april 2003 atacama expedition hci field research report angela wagner kristina mcblain

April 2003 Atacama Expedition

HCI Field Research Report

Angela WagnerKristina McBlain

HCI Field Report: April 2003 Atacama Expedition Slide: 2 of 30

Project Description

Intro

ductio

n

Human-Computer Interaction (HCI) Team Kristina McBlain

MHCI (2003)Human Computer Interaction Institute School of Computer Science Carnegie Mellon University

Angela Wagner MHCI StudentHuman Computer Interaction Institute School of Computer Science Carnegie Mellon University


Project Description

Intro

ductio

n

Research Goal Understand the complete system for human monitoring

of the robot during autonomous navigation and make recommendations for system improvements.

Research Foci Operator Interface (OI) Graphical User Interface (GUI)

OI GUI usage patterns

Robot monitoring system including electronic and “Human to Human” communication

Field process usability

Research Sites Pittsburgh, PA, USA

Atacama Desert, Chile (April 13 - April 17)


Data Collection

Pittsb

urg

h

Use of Existing Operator Interface Interviewed Operator Interface GUI developer

Interviewed OI GUI users from Arctic expedition (2001)

Observed use of OI GUI

Documented current OI GUI

Complete System Structure Studied interaction of

system processes

KSM

screen shot changed


Data Collection: Operator Interface

Pittsb

urg

h

KSM

This slide added and could be displayed while giving info on previous slide if desired....


Data Collection

Ata

cam

a

Autonomous Navigation Observation Observe experienced team members monitoring

operator interface during autonomous navigation

Observe team members observing traversing robot

Observe code integration process

Video tape interface usage during autonomous navigation as available (8 hours)

Use “think aloud” technique during observations

Identify communication pathways

Identify additional tasks occurring in parallel


Environment During Observation

Ata

cam

a

Hyperion’s Team’s Navigation Goals Code integration and debugging

Autonomous navigation

Limit contact between OPS tent and robot observers

Field Team Deployment Pattern OPS tent: 1 operator interface monitor and field

communication contact, 1 mission planner developer, 1 health monitor developer

Robot: 2-3 robot observers, 1 field communication contact

Health Monitor Developer

Mission Planner Developerand Plan Creator



Ata

cam

a

Operations Tent In Field with Robot

OI Monitor andField Comm. Coord.

HCI TeamMember - Video

and Notepad

Robot Field Observers (2-3)1 Field Comm Coord

C

HCI TeamMember - Camera & Notepad



Ata

cam

a

Portable HCI Lab Video camera recording ops tent communication and

operator interface GUI

Note taking: laptop and notebooks

Food tent conference center

4-Runner workstation

Digital camera


Data Analysis

Post D

ata

Colle

ction

Video Coding View tapes and record significant events, providing

traceable support for recommendations

Event Types Coded Communication: electronic and verbal

Tool use: operator interface and others

Operator Interface: breakdowns and redundancies

Communication Modeling Tracked electronic communication between system

processes and operator interface during autonomous navigation

Tracked verbal communication within ops tent and between ops tent and field reps during autonomous navigation


Autonomous Navigation: Electronic Communication

Data

An

aly

sis

Robot

OI GUI

Health Monitor

Mission Planner

Navigator

Vehicle Controller

Mission Executive

State Estimator

Stereo Mapper

Micro-raptorCVS


Autonomous Navigation: Electronic / Human Comm.

Data

An

aly

sis

Robot

OI GUI

Health Monitor

Mission Planner

Navigator

Vehicle Controller

Mission Executive

State Estimator

Stereo Mapper

Micro-raptorCVS

OI Operator (Engineering)

Mission Planner(Science)

Legend


Autonomous Navigation: All Communication

Data

An

aly

sis

Robot

OI GUIHealth Monitor

Mission Planner

Navigator

Vehicle Controller

Mission Executive

State Estimator

Stereo Mapper

Micro-raptorCVS

Mission Planner Debug & Develop

MP Code Monitor

Ops Tent Comm. Lead

Code Integration Coordinator

Hardware Troubleshooting Assistant

Data Collection Configuration Lead

Health Monitor Debug & Develop

HM Code Monitor

Robot Situation Observers (1-3)

Robot Observer Communication Lead

Robot Hardware Troubleshooting Lead

Camera Data Collection Lead

Micro-raptor Expert

2-way Radio

Dennis Cam Experiment Set up and Data Collection

OI Operator (Engineering)

Mission Planner(Science)

Process arbiter

= expected human role= not expected human role

= not expected device role

= human-human interaction

= human-software interaction

= human-hardware interaction


Assumptions: Long Term Goals

Motiv

atio

n

Transparent Robot Robot operations become transparent and science data collection becomes focus

Robot system developers are not routinely involved in experiments

Non-Specialist Operator Interface Monitor Operator Interface monitors are not developers of the system or specialists in robot’s system

Operator Interface monitors have technical experience and have received some training in robot’s system


Assumptions: Near Term Goals

Motiv

atio

n

Standardized Testing Procedures Rely on electronic communication (primarily OI) to monitor robot and troubleshoot

Create redundancies in specialist knowledge of software and hardware

Code Maintenance During Field Trials Create shared knowledge base of code structure and function

Customize information tracking tools to facilitate troubleshooting and development


Foci

Reco

mm

en

datio

ns

OI GUI Organization

Consolidate tools

Update data displayed

Functional breakdowns

Field Process Usability Verify field process accommodates anticipated usage


OI GUI: Organization

Reco

mm

en

datio

ns

Expected

Observed Second GUI with camera data (stereo)

KSM

when was this pic taken - atypical as I recall with hm completely showing and nav cam not visible.


