 watt hours

Width 14.1 in.

Obstacle Height 3 in. max Protection

Fused, Switch, Low

Voltage

Height 13.4 in. Run Time 60 min. (typ.) Remote Kill dist. 53 ­ 28 ft. (env.

dependent)

Wheel Base 15.4 in. Max incline 38 deg. Remote Mobility Line of Sight

Track 11.9 in. Max decline 37 deg. Charge time ~130 min.

Turning rad. 0.0 in. Max roll angle 24 deg Wireless Freq. 2.4 GHz

Weight 23 lbs. Max Torque 150 in. lbs. Onboard Storage 16gb, expandable Payload 25 lbs. Max Power .107 hp Suspension Rocker, Non­damped

2. Assessment Challenges

The original plan for victim assessment in TARRo2 was to wirelessly stream the video feeds to FRI, where QR detection algorithms would be performed. This method was initially chosen for two reasons:

1. Computationally heavy tasks would be performed in stationary locations, with more computing power.

2. Cost of each additional robot would be reduced in a multi­robot system. Video transmission and QR detection were successfully developed and tested during the build phase. During integration however, it was found that the 2.4 GHz control radio communication signal would interfere with the 2.4 GHz WiFi connection between the robot and FRI. Though it was possible to operate the robot without the need for control radio, members decided that a different solution should be found, as the radios of other teams would similarly interfere during competition. This issue was resolved by moving the QR detection software onto the robot, and only updating FRI while WiFi was available. Operator awareness of victim detection is compromised, but TARRo maintains overall performance as determined during field testing.

3. Navigational Challenges During the 2015 Rescue Robotics Competition, it was noted that TARRo1, and other ground rover devices faced navigational issues in every trial. Guided solely by navigating to nearby victims, TARRo1 routinely became trapped in “local minimums,” recursively visiting the same group of victims. Runs were limited on time, so human intervention was required to break out of the minimums. During the design phase of TARRo2, it was determined that a system designed to autonomously break local minimum issues would be a priority. Research and ideation was

TARRo2 DESIGN BRIEF 18

completed on various positioning systems, including WiFi positioning, Bluetooth beacons, dead reckoning, and one novel approach using concepts from a sextant in celestial navigation. Included in these discussions were possible combinations of the systems to offset the inherent shortcomings of each. This would be accomplished through the use of Kalman filters, which develop error covariances with sensor’s sampling data to increase state estimation accuracy. As the complexity of the navigational challenges, along with those of the other systems increased, the practicality of dedicating resources to this effort within the given time constraints diminished. It was decided that an off the shelf navigational unit would provide the solution for this stage of TARRo development, and consequently the ArduPilot 2.6 was chosen as the main navigational device.

4. Mobility Challenges The product of last year, TARRo1, received a “Did Not Finish” (DNF) at field trials due to a fault in the Mobility System. A Terminal Fault Analysis completed on TARRo1 determined that the friction­based motor housing did not hold up to the cyclical torque applied by the motors in the terrain of the competition (Appendix G). TARRo2 remedies this issue with a new drive configuration, including interference fit motor mounts. The design goal of this year’s Mobility System is to maintain contact of all wheels to the ground whenever possible, ensuring traction of the drive wheels in as many terrain conditions as possible. The final design utilizes a rocker suspension for the caster wheels at the rear of the platform, facilitating passing over obstacles of up to 4 inches while maintaining contact of all four wheels. Drive wheels were reduced from four to two and placed at the front of the robot bearing the majority of the weight. A single axis was chosen to reduce drag during turning and allow encoders to more accurately represent vehicle movement. Testing has determined that moving the center of rotation to the front of the robot best complements the inputs from the Navigation System. In certain situations, namely hills with loose or slippery terrain, it was found that smooth rubber tires failed to maintain traction. This was resolved by using a wider tire design with deep reliefs in the tread. The larger contact area prevented the tire from “digging” into loose terrain, while the reliefs work similar to a paddle, providing traction in difficult situations.

TARRo2 DESIGN BRIEF 19

VI. CONCLUSION In the development of TARRo2, RAG invested a considerable effort to ensure a holistic approach to the challenge of rescue robotics. Going beyond building an autonomous robot to complete a single task, the team focused on the ecosystem in which a robot of this nature would ultimately function. This led to studies beyond the engineering and computer science topics that generally encompass robotics. The goal in this effort was to truly determine the primary functionality of the TARRo system, the scope which would dictate the direction of project. Answering this question required the team to expand their learning into subjects such as economics, sociology, history and ethics. This mindset permeated throughout the development cycle, shaping priorities and serving as a prime directive in the various decision making processes. During the design and construction phases of TARRo2, the various system teams gained experience in communication and interaction with members of other systems, on whom they depended on for cohesive inputs and outputs. These lessons serve to benefit the team in future projects where progress relies on the combined efforts of each individual. During this year, the members of RAG successfully completed the design, construction, integration and testing of TARRo2 . Equally as beneficial, the team gained first hand experience learning from mistakes, applying textbook knowledge to a real world project, and growing through self­driven discovery.

TARRo2 DESIGN BRIEF 20

VII. APPENDICES

TABLE OF CONTENTS Pg #

Appendix A: Works Cited……………………………………………………………. 21

Appendix B: TARRo2 Systems Map…………………………………………………. 22

Appendix C: TARRo2 Power Accounting…………………………………………… 23

Appendix D: Battery Capacity Testing……………………………………………….. 24

Appendix E: Bill of Materials………………………………………………………... 25

Appendix F: Weight Analysis……………………………………………………….... 26

Appendix G: Failure Analysis of TARRo1…………………………………………... 27

Appendix H: Table of Acronyms …………………………………………………….. 30

TARRo2 DESIGN BRIEF 21

APPENDIX A ­ Works Cited iRobot History. (n.d.). Retrieved April 02, 2016, from http://www.irobot.com/About­iRobot/Company­Information/History.aspx Jacoff, A. (2013, October 1). Standard Test Methods For Response Robots. Retrieved February 20, 2016, from http://www.nist.gov/el/isd/ks/upload/DHS_NIST_ASTM_Robot_Test_Methods­2.pdf SSI Staff. (2014, May 15). Report: Video Surveillance Market to Reach $42B by 2019. Retrieved March 15, 2016, from http://www.securitysales.com/article/report_video_surveillance_market_to_reach_42b_by_2019 Service Robot Statistics. (n.d.). Retrieved March 16, 2016, from http://www.ifr.org/service­robots/statistics/

TARRo2 DESIGN BRIEF 22

Appendix B ­ TARRo2 Systems Map

TARRo2 DESIGN BRIEF 23

Appendix C ­ TARRo2 Power Accounting The voltages used for each component were determined using the specifications in the respective datasheets. In cases where a component was compatible with a range of voltage levels, the appropriate selection was made from the three levels available on TARRo2: 7.2v, 5v, or 3.3v. A multimeter was inserted in series with the component circuit to determine current draw during testing. Values provided are the typical consumption. Text in red were not empirically tested by RAG; those current values were provided by manufacturer datasheets.

TARRo2 DESIGN BRIEF 24

Appendix D ­ Battery Capacity Testing A 6­ohm, 50 watt resistor was attached to a battery, along with a multimeter attached in parallel.

Voltage measurements were taken at 5 minute intervals, and current was derived using Ohm’s

Law. A Riemann sum was taken of the current * time, used to estimate the area under the curve,

the capacity of the battery. The particular battery shown is a 2 year old 7.2v Tenergy model, rated

at 4200mah. The data shows the battery’s voltage dropping off after discharging only 1800mah,

indicative of a dying battery. Following this test, new batteries with 5000 mah capacity were

purchased.

TARRo2 DESIGN BRIEF 25

Appendix E ­ Bill of Materials The Bill of Materials (BOM) was used to catalog the various components and current market

prices during the build of TARRo2. The following BOM is of the final components to be used

during competition. Various consumables are not noted as they were considered stock to RAG.

This included fastening hardware, wire, laser cut material, and 3D printing filament.

TARRo2 DESIGN BRIEF 26

Appendix F ­ Weight Analysis The weight analysis table was populated during the construction phase of the TARRo2 project.

Consideration to the benefits of performance were weighed against the weight penalty of each

component, specifically in the selection of drive motors and batteries. Additionally, the table is

divided into each platform level, which assisted in choosing component placements that would

best benefit TARRO2’s center of gravity.