OI GUI: Organization: Decision Points

Reco

mm

en

datio

ns

Decision Point 1: User Profile/sHigh level

More granular

Do these profiles / roles change full project timeframe?

Decision Point 3: User Location (Remote or co-located (with other users and with robot))

Decision Point 2: Information needed by each User Profile at each timeframe in project

Decision Point 4: GUI environment configuration (1 monitor or multiple, OS, GUI dev decision/resulting advantages and constraints)


OI GUI: Organization: Near Term

Reco

mm

en

datio

ns

Robot Monitoring with Concurrent Development

Plan for development needs

Define roles

Define “expected” tasks

Plan to rely on electronic communication and orient interface to autonomous navigation

Improve existing tool usability and function

Consolidate tools

Allocate screen space for non- OI tools

Update data displayed

Address identified functional breakdowns

Deprecate use of interface components that may not be feasible with longer-term remote bandwidth limitations

Second computer / monitor


OI GUI: Near Term: Consolidate Tools

Reco

mm

en

datio

ns

Integrate Driving Tool and Kinematics Tool Combine all visual representations of rover state

Transform driving commands into collapsible component of new integrated tool

Display current roll and pitch limits in addition to current roll and pitch

Consider creating easy way to access average rover speed in addition to commanded (target) rover speed for use in mission planning.

Consider eliminating or hiding infrequently used data.



Reco

mm

en

datio

ns

Allocate Screen Space for Non-Operator Interface Tools Other tools (Micro-Raptor, telemetry manager log) are

being used on a different desktop display on the same screen that the operator interface is being displayed, causing the operator interface monitor to miss key troubleshooting information

Include another physical screen for viewing these tools

Prioritize GUI windows to allow these tools coexist on one screen with the operator interface



Reco

mm

en

datio

ns

Create Stereo Camera Image Tool Users monitoring the operator interface rely heavily on

the stereo camera image in “Chris’ Tool”

Under tighter communication constraints, stereo camera image displays may not be possible

Terrain Sensor already displays stereo map and stereo disparity, remove redundancy with “Chris’ Tool”

Include a panel to only display stereo camera image to allow the removal of stereo camera image data without loosing stereo map and stereo disparity

Goal of one OI GUI for each expedition.


OI GUI: Near Term: Update Data Displayed

Reco

mm

en

datio

ns

Incorporate Mission Planner Data Include Mission Planner state in Operator Interface

including Planning, Executing Plan, and Out of Contact

Include Mission Planner data in Local Map Tool

Display time last plan was uploaded

Display current waypoint and total number of waypoints

Display time last waypoint was reached

Remove Unused Information Remove Sun Yaw in Local Map Tool Heading Indicator

Remove “Laser Goodness Map” in Terrain Sensor Tool

Remove unnecessary Health Monitor faults



Reco

mm

en

datio

ns

Redesign Health Monitor Interface Logically organize the faults to facilitate error handling

Adapt the look and feel of the faults to facilitate trigger detection and clarify enabled vs. disabled faults

Allow customizable view to decrease screen clutter and data redundancy and minimize cognitive load of operator interface monitor

Change placement of and access to “Change Fault Constraint”

Remove faults connected to processes which are no longer used in the system



Reco

mm

en

datio

ns

Redesign Health Monitor Interface Provide time stamped message indicating the most

recent data refresh and current status (Is the Health Monitor “alive” and in sync with OI)

Create log of triggered and un-triggered faults with time stamp to enable review of missed triggers

Provide access to telemetry log data relating to triggered fault condition

Consider audio alert for e-stop if usage pattern will involve moving between screens in a manner that may hide the visual cue.


OI GUI: Near Term: Functional Breakdowns

Reco

mm

en

datio

ns

Improve E-stop Button Function E-stop release on single click

Provide Operating Mode Status Include error message when Operating Mode change is

not possible

Include status message when Operating Mode is changed by a process

Enhance E-stop Information Display e-stop sender and time stamp in main OI tool

Window 1: displayed information


OI GUI: Near Term: Organization

Reco

mm

en

datio

ns

Main robot monitoring section: kinematics, stereo cam, local map

E-Stop: source Health MonitorAccess log for e-stop

Mission Planner

Window 2: displayed information

Left Stereo Image Comment Log

Select item drop-down Select item drop-down

E-stop history

Drive Tool

modes


OI GUI: Organization: Longer Term

Reco

mm

en

datio

ns

Science and OI interface integration

Non-developer OI use

Increase sophistication of data visualizationMake “the windshield” a reality

KSM

add in our swag at this information for eng side in intermediate term. Make sure to mention that in field development should not be ignored.


Field Process Usability: Verification

Reco

mm

en

datio

ns

Manage Software Interdependencies Document interaction between processes

Implement process for reverting to most recent, stable version of code quickly

Manage Inter-Process Communication Create an IPC message tracking system to reduce

debugging time and facilitate code management

Create process to verify needed research data is being captured

Manage Process Starting and Stopping Provide system checks for process boot sequence and

enhance efficiency of re-boot


Questions

Qu

estio

ns ?

??

april 2003 atacama expedition hci field research report angela wagner kristina mcblain

Documents