TARRo2 DESIGN BRIEF 27

Appendix G ­ Failure Analysis of TARRo1

Failure Analysis of TARRo1

By: Jonivan Artates

For: 2015 ASEC Robotics Team

Abstract:

On May 31, 2015 TARRo competed at the UCI Rescue Robotics Competition in Aldrich

Park. During the first run, it was noted that TARRo had difficulty turning and one of the driven

wheels had stopped in providing propulsion to the robot. TARRo was withdrawn from the

competition before a second run was attempted in order to minimize hardware loss. A post event

investigation suggests that a pinch bolt holding the left rear motor, lacking sufficient torque to

constrain the motor, was the single­point failure (SPF) which caused a cascade failure to TARRo.

Observations:

May 31, 2015 UCI Rescue Robotics Competition

­ During run one, TARRo lacked sufficient torque to turn under own power

­ During run one, TARRo had only one of two wheels on the left side rotating

­ Proceeding run one, while on a bench, TARRo was unable to drive the rear left wheel, while the

front left, wired in parallel was driven (wheels unladen)

June 9, 2015 Competition Debrief

­ Testing each of the four motors, each had a resistance similar to the others except the left rear,

which had an open circuit (infinite resistance)

­ While removing the pinch bolts securing the motor to the mount, it was noted that they were

easily loosened using an allen wrench; only slightly tighter than finger tight

TARRo2 DESIGN BRIEF 28

Figure 1: Assembly of motor and mount into chassis

­ While removing the motor, the motor leads had been twisted together and the red wire had become disconnected from the motor tab at the solder joint. The red motor lead was bent in the direction of the lead, as shown:

Figure 2: Motor leads and tab

­ Applying electricity directly to the motor drove the output shaft

TARRo2 DESIGN BRIEF 29

Analysis:

The cause of the open circuit was a break at the joint where the red lead was soldered to

the motor lead. The cause of the break seemed to have been through the wire leads twisting and

causing excessive force at this joint. The bent motor lead supports this hypothesis.

The twisting of the wire leads would be the result of the motor rotating in its housing,

while the wheel and wire leads were fixed. The method of securing the motor in the housing is to

pinch the motor in a cylindrical mount, with cutouts on opposing sides (see Figure 1). As the

inner diameter (ID) of the mount is larger than the outer diameter (OD) of the motor, pinch bolts

are used to close the gap then increase the frictional force between the smooth plastic mount and

the smooth metal motor casing. This system lacks a key to interfere with motor rotation. Without

adequate torque on the pinch bolts, the motor can rotate within the housing if the wheel applies a

greater moment than the housing. The ease of removal of these bolts indicates that the torque on

the pinch bolts was less than necessary to fix the left rear motor.

Other Considerations:

1) Installing suspension on TARRo would allow more driving wheels to be in contact with the

ground at a time, reducing the power and therefore moment each motor undergoes to drive the

robot.

2) Reducing the weight of TARRo would reduce the power consumption of each motor and

therefore moment each motor undergoes to drive the robot.

3) Additional driven wheels would reduce the moment on each individual motor.

TARRo2 DESIGN BRIEF 30

Appendix H ­ Table of Acronyms

A ­ Amps IVC ­ Irvine Valley College

ASEC ­ Applied Science and Engineering Club

MDF ­ Medium density fiberboard

BOM ­ Bill of Materials NIST ­ National Institute of Standards & Technology

DNF ­ Did Not Finish NiMH ­ Nickel Metal Hydride

DGPS ­ Differential Global Positioning System

OD ­ Outer diameter

ESC ­ Electronic speed controller PID ­ Proportional­integral­derivative

EMI ­ Electromagnetic interference PWM ­ Pulse width modulation

FDM ­ Filament deposition modeling RAG ­ Robotics Activities Group

FRI ­ First Responder Interface SLAM ­ Simultaneous Localization and Mapping

FLIR ­ Forward Looking Infrared SPF ­ Single­point failure

GPS ­ Global Positioning System TARRo ­ Triage Assistance Rescue Robot

IMU ­ Inertial measurement unit UCI ­ University of California, Irvine

ID ­ Inner diameter